Sensemaking in Investor Networks: The Interactions between Financial Market Participants and the European Central Bank Christoph Fang-Dse Wu Wolfson College University of Cambridge This dissertation is submitted for the degree of Doctor of Philosophy May 2020 Declaration This thesis is the result of my own work and includes nothing which is the outcome of work done in collaboration except as declared in the Preface and specified in the text. It is not substantially the same as any that I have submitted, or, is being concurrently submitted for a degree or diploma or other qualification at the University of Cambridge or any other University or similar institution except as declared in the Preface and specified in the text. I further state that no substantial part of my dissertation has already been submitted, or, is being concurrently submitted for any such degree, diploma or other qualification at the University of Cambridge or any other University or similar institution except as declared in the Preface and specified in the text. It does not exceed the prescribed word limit of 80 000 set by the Department of Sociology. Sensemaking in Investor Networks: The Interactions between Financial Market Participants and the European Central Bank Christoph Fang-Dse Wu Abstract Central banks have taken centre stage in financial market discourses over the past decade amid unconventional monetary policy and financial crises. They are increasingly active in financial markets and thereby rely on market participants to help with and amplify the policy transmission. But how do market participants actually interpret and make sense of central banks as market actors? How do central banks interact with and influence behaviour of systemically important market participants and do these, in turn, influence central banks themselves? In this context, are central banks in the position to actually implement and achieve what their policies are set out to do? These are the questions the thesis seeks to answer. The aim of this thesis is to highlight an area that has so far been underemphasised in the social scientific study of central banks, specifically, how active market participants, the main interlocutors of central banks, adjust investment practices, research methodologies and decision-making processes to the unconventional monetary policy of the European Central Bank (ECB). By doing so, the thesis’ focus is on the portfolio managers at systemically important asset management companies as the main protagonists, not the central bank itself. The focus of the research spans the asset purchases under the Public Sector Purchase Program and the Corporate Sector Purchase Program from 2015 – 2018. Utilising a multi- method approach, the thesis incorporates a quantitative holdings-based analysis of the corporate interlocking networks created by the purchase programs, a social network analysis of the ECB’s Bond Market Contact Group, in depth qualitative interviews with senior Portfolio Managers at systemically important asset managers in Europe and Asia and documentary analysis of interviews, statements and reports of the ECB, investment banks and asset managers. The thesis is grounded in and aims to contribute to organisation studies, the social studies of finance and economic sociology, particularly with reference to networks. It shows the ways in which both monetary policy and investment decision-making is shaped by social processes of sensemaking in expert networks, and by highlighting the unintended consequences of policy actions, critically assesses the role of these experts. Acknowledgements This journey has been invigorating and to this end, I would like to acknowledge the many people that have helped throughout this project. My deepest gratitude goes to my wife and son, Yuni and Conrad, for their love, support and patience. I thank my parents, Chung-Hsiung and Sigrid, for their unconditional support and trust, and my sisters, Stephanie and Andrea, for always being there. I am particularly grateful to my supervisor, Christel Lane, for guiding me over the past three years, for reading all my drafts and for becoming a close friend in the process. My faculty advisor, Stuart Hogarth, has helped me immensely and I cherished his positive energy. For their feedback and help, I also thank Zsuzsanna Vargha and Hendrik Vollmer very much. I am very grateful to all interview participants in this study. I would also like to extend thanks to my fellow PhD students Andrew, Amarpreet, Arsenii, James, Liran, May, Naim, Niamh, Rodrigo, Shuting, Valentina and Yukari for sharing experiences at various stages of the PhD. I would also like to thank Luke Martell, Donald Slater, Samuel Knafo, Anastasia Nesvetailova, Matthew Sparkes, Jan Sparsam, Benjamin Braun and Geoffrey Ingham for feedback. I am also indebted to my friends and old colleagues, Dylan, Christian, Paulo, Emil, Samir, Leo, Garett, Ondrej, Landy, Chris, Thomas, Yoojeong, Sanu and Ranjit. I would also like to thank Meg Westbury, Laura Jeffrey and Martin Vestergaard at Wolfson and the staff of the Judge Information Centre, SPS and Marshall Library. This thesis is dedicated to my parents for their love, values and inspiration. Table of Contents LIST OF FIGURES ..........................................................................................................................................I LIST OF TABLES ...........................................................................................................................................I LIST OF ABBREVIATIONS ..............................................................................................................................I 1. INTRODUCTION, THEORETICAL FRAMEWORK AND METHODOLOGY .................................................... 1 1.1 INTRODUCTION ..........................................................................................................................................1 1.2 CENTRAL BANKS, MONETARY POLICY AND MARKET PARTICIPANTS .....................................................................3 1.2.1 A central banking primer..................................................................................................................3 1.2.2 The European Central Bank and its monetary policy .......................................................................7 1.2.3 The Asset Management Industry and Portfolio Managers ........................................................... 10 1.3 THEORETICAL FRAMEWORK AND LITERATURE REVIEW .................................................................................... 13 1.3.1 Economic Sociology and Social Network Analysis ........................................................................ 14 1.3.2 Organisation Studies and Sensemaking ........................................................................................ 24 1.3.3 Social Studies of Finance and Behavioural Finance ...................................................................... 29 1.3.4 Discussion ...................................................................................................................................... 34 1.4 METHODS AND DATA .............................................................................................................................. 40 1.4.1 Network analysis and data............................................................................................................ 42 1.4.2 Semi-structured Interviews, reports and transcripts..................................................................... 44 1.5 CHAPTER STRUCTURE ............................................................................................................................... 47 2 SENSEMAKING AND INVESTOR FRAMES ........................................................................................... 49 2.1 INTRODUCTION ....................................................................................................................................... 49 2.2 FRAMING IN THE ANALYST LITERATURE ........................................................................................................ 51 2.3 ASSET MANAGEMENT AND MIFID II .......................................................................................................... 53 2.4 INVESTOR MEETINGS AND THE CHANGING LANDSCAPE OF INVESTMENT RESEARCH ............................................... 56 2.5 PORTFOLIO MANAGERS’ SENSEMAKING AND INVESTOR FRAME CONSTRUCTION .................................................. 66 2.6 CONCLUSIONS ........................................................................................................................................ 75 3 CORPORATE INTERLOCKS AND EMBEDDEDNESS: TOWARDS A HOLDINGS-BASED MODEL OF FINANCIAL NETWORKS .............................................................................................................................................. 77 3.1 INTRODUCTION ....................................................................................................................................... 77 3.2 FINANCIAL NETWORKS ............................................................................................................................. 79 3.3 CENTRALITY IN FINANCIAL NETWORKS ......................................................................................................... 81 3.4 EMPIRICAL USAGE ................................................................................................................................... 84 3.5 NOTATIONS AND OVERVIEW ..................................................................................................................... 85 3.6 CONCLUSIONS ........................................................................................................................................ 88 4 CORPORATE CONNECTIONS AND SOCIAL IMITATION IN THE EUROPEAN CORPORATE BOND MARKET 91 4.1 INTRODUCTION ....................................................................................................................................... 91 4.2 BACKGROUND AND HYPOTHESES ............................................................................................................... 93 4.3 THE CASE STUDY OF THE EUROPEAN CENTRAL BANK’S QE .............................................................................. 95 4.3.1 ECB as market participant node .................................................................................................... 95 4.3.2 Node considerations – Active and Passive .................................................................................... 97 4.3.3 Data Collection .............................................................................................................................. 98 4.4 RESULTS ................................................................................................................................................ 99 4.4.1 Network demographics and statistics ........................................................................................... 99 4.4.2 Network Structure ....................................................................................................................... 100 4.4.3 Centrality ..................................................................................................................................... 106 4.4.4 Herding and Imitation ................................................................................................................. 108 4.4.5 Analysis........................................................................................................................................ 111 4.5 CONCLUSIONS ...................................................................................................................................... 112 5 HOMOPHILY AND HOME BIAS IN THE NORTHERN AND SOUTHERN BLOCS OF THE EUROZONE ......... 114 5.1 INTRODUCTION ..................................................................................................................................... 114 5.1.1 Homophily and Home Bias .......................................................................................................... 115 5.1.2 European Quantitative Easing and the Public Sector Purchase Program ................................... 119 5.1.3 Data Collection ............................................................................................................................ 123 5.2 RESULTS .............................................................................................................................................. 124 5.2.1 Network demographics and statistics ......................................................................................... 124 5.2.2 Whole network homophily measures ......................................................................................... 125 5.2.3 Homophily in nodes and home bias in local bond markets......................................................... 129 5.2.4 Regional Homophily, Southern Bias ............................................................................................ 133 5.2.5 Notable national sovereign bond market characteristics ........................................................... 135 5.2.6 Analysis........................................................................................................................................ 136 5.3 CONCLUSIONS ...................................................................................................................................... 137 6 EXPERT NETWORKS AND SENSEMAKING: THE CASE OF THE BOND MARKET CONTACT GROUP ......... 140 6.1 INTRODUCTION ..................................................................................................................................... 140 6.2 A BACKGROUND TO CENTRAL BANK COMMUNICATION: FROM SECRECY TO TRANSPARENCY ............................... 142 6.3 EUROPEAN CENTRAL BANK COMMUNICATION ............................................................................................ 146 6.3.1 Communication during the Mario Draghi Era (2011 – 2019) ..................................................... 148 6.3.2 The ECB’s Bond Market Contact Group and Expert Networks .................................................... 150 6.3.3 Diffusion of the Purchase Programs in the BMCG network ........................................................ 158 6.4 INFORMATION EXCHANGE IN ASSET MANAGER NETWORKS ............................................................................ 166 6.5 CONCLUSIONS ...................................................................................................................................... 171 7 ADAPTIVE FRAME CONSTRUCTION: STRATEGIES AND TACTICS ........................................................ 175 7.1 INTRODUCTION ..................................................................................................................................... 175 7.2 STRATEGIC AND TACTICAL ADAPTATION ..................................................................................................... 176 7.2.1 European Corporate Credit and the CSPP ................................................................................... 177 7.2.2 Sovereigns and Rates .................................................................................................................. 184 7.2.3 Equities ........................................................................................................................................ 188 7.3 COMMUNICATION AND INFORMATION EXCHANGE ....................................................................................... 190 7.3.1 ‘Whatever it takes’ ...................................................................................................................... 190 7.3.2 Information and expert networks ............................................................................................... 192 7.4 REFLECTIVE SENSEMAKING ...................................................................................................................... 194 7.4.1 Communicating the exit .............................................................................................................. 194 7.4.2 End of QE? PM evaluations ......................................................................................................... 197 7.4.3 Analysis........................................................................................................................................ 200 7.5 CONCLUSIONS ...................................................................................................................................... 201 8 FINDINGS AND CONCLUSIONS ....................................................................................................... 204 8.1.1 Findings ....................................................................................................................................... 204 8.1.2 Summary and conclusions ........................................................................................................... 209 APPENDIX A – FORMAL INTERVIEW PARTICIPANT DEMOGRAPHICS ......................................................... 215 APPENDIX B – GLOSSARY OF TERMS ....................................................................................................... 215 REFERENCES .......................................................................................................................................... 218 i List of Figures Figure 1.1 Monetary Policy Transmission Mechanism. ............................................................ 4 Figure 2.1 Decision-making processes in investor networks. .................................................. 72 Figure 2.2 CFTC Net Speculative Positioning (Futures & Options report). ........................... 75 Figure 4.1 Network density at 𝑘1 − 4 ownership levels. ...................................................... 102 Figure 4.2 STOXX 600 top 20 holdings-based network ....................................................... 103 Figure 4.3 CSPP top 20 holdings-based network .................................................................. 103 Figure 4.4 CSPP Network with >5% partition and weighted edge size (Apr 2018 - Feb 2019) ................................................................................................................................................ 104 Figure 4.5 CSPP Node Chart listed by Closeness Centrality................................................. 107 Figure 4.6 Snapshot of largest ETFs tracking the Bloomberg Barclays Euro Corporate Bond Index ...................................................................................................................................... 107 Figure 4.7 Diagram of ECB monthly net purchases by asset purchase program................... 110 Figure 5.1 Monthly net purchases by agent at the start of PSPP and expansion of policy, based on ECB illustration (European Central Bank 2016) .................................................... 121 Figure 6.1 BMCG membership affiliation network (2013 – 2019) by city ........................... 153 Figure 6.2 Social Membership Network of the BMCG (2013 – 2019) ................................. 154 Figure 6.3 Institutional affiliation in the BMCG membership network (2013 – 2019) ......... 158 Figure 7.1 Screenshot of corporate bonds from interview participant................................... 177 Figure 7.2 Screenshot of credit fundamentals from interview participant ............................. 178 Figure 7.3 Screenshot of 2015 European Bond Yields from interview participant ............... 185 Figure 7.4 Screenshot of US Investment Bank Forecast of the S&P 500 US equity index based on Central Bank purchases .......................................................................................... 190 List of Tables Table 4.1 Comparable network statistics at >0% and >2% ................................................... 101 Table 4.2 Nodes table ranked by Closeness Centrality Score ............................................... 107 Table 4.3 Pearson Correlation CSPP, iShares Bloomberg Barc Euro Corp, STOXX 600 February 2019 ........................................................................................................................ 110 Table 4.4 Strength of Ties test over the observation period .................................................. 111 Table 5.1 PSPP network statistics by country at >0% and >2% threshold ............................ 125 Table 5.2 Whole-network PSPP E-I Index and HHI, April 2018 – March 2019 .................. 128 Table 5.3 Statistics of Sovereign bond markets in selected countries ................................... 130 Table 5.4 Inbreeding homophily: Frequency of in-country connections by country breakdown (%) (n=530) ............................................................................................................................ 131 Table 5.5 PSPP in-country connections (actual vs expected) for eurozone members ........... 132 Table 5.6 PSPP Individual Country Network Statistics: Bloc influence ............................... 134 Table 5.7 Herfindahl Index Values, March 2019 .................................................................. 136 List of Abbreviations ABS – Asset - Backed Securities AM – Asset Management Respondent AMC – Asset Management Company BMCG – The ECB’s Bond Market Contact Group ii BOE – Bank of England BOJ – Bank of Japan BTP – Buoni del Tesoro Poliennali CFA – Chartered Financial Analyst CFTC - Commodity Futures and Trading Commission CSPP – Corporate Sector Purchase Program DB – Deutsche Bank DCM – Debt Capital Markets EBITDA – Earnings before interest tax depreciation and amortisation EC – European Council ECB – European Central Bank ECJ – European Court of Justice EFSF – European Financial Stability Facility EMH – Efficient Market Hypothesis EMU – European Monetary Union ESCB – European System of Central Banks ESMA – European Securities Market Association ETF – Exchange Traded Fund EU – European Union EV – Enterprise Value Fed – The Federal Reserve of the United States of America FICC – Fixed Income Currency and Commodities FIX – Financial Information Exchange GC – Governing Council GFC – Global Financial Crisis HF – Hedge Fund Respondent HHI – Herfindahl-Hirschman Index HICP – Harmonised Index of Consumer Prices I – Interviewer IB – Investment Bank Respondent ICMA – International Capital Market Association IG – Investment Grade LIBOR – London Interbank Offered Rate MiFID – Market in Financial Instruments Directive MS – Market Share NCB – National Central Bank OTC – Over-the-counter PIMCO – Pacific Investment Management Company PM – Portfolio Manager PSPP – Public Sector Purchase Program QAP – Quadratic Assignment Procedure QE – Quantitative Easing RPA – Research Payment Account SSA – Sovereigns, Supranationals and Agencies TARGET – Trans-European Automated Real-Time Gross Settlement Express Transfer System US/USA – United States of America YTM – Yield to Maturity 1 1. Introduction, Theoretical Framework and Methodology 1.1 Introduction Over the past decade, central banks have been in the limelight, fighting financial crises, conducting bank bailouts and implementing unprecedented monetary policy. Looking back to 2010 at the start of the eurozone crisis, individuals exposed to the Irish, Greek, Spanish and Portuguese economies lost meaningful percentages of their household wealth, jobs and homes. The image of Greek citizens queuing at cash machines in Athens to withdraw their allotted daily cash comes to mind. Later in 2014, savers in Germany started to lose income generated from their savings and were forced to pay interest instead as rates turned negative. In 2017, the credit rating agency Fitch estimated the total outstanding debt with negative coupons, i.e. guaranteeing a loss to the investor if held to maturity, to be at US$ 9trln.1 At the end of 2019, the German 2-year government bonds were yielding -0.7%, implying that investors were willing to pay the German government for holding their bonds. The past decade has seen an unprecedented rise in the ECB’s balance sheet to almost €5tln, ~40% of GDP.2 Current research has focused on central banks as the main protagonist in financial markets. This thesis will address the market participants in the asset management industry, whom central banks crucially rely on in their market-based monetary policy. The key focus of the study is on the decision makers in these systemically important asset management companies (AMCs). By applying a network approach to this analysis, the European Central Bank (ECB) is seen as just one actor in the financial and social networks it is active in. The thesis explores occasions of sensemaking among market participants, as well as meetings in which individuals cooperate and make sense of market developments. It does not perform a symmetrical analysis of internal sensemaking meetings at the ECB, such as found in Abolafia’s (2004, 2010, 2020) work on the US Federal Reserve. In taking this approach, the thesis adds to the existing literature by focusing on the interactions of the ECB with its main 1 See Quartz, June 2017, https://qz.com/1005720/negative-interest-rates-the-world-is-awash-in-9-trillion-of- bonds-that-are-guaranteed-to-lose-money/ [accessed 03.04.2020]. 2 https://www.ecb.europa.eu/pub/annual/balance/html/index.en.html [accessed 03.04.2020] 2 interlocutors, active market participants making investment and asset allocation decisions informed by their interactions with the central bank. My main motivation for studying these relationships derive from my own experiences as a portfolio manager and circle around the question of why central bank interventions in capital markets have changed behaviour of investment professionals so profoundly over the past 10 or so years. Are central banks in the position to actually implement and achieve what their policies are set out to do? What are the assumptions that are guiding these decision-making processes? How do central banks interact and influence behaviour of systemically important market participants and do these, in turn, influence the central bank itself? How do portfolio managers (PMs) make sense of the ECB and its communications? These are the research questions this PhD thesis seeks to answer. To tackle these questions, I look at the ECB’s monetary policy and communication through the eyes of PMs. The research focuses on the period of quantitative easing under Mario Draghi’s ECB presidency, the monetary policy during two purchase programs, the Public Sector Purchase Program (PSPP) and the Corporate Sector Purchase Program (CSPP). As part of the analysis, the thesis examines two key policy outcomes, the portfolio rebalancing effect and international diversification. It is analysed how market participants interpret, adjust and interact with the ECB and what shape behaviours in financial markets take as a consequence. Section 1.2 provides a general introduction to central banks and their role in and interaction with capital markets. Using literature from economics and central banks, sub- section 1.2.1 is a primer on central banks and discusses the main theoretical assumptions of monetary policy and the connection it has with financial markets. Sub-section 1.2.2. outlines the history and the recent policies of the ECB. It also introduces two of the policy objectives which will be analysed in chapters four and five. Sub-section 1.2.3 introduces the asset management industry to the reader and underlines its importance to central banks in its monetary policy transmission. Section 1.3 outlines the theoretical framework of the thesis by reviewing the relevant literature and concepts in the economic sociology and social network analysis (1.3.1), organisation studies (1.3.2) and the social studies of finance and behavioural finance (1.3.3). A discussion of the theoretical framework is offered in 1.3.4. Section 1.4 introduces and summarises the methods and data used and section 1.5 outlines the chapter structure. 3 1.2 Central Banks, Monetary Policy and Market Participants 1.2.1 A central banking primer A nation’s central bank regulates the money supply, the monetary base and sets interest rates for a given economy.3 Monetary policy is aimed at certain objectives, most notably price stability and/or full employment. Economic theory suggests that price stability provides substantial benefits to the economy and makes monetary policy the key tool to achieve macroeconomic stability (Mishkin, 2007, p. 39; Scheller, 2004, p. 45). Each national central bank may have different pre-defined objectives, for instance the Federal Reserve of the United States of America (the Fed) has a dual mandate of price stability and full employment. The ECB has a sole objective of price stability (see section 1.2.2). There are additional responsibilities of central banks, for example the Bank of England (BOE) also has a mandate to ensure financial stability, while the Fed, regulates and oversees the domestic commercial banks. In most cases, central banks have also become what Walter Bagehot (1873) termed the ‘lender of last resort’ in times of financial crisis by providing emergency liquidity. There are conventional and unconventional monetary policy tools to achieve central bank objectives. Conventional tools consist of setting interest rates, at which central banks lend to commercial banks, and open market operations, the purchase and sale of government bonds and repurchase agreements with banks to increase or decrease money supply.4 Over the past 10 years, central banks have also increasingly utilized unconventional monetary policy tools to achieve their objectives. That is, for instance the use of the central bank’s balance sheet to purchase assets other than government bonds, such as mortgage backed securities, equities or corporate bonds to affect overall price levels in the economy through higher asset prices. The ECB, for instance, launched the PSPP, a sovereign debt purchase program in 2015 and the CSPP in 2016, which I expand on later. In order for central banks to achieve their quantitative targets for a given economy, they rely on financial markets to reflect the intended price levels in the chain illustrated in Figure 1.1 below. Central bankers’ assumptions are that the unconventional monetary policy relies on 3 The European Central Bank is the first supranational central bank to do this across the eurozone economies. 4 Lower interest rates should spur lending demand and vice versa, while buying government bonds releases money held by other market participants thereby increasing money supply. Repurchase agreements are traded on the Repo market. 4 the transmission mechanism through different channels in capital markets: First, the implementation of purchase programs solidifies the central bank’s perceived commitment to its policy and if credible, impacts inflation expectations. Second, with the ECB buying, for example, long dated sovereign bonds, market participants are forced to rebalance portfolios and move into riskier assets, either with longer maturity, or to riskier corporate debt. As a result, financing conditions for corporates improve with access to debt markets and they may subsequently choose to invest into their business and expand. Third, commercial banks are forced to hold reserves with the central bank at, currently, negative rates in the eurozone, and are thereby incentivised to seek higher returns by increasing loan supply to the real economy. Fourth, with lower rates, the Euro is likely to weaken thereby increasing the chances of imported inflation; imported goods become more expensive in local currency terms. Lastly, with greater asset price inflation, economic agents experience higher household wealth and may be incentivised to consume more (Strasser, 2018, pp. 3–14). In monetary theory, these five aspects of the transmission mechanism help achieve the inflation targets of the central bank. A significant lag in the transmission mechanism or inefficiencies within the chain lessens the effectiveness of monetary policy. Central bank communication has been assumed to help the transmission mechanism (see e.g. Siklos & Sturm, 2013). The research in this thesis critically assesses these assumptions with the use of sociological methods. Figure 1.1 Monetary Policy Transmission Mechanism. Source: ECB. 5 With the introduction of unconventional monetary policy, reliance on financial markets to implement monetary policy has grown further. Central banks have “moved away from administrative and rules-based instruments towards reliance on money market operations which they conduct as one participant [emphasis added] in financial markets” (Laurens, Arnone, & Segalotto, 2009, p. 1). The ECB’s balance sheet for instance, stood at around EUR €4.7tn at the end of 2018 compared to only €2.2tn in 2014, the result of its various purchasing programs.5 In comparison, as of the end of 2019, the world’s largest AMC Blackrock manages around $7.4tn and the largest listed European Asset Manager we are aware of, Amundi, invests around €1.6tn of assets. 6 Both the ECB and the Bank of Japan (BOJ) have also introduced another unconventional monetary policy tool in the form of negative interest rates. Negative interest rates force commercial banks to pay interest on reserves it deposits with the central bank. To offset this loss, the commercial bank is incentivised to lend excess funds to companies in the real economy to generate a higher return. The other consequence, however, is that the commercial bank may be forced to pass on this interest cost to the customer that deposited the money into their bank account, thereby eliminating profits from the saver’s interest income. Excess deposits, then, turn into a ‘hot potato game’ that get passed around and are supposed to encourage risk taking.7 Amid the rise in unconventional monetary policy, the biggest trend in central banking has been increased institutional transparency and independence; both qualitatively and quantitatively documented (Blinder, 2004; Dincer & Eichengreen, 2007; El-Erian, 2016; Siklos, 2011). The initial premise for independent central banking was not to succumb to short term pressures of politicians’ election cycle and to form an independent longer term policy strategy that ensured price stability. Central banks have in general, transformed from government tools of economic policy into independent bodies that pursue the ‘public good’ of price stability (Laurens et al., 2009). While instrumental independence, independence in the choice of tools to achieve policy objectives, is seen as desirable by economists, goal independence, central banks setting independent monetary policy goals, is more problematic, 5 https://www.ecb.europa.eu/pub/annual/balance/html/index.en.html [accessed 12.02.2020]. 6https://s24.q4cdn.com/856567660/files/doc_financials/2019/Q4/BLK-4Q19-Earnings-Supplement.pdf , https://about.amundi.com/Discover-Amundi [accessed 12.02.2020]. 7 I thank an interview respondent for this analogy. 6 as it creates “…a fair amount of tension in a democratic society because it allows an elite group to set goals of monetary policy” (Mishkin, 2007, p. 51). With increased independence and power, the need to communicate and account for central bank action rose significantly, especially after the Global Financial Crisis (GFC) in 2008 and the implementation of unconventional monetary policy. Subsequently, Fed Chair Bernanke instituted regular press briefings and introduced the ‘dot-plots’ - future interest rate estimates of each member of the board of governors (El-Erian, 2016). Central bank communication increased in the amount and type of data made available, the frequency of releases and press briefings by central bankers. The notion of central bank communication remains complex and as Blinder concluded, there is no consensus on what works best in communication while there remains a need to pay more attention to perceptions of central bank communication by the public (Blinder, Ehrmann, & Fratzscher, 2008). Communication itself has become a monetary policy tool, shaping market expectations of the future rate path as well as amplifying the transmission mechanism. The ECB also goes to significant length to record the (un)favourableness of the press coverage following monetary policy decisions.8 A recent analysis of its own communication policy concluded that “communication can help making [sic] central banks transparent and thereby contribute to their accountability and to the management of expectations of economic agents…the central bank should strive to minimize any uncertainty about its own behaviour, to the extent possible” (Coenen et al., 2017, p. 3). The thesis examines and challenges existing conceptions of central bank communication in chapter six. A minimum of four key tenets of recent trends in central banking are relevant to sociological enquiry. First, the increasing use of unconventional monetary policy has an impact on financial markets as central banks’ balance sheets start to dwarf the size of institutional AMCs. Sociologists need to look at how this changes behaviour and decision-making processes at large institutional investors and examine the network structure of the financial market more generally. Second, communication becomes a larger part of central banks’ ‘explaining’. As monetary policy tools have become more complex to understand and larger in number, central banks have simply more explaining to do. This takes the shape of educating and guiding market participants on future behaviour or specific conditionality of such expected behaviour. In 2017, Mario Draghi acknowledged that he did not expect the 8 See e.g. Chapter 6 in Siklos and Sturm (2013). 7 extent of the market reaction following his famous quote “within our mandate, the ECB is ready to do whatever it takes to preserve the euro. And believe me, it will be enough” (Draghi, 2012). This invites sociological analysis of central bank communication and how market participants interpret and interact with it. Third, while central banks have been increasingly protected and independent of political interference, it is not clear how independent from financial market participants the central bank really is. Given the dialectic relationship of forming expectations of the future interest rate path and pace of purchase programs among market participants, its position becomes increasingly complex. The more information the central bank discloses about the likely future path of policy, the more market participants are likely to allocate capital to reflect this path. It then becomes increasingly difficult to react to changes in incoming economic data without causing a market upset.9 In the end, at the heart of monetary policy lies the transmission mechanism elaborated above. The transmission mechanism is in essence a sociological puzzle in that monetary policy can only be successful if market participants act in the central bank’s predicted way. Sociology should question the assumptions and analyse the consequences of central bank interactions with the asset management industry. The empirical chapters in this thesis deal with these research questions in detail. 1.2.2 The European Central Bank and its monetary policy Over the past 60 years, there have been numerous attempts at a monetary union in Europe, starting from the treaties of Rome in 1958, to the Marjolin Memorandum of 1962 and the introduction of the European Currency Unit in 1979. Ultimately, the 1989 decision of the European Council to initiate the European Monetary Union (EMU) by the end of the 20th century followed (Scheller, 2004). As opposed to National Central Banks (NCBs), the ECB is distinct in that it operates at a supranational level within a community 19 autonomous states as of 2019. It encompasses only those EU member states, that have adopted the Euro as their national currency. The ECB is the core of the European System of Central Banks (ESCB) and is responsible for monetary policy in the eurozone. As of 2019, the eurozone consists of Germany, Italy, France, Belgium, Greece, Spain, Ireland, Luxembourg, Netherlands, Austria, Portugal, Finland, 9 The US Fed faced this issue with the ‘Taper Tantrum’ in 2013 or indeed in 2015/2016 when rates were not raised four times as indicated. See a recount of events in (Draghi et al., 2017). 8 Malta, Slovakia, Slovenia, Cyprus, Estonia, Latvia and Lithuania. The ESCB consists of the eurozone members as well as the NCBs of the remaining European Union members, namely Bulgaria, Croatia, Czech Republic, Denmark, Hungary, Poland, Romania and Sweden. 10 The ESCB itself has no legal persona. The ECB has its legal framework stated in Article 107 of the EC Treaty and is subject to international law. While not a European Community institution and hence independent of the Commission, it is able to conclude international agreements in its fields of competence. To shield the bank from political interference, Article 108 stipulates that the ECB shall explicitly not seek instructions from community bodies or governments of member states. The NCBs of Euro member states retain their legal framework within the national law of the respective countries. The NCBs fund the ECB and are required to execute the tasks delegated by the ECB. Each NCB is a ‘shareholder’ in the ECB, capital has been provided in line with the ‘capital key’ and monetary income is shared accordingly.11 In line with modern economic thinking, the ECB’s single objective is maintaining price stability within the eurozone. Price stability is not defined by the treaty but by the ECB itself and was set at a “year-on-year increase in the Harmonised Index of Consumer Prices (HICP) for the Euro area of below 2%” (Scheller, 2004, p. 77). It is also within the general mandate to support economic objectives of the Community. At the centre of the ECB is the Governing Council (GC) which consists of 25 members who each have one vote on monetary policy decisions. As of 2019, six Executive Board (EB) members and 19 NCB governors constitute the GC. Decisions are passed with a simple majority, in specific cases a two third majority may be required, such as the use of operational methods of monetary control other than those specified in the Statute of the ESCB. Unanimity is required to amend the statute of the ESCB. The EB consists of the president, vice president and four other members of recognised standing and professional experience in monetary and banking matters (Scheller, 2004). The GC nominates three EB members while the remaining three members are nominated by the eurozone governments. The EB prepares the agenda for the GC meetings and has the right of initiative for decisions 10 https://www.ecb.europa.eu/euro/intro/html/index.en.html [accessed 12.02.2020]. 11 The capital key is the GDP weighted ECB account by member country https://www.ecb.europa.eu/press/pr/date/2018/html/ecb.pr181203.en.html [accessed 10.12.2019]. 9 by the GC. It is responsible for monetary policy implementation in the Euro area, instructions to the NCBs, managing current business of the ECB and publishing all relevant periodicals and balance sheet releases. As part of the discussion in section 1.2.1, the ECB pays close attention to financial market events and uses market signals actively as input into monetary policy making: “when buying and selling bonds, financial market participants implicitly express expectations about future developments in interest rates and prices. Using a variety of techniques, the ECB can analyse financial prices to extract the market’s implicit expectations for future developments” (Scheller, 2004, p. 84). The ECB also manages contact groups in which senior experts from the asset management industry meet with ECB representatives regularly to discuss market conditions, potential policy and investment actions, and likely courses of action given varying market scenarios (see chapter six). These include the Bond Market Contact Group (BMCG), Money Market Contact Group (MMCG), Foreign Exchange Contact Group (FECG) and the ECB Operations Managers Contact Group (ECB OMG). Under Mario Draghi’s presidency (2011 – 2019), the ECB has embarked on unconventional monetary policy in the form of various Asset Purchase Programs (APPs). These purchase programs faced significant backlash from eurozone member states, especially Germany, given the inclusion of local and regional sovereign bonds and corporate bonds, while the ECB usually lends money to commercial banks instead through repurchase agreements.12 ECB President Mario Draghi framed these purchase programs as pure monetary policy with the sole pursuit of the official price stability mandate; in other words the ends justify the means. The ECB announced the sovereign debt purchase program also known as PSPP in January 2015, initially to buy EUR 60bn of securities per month from March 2015 onwards until at least September 2016, dependent on future monetary conditions.13 Given that interest rates were already at zero, the judgment was such that QE in the form of government bond purchases was the best monetary stimulus for the eurozone. The following year, in June 2016, the ECB initiated the CSPP, in which it announced the purchase of €-denominated Investment Grade (IG) corporate bonds with the aim of reducing Euro credit risk premium to incentivise investors to shift allocations to riskier bonds or longer maturities. This thesis 12 This is a unique problem with the eurozone, as it has nineteen different member states, and the incentives could be for weaker states to build larger deficits at the cost of fiscally stronger states. 13 See https://www.ecb.europa.eu/mopo/implement/omt/html/pspp.en.html [accessed 10.12.2019]. 10 analyses these two APPs in detail since they are the ECB’s most experimental and unconventional stimulus programs to date. Amongst the key policy objectives of the ECB is a mechanism referred to as portfolio rebalancing effect, which assumes that should a central bank purchase assets as part of a program and thereby drive prices of bonds higher and yields lower, investors will sell those assets to the central bank and move into riskier assets with higher yields. The specific policy objective is analysed in chapter four. Another key policy objective of the ECB is to encourage cross-border lending and investing in the eurozone. If successful, this decreases both dependency on certain bond holders and diversifies the investor base. This policy objective is analysed in chapter five. 1.2.3 The Asset Management Industry and Portfolio Managers The asset management industry refers to a group of companies that invests and manages portfolios of investments on behalf of clients. The importance of this industry to central banks has been discussed in the illustration of the transmission mechanism in section 1.2.1. The resultant behaviour and response to monetary policy by central banks is conducted within this network of market participants. Funds are the key vehicles in which investments are accumulated on behalf of the AMCs’ clients. Generally speaking, funds are investment products that allocate assets in an often predefined selection of investments. However, some funds with a discretionary mandate, such as hedge funds, may have more loosely defined selection criteria for securities included in the vehicle. There are various categories of funds with different asset classes (equity, credit, currency, property and alternative assets), but also with different styles, themes or geographic focus. Passive funds aim to replicate an index of securities and active funds focus on outperforming a selected benchmark index or target absolute returns. The distinction between passive and active investors is important when looking at the monetary policy in recent years. The low interest rate environment and reduction of risk has led to rising stock markets with low volatility, 14 which in turn led more investors to allocate funds to equities; a riskier asset class. The most efficient way to invest in equities has been the purchase of passive investment vehicles replicating returns of broad equity indices such as the S&P 500 given low management fees, lower transaction costs and 14 Volatility is a statistical measure of risk. 11 higher daily liquidity. Central bank sponsored appreciation in equity prices in turn spurred growth in passive investment vehicles such as Exchange Traded Funds (ETFs).15 Funds form not only the investment vehicles but can also shape the AMC itself, employing managers of the fund, analysts, investment professionals, investor relations and administrative staff. Looking back at the era of 2006 – 2008, traditional fund houses started to list as shares on the exchange, just like the securities issued by the companies they themselves invest in. For instance, Blackrock, State Street, Fidelity or Amundi are all listed companies on the stock exchange. 16 In other cases, insurance companies and financial conglomerates buy fund management houses to diversify their business and acquire expertise in asset management. This aligns the interest of PMs with the AMCs as they receive company shares as part of the remuneration, which makes them important research subjects as key decision makers. The asset management industry has traditionally been a highly profitable business given the elevated level of expertise, asset- and employee-light business model and the efficient use of capital employed. Simply put, billions of dollars can be traded by a single individual in financial markets on a daily basis, while a cement plant with thousands of employees may need a few years to generate a return on its initial investment. Between 2010 and 2015, operating margins for the asset management industry were ~36%, almost twice those of consumer and technology services (Financial Conduct Authority, 2016, pp. 10–11). While central bank balance sheets have grown significantly over the past few years, the largest AMCs have done so too. It is estimated that Blackrock and Vanguard, the two largest asset managers, will manage a combined US$20trln by 2023, roughly the size of the US economy today.17 To study these institutions and their interactions with the ECB is not only important given their size, but is also due to the fact that retail investors utilise these investment vehicles. The asset management industry has to consider central banks in a variety of ways. In the period before the financial crisis of 2008, central bank communication was mostly a domain 15 ETFs are passive investment vehicles that are less costly and more liquid than other types of passive funds. 16 Vanguard Group, with assets under management of around US$ 4.3trln is an exception and unlisted. 17 See Bloomberg article https://www.bloomberg.com/news/features/2017-12-04/blackrock-and-vanguard-s-20- trillion-future-is-closer-than-you-think [accessed 03.04.2020]. 12 for bond and currency investors given the focus on the expected paths of interest rates. With the introduction of QE and global central bank balance sheet expansion, communication became more important for equity investors as the excess liquidity started to fuel the stock market. While the legality of buying equities outright for the Fed and the ECB remains unclear, the BOJ embarked on purchasing equities directly through ETFs. The focus on stemming systemic risk, lending as last resort and thereby suppressing market volatility has amplified the behavioural pattern and Wall Street mantra of ‘don’t fight the Fed’ (Blinder, 2004). The role of PMs in the asset management industry is crucial as will be discussed in chapter two. PMs manage and select securities included in the funds and in the process interact with a variety of counterparties in relation to trading, research, marketing activities and profitability. As mentioned, senior PMs are often shareholders in the funds and the profitability of the business or the share price performance is aligned with the personal compensation. PMs are often required to answer for the performance, security selection, the investment process and administrative duties of a given investment vehicle to clients, fund selectors, fund of fund clients or financial authorities.18 They also manage staff that aid in the analysis and trading of financial instruments such as buy-side analysts or execution traders. Buy-side analysts would help with the security selection while execution traders would help with the portfolio implementation. Buy-side analyst perform similar roles to the PM in that they read sell-side research reports, speak to and meet companies in company-fund manager meetings. However, the PM still decides which securities to include in her portfolio. The interest of the fund and the interest of the PM are uniquely aligned in that the PM’s bonus, remuneration and the likelihood of keeping her job depend on the fund’s performance. In the case of a fund’s underperformance, clients may redeem investments in the fund which can lead to management replacing a PM. In contrast to the corporate executives in previous studies in economic sociology, PMs see the immediate impact of their decisions in the daily price changes of their portfolio of securities. When selecting securities to be included in the portfolio, PMs may consult additional sell-side analysts, those analysts servicing the AMCs and employed at either investment banks or independent research providers. 18 Fund of funds invest across funds or hedge funds. Fund of funds regularly meet with PMs running the funds they invest in. 13 PMs are supported not only by buy- and sell-side analysts, but also by risk managers and operations staff. In some structures, particularly in funds investing across different asset classes, a Chief Investment Officer (CIO), would engage in asset allocation decisions, overseeing PMs dedicated to particular funds or asset classes, e.g. equities or fixed income. 1.3 Theoretical Framework and Literature Review The theoretical framework of the thesis was inductively derived, subsequently developed and refined during my work in financial markets and particularly, the PhD fieldwork conducted. The initial findings from the network analysis led me to adopt the concept of sensemaking to account for decision-making in the networks I was analysing. Both network analysis and sensemaking offer the conceptual and methodological tools to account for what I deem an accurate depiction of investment decision-making, as heterogenous market actors try to make sense of the world, purposefully engage with their counterparts in networks and adjust views and behaviour. The thesis conceptualises the central bank as just one actor in financial networks of active market participants. It integrates sensemaking into the decision-making process of financial market participants and refines the concept of frames. Situating divergent investor frames within the decision-making process of the market participants themselves re-introduces agency into financial networks. While sensemaking processes unfold on an interorganisational basis, frame construction - a product of such occasions of sensemaking - is grounded within the organisational boundaries of the agent, rather than being imposed by outside experts or analysts. The empirical data collection to study the interactions of the ECB with PMs and the impact of its policies is focused on affiliation networks, which were traditionally used to examine interlocking directorates. The affiliations in this thesis look at co-investor networks where AMCs are connected by their shared investments. Within these so-called interlocking corporate networks exist the social networks of the senior decision makers representing the institutions and forming multiplex ties, that is multiple forms of ties between a senior elite of individuals. Financial networks remain dynamic and contested environments as the experts aim to realise their interest in the market. In order to outline this framework, the following sections will review relevant concepts and literatures that are addressed in this thesis. 14 1.3.1 Economic Sociology and Social Network Analysis The historical origins of a sociological analysis of financial markets could probably best be drawn from Max Weber’s involvement in the regulation and politics of the early corn futures exchanges.19 Weber was the first professor to have given lectures on the financial exchange and its inner workings in German history (Weber, Borchardt, & Meyer-Stoll, 2000, p. 103). What resembled today’s securities trading only emerged in the second half of the 19th century through futures trading ‘Termingeschäft’ (Weber et al., 2000, pp. 11–14); speculation ensued. The key premise for Weber was that financial markets (the exchanges) constitute an important pillar for economies that have moved past pure subsistence. It gave farmers an idea about what their stock was worth and provided the ability to economically plan, as well as monetise inventories (Tribe, 2002; Weber et al., 2000). Markets exist if there is struggle between not only buyers and sellers (‘interest struggle’), but also between different sellers of a security and between different buyers respectively (‘struggle of competition’) (Swedberg, 2000). This distinction highlights the area of co-investor networks, rather than trade between buyers and sellers, I seek to explore. Another important aspect of Weber’s analysis is his comparison between different exchanges in Paris, New York, London, Berlin, Hamburg, Leipzig and the aspects of the social network that they encompass, for instance the notion of ‘gentlemen agreements’ in London. "Today's [economic] structure binds each individual to countless others via uncountable threads. Each person tugs on the network of threads, in order to arrive at a position where he wishes to be and where he believes his place to be, but even if he were a giant, and had many threads in his own hand, he would far more be tugged by others over to a place that is actually open for him" (Weber, 2000, p. 321). Hence the notion of networks took shape in both Weber’s early work on financial markets and interest groups. The new economic sociology of the 1980s brought significant innovation to the sociological study of markets and utilised the concept of networks in studying the economy. Emirbayer (1994) contended that this research tradition enabled a relational approach to markets. Foremost, this research tradition focused on how networks influence the economy in that social ties and economic exchange are interwoven and that networks can be formal exchange 19 For a comprehensive overview, see “Einleitung” in Weber, M., Borchardt, K. and Meyer-Stoll, C. (2000, pp. 1–120). 15 structures in which parties become interdependent. Granovetter’s embeddedness theory formed the foundation of this research tradition, as markets were studied as networks and as social-structural arrangements embedded in institutional rituals and belief systems (Preda, 2007, p. 512). The social structures of networks refer to the non-randomness of relations between actors in a network and “assumes that social life… is rooted in the structure of social positions and relations and must be explained by analyzing these patterns…To understand the dynamics of social structures requires knowledge of the social process involved in implementing social change” (Blau, 1982, p. 275). A critique of this approach is arguably the advent of electronic trading, the anonymization of actual exchange (Preda, 2007, p. 525) and the decline of physical presence during the time of exchange, the fall of the ‘embodied presence’ (Knorr Cetina & Bruegger, 2002, p. 909). The focus of this thesis is not on the point of trade in financial markets but on the social structures that shape decision-making. In other words, the interest is not on the point of consociation in Weber’s terms, but on the relationships that are represented by the shared material interest observed in financial networks. Baker (1990) referred to these two distinct forms of ties as the relationship interface and the transaction interface, where the former is characterised by its longer-term and the latter by its more short-lived and episodic nature. The thesis looks at this relationship interface. As a result and despite the proliferation of electronic trading, the sociological body of research and methods of study from the new economic sociology tradition remains highly relevant today. As Powell and Smith-Doerr (2005, p. 380) argue, the conceptual toolkit of network analysis in economic life and its focus on empirical methods bring together diverse sociological work and hence formed a building block for what Urry (2004) referred to as ‘new social physics’, with quantitative scientists entering the discipline in the late 1990s. Harrison White’s (1981, 2008) economic sociology dealt with the influence that social contexts have on economic action, particular in production markets. So called ‘contexts’ resemble the social structure of any given transactions (White, Boorman, & Breiger, 1976, p. 735). White contended that “markets are self-reproducing social structures among specific cliques of firms and other actors who evolve roles from observations of each other's behaviour. I argue that the key fact is that producers watch each other within a market” (White, 1981, p. 518). Competitors would gauge each other’s actions and adjust their own behaviour. White’s PhD student Mark Granovetter (1973), first challenged the assumption in 16 labour economics that the distribution of labour is based on free market principles, in concluding that most job seekers found jobs with the help of acquaintances, relying on so called ‘weak ties’, rather than on close friends. Granovetter’s contribution rejected the traditional assumptions of free market dynamics in labour economics and at the same time, paved the way to conceive of markets as social networks. Granovetter (1985) later introduced the notion that economic action in markets is socially embedded and thereby challenged the neo-classical conception of homo economicus further. Networks form informational ties in markets and bind economic exchanges with social ties (Powell & Smith-Doerr, 2005). The notion of an embedded market approach forms the basis for a social network approach to the empirical analysis of financial markets. PMs are affected by their social relations with other market participants and these relations have a primary impact on the investment decision-making processes. This is a key assumption for the theoretical framework of the thesis, taken from Granovetter’s (1985) work on the embeddedness of economic action.. Arguably, Granovetter’s idea of well-connected groups within networks, known as clusters, was first modelled in Watts and Strogartz’s (1999) article on clustering and small worlds. Networks consist of multiple well-connected clusters rather than being randomly distributed. Granovetter’s weak ties are the nodes that bridge these large well-connected clusters. A jobseeker would interact with her friends on a regular basis but would likely hear of a new employment opportunity through weak ties, those ties that have fewer overlapping contacts with her and hence hear of different opportunities. As applied to financial markets, strong ties between market participants who share a lot of investments are more likely to encourage regular encounters at investor meetings for selected companies, following the same news- flow pertaining to the investments and speaking to the same experts creating different clusters of market participants in financial markets. Weak ties, on the other hand, can offer PMs innovative ideas and inputs. In his empirical work, Ronald Burt (1992) observed that networks mostly consist of highly connected clusters of nodes, nodes that as a result are ‘redundant’ because the same contacts are shared. So called ‘structural holes’, nodes which can act as broker between different clusters can benefit from their position, either through access to different information from the various clusters they are connected to or form the ability to facilitate exchange between members of distant groups. Burt (1992, p. 28) concludes that Granovetter’s weak ties are 17 about the strength of the relationship that spans a chasm between two different clusters, while structural holes are about the chasm spanned itself, and the informational benefits this brings. Brokers can facilitate exchange between disconnected actors and also serve as innovators given their ability to combine new findings gathered from various stages in a production supply chain and actively engage in divergent networks (Burt, 2010). Burt’s (1983) early work focused on the study of interlocking directorates in US businesses, particularly in the manufacturing economy. An interlocking directorate occurs when a person affiliated with one organisation sits on the board of directors of another organisation (Mizruchi, 1996, p. 271). Since the 1950’s sociologists and organisation theorists have researched the effects interlocks of corporate elites have on corporate decision-making and ownership structures (Stearns & Mizruchi, 2005, p. 297). Useem (1980, pp. 54–58) examined the relationship between class and social cohesion amongst the elites in the interlocking directorates of US corporates in the 1980s. Drawing from diverse studies of communities, social clubs, prestigious boarding schools, university boards and boards of orchestras, he (1980, p. 55) found that those directors participating in the interlocking networks exhibit higher social cohesion, as indexed by friendships. Cohesion and unity between senior directors of different institutions can result in influence concerning capital allocation decisions in the financial industry (Mintz & Schwartz, 1981, 1985, p. 151). In the research findings of this thesis, the investing elite meets at various junctures, such as company-fund manager meetings, fund manager-analyst meetings and the ECB’s BMCG meetings, bringing together senior-decision makers that operate in different networks. Social network analysis has been used to analyse interlocking directorates. Interlocks occur due to a variety of reasons, namely collusion, co-optation and monitoring, legitimacy, career advancement, and social cohesion (Mizruchi, 1996, p. 273). By appointing representatives into decision-making positions of an organisation, Selznick defined co-optation as absorptions of potentially disruptive elements into an organisation’s decision-making processes (Mizruchi, 1996, p. 274). Burt also adopts this definition. He describes the co- optive potential of directorate ties are access to trustworthy information and exercise of influence (Burt, 1983, p. 81). Hence, co-optation in corporate interlocks also serves as a means of intercorporate influence and means of communication (Mariolis & Jones, 1982, p. 572). Monitoring on the other hand, specifically refers to a means of inter-corporate information exchange. Furthermore, chapter four shows the effect of interlock centrality 18 (Mizruchi, 1996, p. 282), the effects network centrality in corporate interlocks have on social cohesion and behavioural uniformity in investment decisions. Chapter six examines the ways in which the individual transcends its affiliated institution through social networks, particularly the affiliation networks of ECB experts, for personal benefits. These findings in chapter six are linked to other studies applying the concept of Burt’s structural holes, such as in Reagans and Zuckerman’s (2008) study of non-redundant contacts or Seabrooke’s (2014) epistemic arbitrage of transnational professionals. The relevant social network literature revolves around the corporate interlocking literature, where the key questions focus on structural holes and profits, co-optation and homophily in co-investing. Sociologists have long highlighted the importance of co-investor networks in different contexts; for instance, Sorenson and Stuart (2001) show that co-investor in Venture Capital networks exchange investment ideas and facilitate a rich information exchange. They also find that geographic proximity plays a crucial role in such network formation and impacts significantly on investment decisions. Likewise, early work from economic sociologists yielded both that board directors (Mintz & Schwartz, 1985) and multiplexity of social ties lead to more co-investments (Uzzi, 1997). In this thesis, I use the concept of interlocks in two ways. First, I apply the concept of interlocking directorate ties to the holdings-based network analysis of chapters four and five. Instead of using board memberships as ties, I apply the interlocking directorate methodology developed in the sociological literature to the securities holdings of AMCs in the networks of the ECB. I expound this application in chapter three. The application focuses on portfolio securities holdings, conceived as the material interest of market participants in the financial networks the ECB is active in. In other words, institutions are interlocked with the ECB when they share the securities holdings the ECB buys as part of its monetary policy. Instead of directors and companies, I analyse institutions (ECB, AMCs, institutional investors) and securities holdings. These ties are empirically observable and quantifiable and depict the strength of ties by utilising the data for valued networks, strength based on the percentage holding. This analysis enables the researcher to use formal criteria to identify network boundaries in a nominalist approach (Lauman, Marsden, & Prensky, 1989; Marsden, 1988), where the corporate interlocks of a given group of securities are based directly on ownership of securities holdings. In corporate interlocking networks, the “structural ties created between the two companies by an interlock are mediated by the individuals who actually sit on two 19 boards” (Mintz & Schwartz, 1985, p. 151). Financial networks are grounded in the interlocks of securities holdings between institutions, but also through individuals (PMs) acting within these relationships and making investment decisions. Within the nominalist boundaries of the interlocks exist other social ties between individuals in the member institutions operating in the social networks of PMs and elite decision makers. Useem (1980, pp. 41–42) already highlighted the importance of stock ownership as one of three ways to influence a corporation’s economic reality. With the use of node attributes, the network analysis in this thesis offers insight into the social differentiation and cohesion of the networks the ECB is active in. Second, I apply interlocks to the study of policy networks organised by the ECB. Instead of directors and board seats, I look at ECB experts and the BMCG membership. In chapter six, this analysis shows how the ECB aims to co-opt senior experts in the asset management industry to an elite group and arranges regular meetings at the ECB to foster social cohesion and establish consensus. In examining whether co-optation leads to better profitability, Mizruchi (1996, p. 275) only found mixed support in this research for a correlation between interlocks and profitability. However, in applying interlocks to the study of financial markets, Cohen, Frazzini and Malloy (2008) analysed the social ties between PMs and corporate board members of investee companies and found that portfolios with connected positions, those where the PM knew a board director, outperformed those portfolios that were non-connected, particularly during episodes of corporate news announcements. Chapters four and seven address the co- optation and monitoring conundrum in the interlock literature, namely whether agents can use interlocks for achieving higher returns. PMs can choose to interlock with the ECB by buying the same positions and thereby absorb potentially disruptive elements of investing in European corporate bond markets. Indeed, in the semi-structured interview data presented in chapter seven, PMs attest to the bifurcation of the European bond universe into ECB eligible and non-eligible issues through the application of the purchase programs. In chapter seven, PMs describe the ways in which to extract economic gains by the analytical input gained into the ECB’s behaviour. Co-optation can also serve as a useful tool to establish consensus in an invited group of market participants, thereby limiting the realm of possible actions in certain future market situations. In studying directorate ties, Burt (1983) found interlocking directorates to be a method to replace direct ownership. Board directors were appointed by the firm’s chief 20 executive and would consent to their proposals. These ties were regarded as a “weakened version of an ownership tie” (Burt, 1983, p. 75), thereby harmonising decision-making among corporates. The implications of social cohesion for profitability have been a topic of more recent studies in financial markets. For instance, Fracassi (2016) found that interlocks lead to increased similarity in capital investments and that the more central a particular node in a network is, the less idiosyncratic risk it takes. When applied to a firm’s profitability, Cohen, Frazzini and Malloy (2012) highlighted the contentious nomination practices of appointing the company’s favoured stock analysts as board members and the resultant poor governance of the corporation as a result. The sociological work on interlocking directorates raises important questions of analysing interlocking ties in conjunction with stock or bond ownership ties. The analysis of network ties between senior PMs at AMCs and the ECB (see chapter six), in other words the European investing elite, is an important complementary analysis to the study of material interest in interlocking corporate securities holdings, as securities holdings are the product of decision-making at the helm of the asset management industry. In the AMC, a PM defines and often directs the securities selection process. The PM is thus a senior decision maker akin to the role of the CEO in the manufacturing economies in the corporate interlock studies of the 1980s. While Burt (1983) found that the need for interlocking directorates diminishes in industries that are highly consolidated, chapter five will show that even in countries where the asset management industry is highly developed with multiple actors such as in Spain, the interlocking holdings networks in the local government securities are very dense. In analysing Baker’s (1984) work, Preda (2007, p. 509) argued that denser networks in financial markets lead to lower volatility, while larger networks bring higher volatility due to the uncertainties in searching and processing information. While Preda (2007) concluded that the advent of electronic trading changes this dynamic, the findings in chapter four underline that the ECB is unintentionally tying together actors in the large network of European corporate bond holders. Given the enormous size of the European sovereign bond market (at €10.06tln as of Q2:19)20 chapter five will show that the market is showing a high fragmentation and clustering of actors based on country domiciles, particularly in the Southern European bloc. 20 AFME Government Bond Data Report q2 2019, p. 3. 21 In the late 1990s, network analysis benefitted from both growth in computing technology as well as the arrival of theoretical physicist, such as Mark Newman or Duncan Watts, utilising and adding to the sociological body of literature. Watts developed the clustering coefficient, a measure of Granovetter’s strong ties: “Without any affiliations, the chance that two will be connected is negligible, the more affiliations they have, and the stronger each affiliation is, the more likely they are to interact” (Watts, 2004, p. 118). This brought together the concepts of homophily (see chapter five) and the occurrences of small worlds in a given network. Inspired by White’s notion of contexts, elaborated earlier, Watts took the group affiliations to shape the context of exchanges in networks. As an illustration, members in a tennis club will be tied through their membership and may play against each other in club tournaments or get acquainted through gatherings at the club. Those actors who are affiliated beyond simple membership, for instance by playing in the same club league, are likely to have more interactions and have more affiliations in common. The concept of homophily in the sociology of networks helps to characterise the empirical approach in this thesis. Homophily, or ‘patterns of inbreeding’, is an important occurrence in networks and refers to the phenomenon that nodes with similar attributes, such ethnicity, educational background or profession, tend to be more closely tied (Lazarsfeld & Merton, 1954; Marsden, 1988; McPherson, Smith-Lovin, & Cook, 2001). It is a universal occurrence that people with similarities or common interest are more likely to form connections. Chapter four will show the level of strong ties in the ECB interlocking corporate networks, with the assumption that nodes who share a lot of friends also have more things in common. In this network of institutions and securities, institutions with shared institutions also hold more shared securities in common. Such Granovetterian ties indicate more density in the network. Chapter five in particular examines the notion of home bias and homophily in the sovereign debt market in the eurozone. Related work on homophily has been conducted by both economists and sociologists. Burt (2000) studied homophily amongst investment bankers over a 4-year time frame and found that bankers have a stronger and longer-lasting relationship with other bankers compared to relationships with non-banker business contacts. Gabrieli and Salakhova (2019) conducted a study of 72 banking groups in the eurozone and their cross-border lending practices. They contended that a denser network with shorter average paths and less clusters bears higher risk of contagion. However, a detailed exploration of the social factors influencing decision-making processes in such networks is 22 lacking. In contrast, Stolper and Walter (2018) focused on homophily in the investment advisory business and discovered a positive relationship between the number of demographic commonalities and clients’ propensity to follow investment advice. The study was drawn from a sample of 2,400 advisory meetings in a German savings bank, focusing on retail clients. Centrality is a measure of the position of a node in a given network. In particular, the thesis will use closeness centrality (Freeman, 1978) in chapter four as this measure works well in both ego and connected networks where all nodes are connected to one central node. Implicitly, Mizruchi’s (1996) conception of interlock centrality is also looked at in chapter six by examining the social networks of experts around the ECB. This is based on the assumption that in the asset management industry, PMs have large discretionary mandates to act on the firm’s behalf, akin to the traditional CEO role in the sociological studies of corporate interlocks. Hence investment decisions are taken by PMs rather than the CEO and they thereby exercise considerable discretion in financial markets. A common criticism of network analysis is its ‘atheoretical’ nature, that it is a method rather than a theory, and that the network analysis needs to provide a model that explains the decisions that had led to the actual occurrence and reproduction of network structures (Emirbayer & Goodwin, 1994, pp. 1412–1413; Erikson, 2013, pp. 220–221; Scott, 2017, pp. 6–8; Wellman, 1988, p. 19). Another criticism of the use of network analysis as a tool to research social structures is that the method does not look at the ‘content of ties’ and does not address what content flows between different nodes that are tied together (Powell & Smith- Doerr, 2005, p. 394). Based on Burt’s (1982) assumption and by grounding the network structures as both an input to and product of social decision-making processes of market participant nodes, the thesis addresses this shortcoming in the analysis of financial networks (see especially chapter three). Holdings-based network analysis of corporate interlocks measures the depth of ties and quantifies different stages of involvement in the ties between market participants and can empirically quantify the economic linkages between different nodes in a market. Podolny (2001) contended that the content of ties can be perceived differently by market participants depending on the levels of investments in a relationship. As an illustration, the ECB did not expect the level of commitment to its asset purchases to lead to replication of its investment behaviour and the bifurcation of investment choices in the sovereign and corporate bond market. To address the question about the content of ties more 23 deeply, the thesis examines sensemaking processes and utilises a qualitative approach in analysing the information exchanges in expert networks of the ECB and systemically important market participants in chapters two, six and seven. By analysing the social networks of the experts involved in sensemaking exercises with the ECB in chapter six, the thesis addresses what is known as the dualism of groups and actors and the fact that “the nature of groups is determined by the intersection of actors within them…while the nature of the actors is determined by the intersection of groups ‘within’ them” (Emirbayer & Goodwin, 1994, p. 1418). As White et al. put it, “analyzing systems of formal organizations will require still further developments of network imagery, and these cannot be divorced from models of elites and the ways in which they may control large social systems through the structure of network access” (White et al., 1976, p. 734). Lastly, the question of network formation, specifically focal and membership closure is not answered by a pure network analysis. Easley and Kleinberg (2010, p. 91) raise two questions on network formation specifically. As applied to the ECB’s corporate interlock network in this thesis, network analysis alone cannot answer the question of whether the node bought a shared security as a result of the ECB holding it (focal closure), or because of the number of connected nodes holding it (membership closure). To complement the quantitative network approach to social relations within a given financial market, the thesis utilises further qualitative data to explore the sensemaking processes in elite social networks of the asset management industry and the ECB. Within the institutions there are single decision makers that execute and enact environments through portfolio implementation on behalf of the institutions. The thesis adopts the network approach to study financial markets, in that economic action is conducted by considering interests of competing agents. Asset managers and their ties with other significant market participants such as the ECB are particularly conducive to network analysis, as “networks are formal exchanges, either in form of asset pooling or resource provision, between two or more parties that entail ongoing interaction in order to derive value from the exchange” (Powell & Smith-Doerr, 2005, p. 379). The social networks of decision makers exist within the corporate interlocking networks. For instance, smaller shareholders are usually put into group-meetings at investor conferences, while larger investors have one-on-one meetings or small, more limited, group meetings with management. Even without asking a single question, listening to other participants in 24 meetings can offer a comprehensive overview of what the key factors are that interlocking ties are paying attention to. It is also very likely that pertinent issues, Weick’s (1995) ‘cues’, will be dealt with directly by management or addressed in conference calls and other investor meetings. These cues are then followed by covering stock analysts from investment banks. Hence previous literature on corporate interlocking networks has emphasised that interlocking networks are “formed by a combination of institutional and individual actions [emphasis added], interpretations of interlock networks must encompass both the structural compulsions and the personal decisions that underlie them” (Mintz & Schwartz, 1985, p. 144). To this end, Mintz and Schwartz (1985) caution on interpreting interlocks in a deterministic fashion, which the thesis circumvents by including qualitative research into sensemaking and decision-making of the key decision-making individuals in the networks. Moving on to the more qualitative research data in chapter six and seven, the thesis brings together the notion of sensemaking with the social processes around investment decision- making. Central bank communication is integrated into the social network approach to markets and portrayed as information exchange in expert networks. This moves away from the conception of communication as data sent from Actor A to Actor B and grounds information exchange as meaning creation in social networks. It thereby incorporates an analysis of the content of ties in the network, a common criticism of network analysis addressed in section 1.4. The sensemaking and framing theories from organisation studies are summarised in the following section. 1.3.2 Organisation Studies and Sensemaking The key assumption of the social network approach to financial markets is that exchange is a social action in which other participants, whether physically present or not, are taken into consideration when decisions are made. Relationships in the network influence the behaviour of an individual node. The previous section also argued that the relationships between co- holders, between multiple buyers of a security, are of a longer lasting nature compared to the transactional relationship between buyers and sellers. Exchange in financial markets involves the process of putting new data into frameworks that structure the decision-making process and resultant relationships between market participants. How such decisions are made and how the different stimuli and data points are processed and put into frames is akin to the conception of sensemaking in organisation studies. 25 There is a line of organisational scholars who engage directly with PMs and the ways in which they deal with uncertainty, how investment practices are shaped and performed in this context (Barker, Hendry, Roberts, & Sanderson, 2012; Chong & Tuckett, 2015; Hendry, Sanderson, Barker, & Roberts, 2006; Roberts, Sanderson, Barker, & Hendry, 2006; Taffler, Spence, & Eshraghi, 2017). Particular attention is paid to the need for PMs to overcome anxiety and the emotional effects of uncertainty through company meetings (Barker et al., 2012), Conviction Narrative Theory (Tuckett & Nikolic, 2017) or mental defences (Taffler et al., 2017). Relevant areas to the thesis of PM interactions with corporate management and other significant market actors were sketched out by Barker et al. (2012). Taffler et al. (2017) offer categorical groupings of how PMs make investment decisions. They distinguish between stock pickers, pure quantitative managers and those pursuing a hybrid strategy. Belief in their particular ‘model’ offers the PM a feeling of confidence and control. The theoretical framework of this thesis incorporates the past socialisations of the individual PM, in that the investor frame, a framework guiding decisions and how the PM sees the world, remains a cognitive and micro-sociological notion. Taffler et al.’s (Taffler et al., 2017, p. 58) concept of the PM ‘model’ does not fully capture how PMs go about the everyday life of managing a portfolio. Indeed, the authors acknowledged that there is particular deviation from and flexibility of this ‘model’. This is a good illustration of the adaptive processes PMs undergo in occasions of sensemaking to put the increasing amount of data and cues in an investor frame. Indeed, Weick’s sensemaking is the process of placing “items into frameworks, comprehending, redressing surprise, constructing meaning, interacting in pursuit of mutual understanding, and patterning” (Weick, 1995, p. 6). Weick distinguishes between sensemaking and pure interpretation in that the former constitutes an activity rather than a product (interpretation). In this process, sensemaking constructs, frames, filters and creates tangible facts and reconstructs relations between these facts. According to Weick (1995, pp. 18–61), sensemaking is grounded in construction, retrospective, enactive of ‘sensible environments’, social, ongoing, focused on and by extracted cues and driven by plausibility rather than accuracy. Based on the thesis research, these different facets of sensemaking form an intricate net of social processes before investment decisions in the asset management industry are made. Data points are placed into frameworks that make sense of events and new data, and re-focus portfolio management decisions. Chapter two discusses sensemaking in particular social gatherings in PMs fact-finding processes. Chapter six examines sensemaking 26 processes between the ECB and PMs of systemically important AMCs. Chapter seven analyses specific empirical examples of sensemaking in the interactions of PMs with the ECB from the fieldwork conducted in Frankfurt, London, Munich, Singapore and Hong Kong. Sensemaking in the literature has often dealt with understanding and conceptualising crisis, equivocal or confusing events (Maitlis, 2005, p. 21; Weick, 1988). This thesis, in contrast, aims to integrate the social processes of sensemaking into the day-to-day decision-making processes of investors unfolding in meetings with extra- and intra-institutional experts, thereby circumventing a focus on extreme situations, market bubbles or crises. Chapter two shows that the context of the investor meetings is changing and that PMs increasingly pay for such meetings given the regulatory change of investment research. Thereby, the fund manager-analyst meetings and information exchanges become more controlled by the sensemaking leaders, as PMs direct the topics and form of the paid conference calls. In turn, company-fund manager meetings become more of a fact-finding mission rather than a company-framed exchange. Sensemaking “involves not merely interpretation and meaning production but the active authoring of the situations in which reflexive actors are embedded and are attempting to comprehend” (Brown, Colville, & Pye, 2015, p. 267). These meetings are occasions of sensemaking in which the aim is to reduce multiple meanings, as “people need access to more cues and more varied cues, and this is what happens when rich personal media such as meetings and direct contact rake precedence over less rich impersonal media such as formal information systems and special reports” (Weick, 1995, p. 99). The result of these processes of sensemaking in the asset management industry is not an interpretation, but a more formal investor frame. By processing new information and selected data points in group meetings with experts, agents construct a frame of how they see financial markets. Chapter two develops this conception of the investor frame, which constitutes the way the PM sees the world, a cognitive framework, that shapes the investment process and how portfolio strategies are implemented. Sensemaking thus serves as an input to frame construction. In Weick’s (1979, pp. 164–166) terms, market participants carefully enact their environment through the investor frame expressed in strategic planning and portfolio implementation. Financial markets are thus a product of the enacted environments of each participating member. For instance, at the launch of a new fund this is done by formulating a strategy through social interactions with clients, the firm and outside analysts. 27 It is not an automated process but rather a strategic action and self-referential negotiation in which the organisation enacts its environment (Abolafia & Kilduff, 1988, p. 180). The concept of sensemaking allows for contradictions, divergence and multiple frames. Market participants are self-referential actors that can put models and rational assumptions in the context of a given social situation and focus on plausibility, rather than accuracy (Weick, 1995, pp. 55–61). Earlier work in sensemaking highlighted the need of managers to immerse themselves into events surrounding their organisation and actively make sense of these events. Placing stimuli into frameworks is hence a key exercise and those frameworks take different shapes, they can be complex or simple, following static state of affairs or sequential developments within categories (Starbuck & Milliken, 1988). These frames can focus on statistical properties or follow static assumptions about the investing world. In chapter seven, I report on market participants making a distinction between traditional assumptions in economic theory and the use of plausibility in investment decision-making. PMs use the ECB’s policy objectives to gauge its likely behaviour and make financial gains by front-running the ECB purchases rather than follow the assumptions of economic theory to rebalance portfolios and take on excessive risk. In this thesis, I am attempting to connect the concept of sensemaking with framing and apply these to the investment process of PMs and ECB members. In their recent article, Brown et al. (2015, p. 272) contended that there remain large uncharted topics in empirical sensemaking research. The thesis thereby focuses on an underemphasised area in empirical research and at the same time attempts to refine these conceptions in the literature. Mitchel Abolafia, a PhD student of Granovetter, studied bond trading (1996), monetary policy decisions at the Fed (2004, 2010, 2020) and the social and regulatory structure of the silver market (1988). With reference to Weber’s premise of conflict and competition in financial markets, Abolafia concluded that the futures traders in his study sought competitive markets in which a single large dominant trader would be detrimental to the functioning of an exchange (Abolafia, 1996, p. 68). Throughout his work on financial markets, Abolafia (2004, 2010) incorporated further concepts from organisation studies, particularly sensemaking and enactment theory inspired by Karl Weick (1988, 1995) to analyse the meaning of actions and the decision-making process in particular historical contexts. Abolafia’s key approach in studying the central bank was to analyse decision-making at the Fed by looking at policy 28 shifts and the preceding discussions to flesh out the dynamics leading into a change in policy. By applying the concept of framing, Abolafia was able to highlight the political struggle between the Fed Chair and the remaining voting members of the governing board (Abolafia, 2004). This highlighted the reputational, political and personal ambitions of voting members of the Fed and the legacy of their role at the institution. In analysing transcripts of the Fed prior to policy changes, his narrative approach to sensemaking, in which the internal meetings of the Fed are taken as sensemaking processes, illuminate the processes in which monetary policy is shaped. I aim to refocus Abolafia’s approach to inter-organisational occasions of sensemaking in the analysis of exchanges in expert networks around the ECB. While Holmes (2013) assumes that the central bank’s narrative shapes interlocutors’ views and decisions, Abolafia sees narrative frame making as a primarily internal occasion of sensemaking in the Governing Council of the Fed and the way an actual policy decision comes to fruition. Abolafia’s framing is also more situational as I expound in this section’s discussion below. Leading on from the previous discussion of Weber’s work on financial exchanges, I define financial markets as the product of enacted environments and posit that its structures are both shaped by and in turn shape market participants behaviour. In contrast to Abolofia’s (2010) study of the internal sensemaking in the Federal Open Market Committee, I argue that sensemaking unfolds within networks of both internal and external experts, inter- organisationally. The networks in European bond markets are shaped by interlocks, both based on material interest in the holdings-based networks and on the social networks around the ECB experts that work within these connected corporations. Sensemaking in financial markets is interorganisational and unfolds at particular junctions in the routines of a PM. In chapter two, I will highlight the examples of company-fund-manager meetings and in chapter six the meetings of the BMCG. The occasions for sensemaking in the meetings of the BMCG serve as an illustration of Maitlis’ (2005, p. 35) guided organisational sensemaking, where sensemaking is both highly controlled and highly animated. These occasions are highly controlled in terms of the strict selection criteria of members admitted to the group and the regularity and location of the meetings. They are also highly animated, in terms of the intensity of content and information exchanged. In her empirical work, Maitlis states that “the presence of stakeholders who were actively engaged in shaping the interpretations of events and issues resulted in a greater circulation of information” (Maitlis, 2005, p. 31). As leaders, the ECB members control the agenda and regularity of the meetings and direct research 29 efforts and collaborations of the various members (stakeholders). Hence investor frame construction becomes a process central to each decision maker within an organisation and undergoes a similar social process for members of the ECB’s GC as well as for each PM at Blackrock, Vanguard or any other AMC. In summary, the thesis looks at specific examples of environments in which sensemaking happens, including company-fund manager meetings, analyst conference calls, the ECB’s BMCG meetings, investor conferences, client meetings and internal investment committee meetings. These events happen when the PM is not observing markets and when ‘glued to the screen’, but in social gatherings which offer space for reflection and frame adaptation. In doing so, I argue that the meetings of the BMCG help sustain the networks of experts and stakeholders and at the same time energise them. With each meeting, the ECB actively influences consensus on courses of actions in the changing contexts of financial markets. It is thus an attempt to control the diversity of interpretations of future events and a direct influence on the asset manager’s investment process. However, PMs also use these meetings to co-opt and influence the ECB itself as will be shown. 1.3.3 Social Studies of Finance and Behavioural Finance Despite using different theoretical and methodological approaches, the thesis’ findings are relevant and contribute to the literature in the social studies of finance and to a lesser extent, behavioural finance. Indeed, the thesis’ use of co-investor and social network analysis, as well as the concept of sensemaking, could help refine and solidify the claims made by this literature. In order to outline the relevance, I discuss the relevant contributions in this section. During the early 2000s, an increasing number of researchers in the social studies of finance started to examine the practices of market actors in financial markets. Earlier ethnographic accounts have looked at the epistemic cultures of traders and how traders utilise both calculative frames and how physical trading became a socio-technical domain. This research had focused on trader pits (Zaloom, 2003, 2006), the advent of electronic trading (Knorr Cetina & Bruegger, 2002) and the use of the Black-Scholes-Merton Model on the option exchange of Chicago (Mackenzie & Millo, 2003). More recently, High Frequency Trading (HFT), algorithmic and automated Trading had been the topic of innovative research in this field (Beunza & Stark, 2004; Lange, Lenglet, & Seyfert, 2016; D. MacKenzie, 2018). 30 Important distinctions were made between those market actors who provide liquidity and those who arbitrage prices and front run utilising HFT strategies, thereby highlighting the importance of market actors’ motivation to extract active portfolio returns by pursuing distinct trading strategies. These accounts have been primary influenced by science and technology studies. This particular area of the research field in the Social Studies of Finance emphasises the shared use of tools and modes of behaviour among these expert groups, so called epistemic communities. Another major strand of this literature has focused on economists, technocrats, sell-side analysts and the performative aspect of their interactions with financial market actors. Particular attention has been paid on how economic theory, calculative frames and socio- technical instruments shape decision-making and more broadly, the economy as a whole. Originally, Michel Callon’s work poses the question of how the rational economic model and its implementation ‘in vivo’ is instituted and achieved (Callon, 1998b) - and not whether it exists, thereby examining the very instruments that are used in the economists’ arsenal. The implicit assumption is that economists and the academic field of economics shape the economy.21 This assumption had spurred significant research in the field of performativity studies, in short, that economists perform and shape the economy through financial models, words or in vivo experiments rather than observing it inductively (Callon, 2007; Holmes, 2013; D. A. MacKenzie, 2006; Muniesa & Callon, 2007). In essence, the idea goes back to Merton’s concept of the self-fulfilling prophecy in that “if men define situations as real, they are real in their consequences” (Merton, 1948, p. 504). Mackenzie and Millo have shown in the study of the Black-Scholes-Merton (BSM) model and the derivatives market, the ways in which economic assumptions can shape asset prices (Mackenzie & Millo, 2003) and impose a ‘frame’ which shape behaviours, expectations and prices. With specific reference to the ECB, Callon’s conclusion that real monetary markets correspond to its use of abstract economic theories (Callon, 2007, p. 324) does not explain historical crisis such as the Greek sovereign debt crisis of 2010 – 2012. The question of why crises occur if economists are supposed to be able to control the economy remains. Simple performativity overemphasises the dominance of economic theory in financial markets and thereby neglects other significant economic agents, e.g. hedge funds or pension funds, who 21 For a comprehensive overview and critical assessment of this argument see Slater (2002). 31 may compete in the same markets.22 Zuckerman (2012) argues that even when a theory is expounded by a prominent economist and then widely adopted, it may not be performative. He further argues that the BSM model is actually not performative in shaping future prices but just a reflection of reality. It was intended to be an ‘engine’ rather than a ‘camera’ (Zuckerman, 2012, p. 227).23 Mackenzie (2006, pp. 19–21) specifies that performativity, in the sense of the self-fulfilling prophecy, does not constitute a perfect resemblance of theory in practice. If adherence to the formula or model leads to consistent losses, no market actor would pursue usage of such model. Indeed, what is less straightforward conceptually and more complicated empirically is the effect of a model on market situations specifically. This depends on the ways in which models are used and who uses them. Hence the performative effect of a theory depends on various social contexts. Indeed, recent contributions continue to take a more nuanced approach to the performative aspects of technocrats, economists or traders. Svetlova (2018) shows the actual use of financial models by practitioners and how different tools are chosen to fit the decisions made by the PM, or as quasi market monitoring tools in which consensus estimates are taken as a pulse. While analysing the DCF model in detail, she (2018) further outlines that the BSM model has so far been the only working example of performativity. Hence it depends on who imposes a frame on a particular field in financial markets for it to actually materialise. For instance, while it is clear that central bank communications have a significant impact on markets, the social context and the interpretation of the situation by market actors plays a major role in what impact this communication may have. The literature on central banks in particular has yielded significant research with the broad assumption of performativity developed in the social studies of finance. The earliest and arguably most significant contribution to the literature on central banks comes from social anthropologist Douglas Holmes (2013), who examines the way central banks communicate with the public and perform the economy through policy statements. Holmes directly applies Callon’s (1998b) concept of performativity unto central bank communication and with that the construction of an ‘economy of words’. He (2013) analyses the Fed, the ECB, the Deutsche Bundesbank (the German central bank), the Central Bank of New Zealand as well as the BOE. Through this multi-sited ethnographic work, he seeks to weave narratives of how 22 The famous example of George Soros breaking the BOE in betting against the Pound in 1992 comes to mind. 23 Based on my own experiences in options trading, I would agree with Zuckerman that the model served more as a pricing tool and made it more difficult for brokers to misprice options in their favour. 32 central bankers use carefully crafted words in order to achieve their policy goals and frame the economic discourse. He adds to the literature not only by including global central banks beyond the Fed, but also in establishing that before central bankers make any policy decisions or announcements, they consult a variety of individuals in networks of interlocutors to get a qualitative sense of the economy (Holmes, 2013, p. 47). These interlocutors are market actors, business managers or bankers with a view of the economy. While it is common industry practice to conduct ‘channel checks’ to ascertain anecdotal data,24 Holmes could have gone into more detail about the extent to which the network of interlocutors in turn influence monetary policy making, thereby circumventing a pure focus on technocratic enactment. In a recent collaboration with the BOE, Holmes explores the idea further with the idea of narrative monetary policy making of the BOE and its meetings with interlocutors across the country (Tuckett, Holmes, Pearson, & Chaplin, 2020). This adds additional credence to their narrative approach to monetary policy making. In her study of the BOJ, Riles (2011, pp. 124–130) describes the relationships between central bankers and such market interlocutors, particularly investors, and the changes to regulate such interactions in the articles of the BoJ (Riles, 2011, p. 153). However, there is a lack of detailed empirical research on these important market actors. Hence, the research gap lies in a more exploratory engagement with actual markets participants, both PMs responsible for making asset allocation decisions and the institutions they represent. Greta Krippner studied in detail Fed Chair Paul Volcker and an era in which communications with financial markets and the public underwent significant change (Krippner, 2011). Krippner’s work incorporate the process of financialisation into thinking about central banks. Politicians were willing to pass the accountability for the state of the economy to the Fed, and in doing so, aided this process of financialisation (Krippner, 2011). Velthuis and Braun started to re-focus the central bank research on ECB audiences. Velthuis (2015) addresses the dearth of ECB studies by analysing the media as another actor in the network around the ECB’s policy during Duisenberg’s presidency. Braun’s findings are embedded within the context of the management of an uncertain future and how expectations among economists are shaped in interactions with the ECB (Braun, 2015). Braun (2016) argued that previous scholars have reduced central bank agency to its communicative dimension and thereby underemphasised actual market conduct. While Braun also applies the concept of 24 Based on the thesis research. 33 performativity, he broadens the discussion from communications to include other forms of central bank agency, for instance in his more recent work on the repo market (Braun, 2018). Braun’s work has spurred further research on the ECB from scholars in political economy. The focus of this literature is on aspects of the financialisation of the economy and the technocratic power of central banks (e.g. Diessner & Lisi, 2019; Walter & Wansleben, 2019). The political economist argument deals with the new role of the ECB as an active market participant itself, exercising infrastructural power through markets (Braun, 2018; Braun, Gabor, & Hübner, 2018). Hence, recent developments in the literature point towards a more nuanced approach, such as highlighted by Wansleben (2018, p. 777), taking a more critical approach to the ‘reductionist’ performativity view. In a recent collaboration, Mackenzie looks to have moved more into the embedded market approach to contextualise financial markets as ‘chains of finance’ (Arjaliès, Grant, Hardie, Svetlova, & MacKenzie, 2017). The authors find that financial markets are primarily embedded in and shaped by decisions taken by individuals across a chain of asset managers, brokers and sell side analysts. The use of models is integrated in the interactions of individuals in such chains. Interestingly, in the concluding chapter, the authors acknowledge that despite its complexity, social network analysis could take the role of systematically analysing the individuals in this investment chain (Arjaliès et al., 2017, pp. 156–157). Likewise, in more recent work, Millo embraced a network or an embedded market approach in a collaborative study of hedge fund managers and the sharing of investment ideas (Kellard, Millo, Simon, & Engel, 2017). The interactions show that decision-making is heavily influenced by nodes in the social network of the decision maker, particularly in the example of the hedge fund manager collaboration around a long-short position in Porsche- Volkswagen. Yet another approach analyses how agents deal with uncertainty in financial markets by forming expectations and ‘imagined futures’ (Beckert, 2016; Beckert & Bronk, 2018). They acknowledge that futures cannot be known and to overcome uncertainty, individuals orient themselves towards expectations that are constructed and shaped in the present. The authors (2018) sketch how, across the economy, actors creatively imagine possible future scenarios and make certain decisions based on those imagined futures. The authors argue that this does not have to be antithetical to rational thought, or omission of empirical facts, but requires a creative process of imagination in combination with the known facts at hand. 34 Both network analysis and sensemaking can refine the studies of central banks by scholars in the social studies of finance. The question of which frames may be more successful than others in shaping financial markets can be tackled by network analysis, in which the centrality of nodes is taken as a measure of influence. By looking at co-investor networks rather than transactions between broker and PM, my empirical findings add to this literature in a novel way. The thesis critically assesses both the rationale for and the policy objectives of the asset purchase programs of the ECB and examines whether the expected outcomes are achieved. Lastly, literatures in behavioural finance are addressed with the empirical study of home bias in chapter five. The chapter introduces the link between the sociological concept of homophily with that of home bias, developed in behavioural finance. By introducing this link, future studies in behavioural finance may benefit from engaging with methods of social network analysis. 1.3.4 Discussion The thesis draws from Weick’s (1995) concept of sensemaking. Very common to the processes of sensemaking is placing stimuli into interpretive schemata or perceptual frameworks (Daft & Weick, 1984; Starbuck & Milliken, 1988). As applied to financial markets, sensemaking is the process by which investors establish what it is that is going on in a certain situation and individual investor frames of PMs are being shaped by both such occasions of sensemaking and their own past socializations of the PM herself. This concept of the investor frame is a reflection of how the investor sees the world and leaves open possibilities for different and incongruous decision-making and interpretation of events, reflective of the financial markets PMs embed their decisions in. While Weick’s early work looked at organisations as interpretation systems (Daft and Weick 1984), PMs are often sole decision makers and hence the organisation becomes defined by their investor frames, expressed in portfolio positions and performance. In this investment context, sensemaking adds a new lens to account for both the multiplicity of frames in financial markets but also how PMs organise their working lives around occasions of sensemaking. I aim to refine and refocus the concepts of sensemaking and frames by integrating the investor frame into the social processes of sensemaking in the day-to-day activities of decision makers in financial 35 markets. Sensemaking unfolds during occasions in which PMs regularly adjust to changing market environments in deliberations with selected experts, analysts and regulators, both intra- and inter-organisationally. These occasions help shape the individual frames, that PMs construct to structure their investment decisions. In his seminal work on sensemaking, Weick (1995, p. 111) defines frames as past moments of socialisation and by connecting these past socialisations with current cues in the environment, meaning is created. He argues that such frames shape the content of sensemaking. Likewise, Starbuck and Miliken (1988) develop ‘sensemaking frameworks’ through which managers filter information and through which managers may interpret and perceive the same cues differently. On the other hand, Abolafia uses Goffman to describe framing as ‘interpretive schemata’ and focuses on ‘how actors make sense of their environment and how the changing environment, in turn, shapes that process” (Abolafia, 2004, p. 350), drawing from what he refers to as interactionist theory in sociology. Abolafia speaks of frames as applied to certain situations policy makers face and which suggest certain actions, such as the decision whether the Fed will bail out of the Bear Sterns investment bank in 2008. He sees framing primarily as a political act, a linguistic contest of which narrative wins, while alternative frames may also have significantly different policy consequences (Abolafia, 2004, p. 351). Throughout Abolafia’s work on the Federal Reserve, the idea of frames is equated to an argument or strategy in reference to a specific situation, often elaborated through a narrative. For instance, he mentions the case of the ‘systemic risk frame’ which is associated with support of a bail out of Bear Sterns while the ‘fairness frame’ would let free markets rule (Abolafia, 2020, pp. 97–99). The ‘systemic risk frame’ is hence a singular strategy in a particular situation. Likewise, he analyses previous episodes at the Fed and brings up the competing frames of targeting monetary aggregates and targeting interest rates (Abolafia, 2004, p. 351). In contrast to Abolafia, the notion of investor frames in this thesis is a representation of ‘how the investor sees the world’. Discretionary decision makers such as the PMs in this study, operate more autonomously compared to the executives in management meetings or policy makers in committees. They develop investor frame to independently interpret and analyse an increasing amount of data and cues. The explanatory possibilities of individual investor frames, which the PM develops across situations and time will yield analytical merit, as 36 shown in this thesis. It leaves room for discrepant interpretations, for instance embedded in past socialisations, different research methods and experiences, but also contradictory frames in financial markets. As seen in co-investor networks of the ECB, different rationales lead to investor affiliations and shared material interest. The staging aspect of investment decision-making becomes relevant where the investor is committed to her investment decisions by both having rationalised those in context of her investor frame, but also by clients holding her to account on that basis. For instance, a fund of fund would invest with interview respondent HF1a for his talent in picking stocks and turnaround stories. Should HF1a decide to buy an index stock as his biggest portfolio holding, he could be accused of ‘style-drift’. Nonetheless, through sensemaking processes, PMs participating in this study adapted to the changing environment and were able to purchase bonds with negative yields, although the mandate would usually not allow for this. The investor frame is hence not limited to simple categories such as what Taffler et al. (2017, pp. 58–59) developed with categories of the ‘Quant Manager’, ‘Fundamental Investor’, and the resultant portfolios. That is also the reason why PMs, particularly those with discretionary mandates, often find it difficult to articulate what it is they do.25 The investor frame, and how the PM sees the world, is a result of and fine-tuned in actively arranged social occasions of sensemaking, past socialisations and idiosyncratic research methods to track the investment universe. Given that there is not an infinite number of securities in a predefined universe which PMs encounter, portfolios can often resemble each other across firms. For instance, PMs find it difficult to express the findings of their sensemaking process in a single ‘model’ or ‘system’ that can be pre-manufactured in a certain category. The portfolio implemented by the PM is also rarely a pure expression of the investor frame. Hence, to the outsider, the approaches that are explicated for staging purposes to the investor may seem, superficially, very similar to those of competitors (Taffler et al., 2017, p. 58). Such co-investors hence share a material interest through their portfolio and can only interdependently achieve their goals. For investments to increase in value, investors need other investors to purchase the securities they hold. Investors into funds, such as fund of fund managers or university endowment fund managers, evaluate PMs’ track record, e.g. the fund’s performance over the past five years, how the 25 The famous fund of fund and hedge fund investor Jeff Tarrant noticed this in the early pioneers of the industry he invested in. 37 manager performed during specific crisis months, such as September 2008 or August 2015, and more interestingly, the idiosyncratic themes the PM played over the years and for which she had become well-known for. Quantitative and qualitative evaluations are used to do this. The staging aspects of the investor frame become evident not only to clients, investors in the fund, but also to other PMs in the network who may adjust investment decisions based on who owns a certain stock or bond. As an illustration, if a large fund such as Fidelity starts buying a small-cap stock, other smaller investors may feel safe to buy it. Likewise, brokers pitching ideas to PMs must be aware of investor frames of their clients. This goes far beyond writing down certain technical requirements, such as stocks having a minimum market value, certain liquidity requirements or sectors that are to be avoided. Indeed, these brokers need to find innovative ideas that they deem of interest and coherent to the PM’s cognitive framework, by speaking to different networks and make sense of often contrasting and incompatible ideas. For instance, index huggers or PMs with a low risk tolerance wouldn’t be interested in contrarian ideas. Burt’s (2010) structural holes often need to make sense of contrasting ideas and knowledge and interpret implicit and tacit understanding of different networks. If one were to apply such frames to Abolafia’s study, the individual members of the FOMC could be seen as hawks and doves, with individual characterisations. For instance, Jerome Powell was previously working in private equity and this would imply certain tendencies to keep interest rates low, bail out market participants and so on. Indeed, interview participants in this PhD research made clear that evaluating statements from different members of the ECB governing council requires an understanding of the deeper context of the individual’s past socialisations. Likewise, the ECB president has to pay particular attention to the philosophical view of other GC members, such as Jens Weidman, when proposing policies around the European QE. In this thesis, the PMs managing portfolios for AMCs, do not rely on committee votes for their individual investment decisions and are hence a unique topic of research to explore sensemaking and framing. Occasions of sensemaking are precedents to individual investment decision-making. There is arguably a tension between a quantitative network approach, in particular on information exchange that is shaped by the structure of a network on the one hand, and the idea of enactment and the construction of such social context through framing and 38 sensemaking on the other hand. A previous debate in economic sociology revolved around criticism of prominent concepts in networks, such as Podolny’s ‘networks as pipes’ and Granovetter’s embeddedness, as abstract and akin to the neoclassical economist view the concepts sought to challenge (Ingham, 1996; Krippner, 2004; Krippner & Alvarez, 2007). For instance, economists tend to look at central bank communication as information transfer, data that is separate of its social context and sent from an agent A to agent B (Morris & Shin, 2008). This tension is also invoked by organisational scholars researching at the intersection of networks and organisations (Kilduff & Tsai, 2003; Landis, 2016). These contributions highlight the need for social network analysis to incorporate the individuality, cognition and perceptions of people that occupy the positions in a network (Kilduff & Lee, 2020). Podolny himself argues that perceptions of ties through which information exchange is enabled depends on how such ties are interpreted and, in turn, play a key role in shaping the network and behaviour (Podolny, 2001). In a direct response to the network criticism, Granovetter also expounds that networks are not deterministic and that network analysis offers a link between micro and macro theories (Krippner, 2004, pp. 113–117). Further, over the course of his work on structural holes, Burt adds an important aspect on network formation and information exchange, in that agents actively engage in the network to innovate and interpret often paradoxical and opposing ideas in order to generate returns (Burt, 1992, 2010). Hence information across networks is interpreted and acted upon in diverse ways. It requires, not rational, but purposeful engagement with the network ties and interpretation and improvisation on behalf of the agents filling structural holes. Likewise Uzzi’s (1996, 1997) work on the textile industry, shows that the social ties between companies and their managers are multiplex. The multiplexity of network ties creates even more influence on individual decision-making, which in turn, reshapes the corporate interlocks and social ties overlaying such networks. If the researcher starts to think of networks in this way, sensemaking becomes a fruitful analytical tool to fill the interpretation of the network environment. In financial networks in particular, the interpretations and decisions made by single individuals create ties between corporations in networks of co-investors. PMs acting highly independently, often need to rationalise decisions cognitively on their own, rather than in executive board discussions. This thesis uncovers the particular dynamics of the multiplexity of social ties between senior decision makers and how these affect the formation and shape of corporate interlocks. It also addresses the question of the content of social ties in financial networks as well as the tensions between the sensemaking and network literature by accounting for the fluidity and dynamic nature of networks. 39 Some of the tensions described in the social studies of finance literature around performativity and counter-performativity can be addressed by network analysis in which centrality can serve as a measure of influence in a given network. Inevitably, looking at affiliation networks allows for a multiplicity of frames and this addresses a previous discussion in economic sociology, best summarised by Slater (2002). By examining holding networks and the social ties of individuals overlaying the corporate interlocks, the analysis focuses on the social interactions and occasions of sensemaking between actors in which relationships develop over time. The influence on each other’s life worlds becomes more sustained as opposed to the point of transaction where buyers and sellers meet shortly. The use of calculative frames and apparatuses are important in the investment decision-making of PMs as argued by the social studies of finance literature, however the ways these are used are multifaceted and increasingly decentralised in that PMs develop their own idiosyncratic tools within their own investor frames, which aid in sensemaking processes but are not televised to other market participants. This is also shown in chapter two, in that the influence of sell-side analysts is waning and with it, the particular choice of calculative devices they pursue. Indeed, such tools are kept secret in fear of imitation and are often adjusted and disposed of, particularly amongst quantitative strategies. Depending on the investment themes, new adoption of self-negotiated monitoring tools becomes a dynamic process. For instance, should the PM become interested in a certain investment theme, diverse sources of information and unstructured data are consulted as expounded in chapter six. Lastly, these tools do not alter the social ties which exist between the European financial elite described in this thesis, but different heterogenous calculations unfold in the co-investor networks and in occasions of sensemaking of these individuals. Chapter six shows the non-linearity of such sensemaking processes and the contradictory ideas that structural holes work with. Indeed, network analysis outlines the overall structure of the network ties but does not address the content of ties. Information exchange is hence non-linear and tacit. By applying corporate interlocking methods to the affiliation networks of the BMCG in this thesis, co-optation is invoked as a research question. Does the regulator co-opt senior investment staff of systemically important asset managers to control the consensus and the variation of investment behaviour or do selected members, in turn, shape the policy content and decisions? From a network perspective, an expert group such as the BMCG, inevitably develops a body of shared perceptions and tacit knowledge (Burt, 2010, p. 3). Sorenson and Stuart (2001) already outlined the importance of co-investor networks for 40 purposes of obtaining and sharing information about possible future investments, but the thesis underlines that the multiplexity of ties enables collaborative sensemaking processes that are used to make such investment decisions. The ways in which the members individually utilise this tacit knowledge will be shown in chapter six and seven. Indeed, BMCG members and ECB experts working in AMCs themselves perform different forms of calculation and act purposefully as a result of engaging in this expert network. Reputation, investment performance, job security, career advancements, and policy performance, emerge from the group. The network approach to the asset management industry also underlines the embedded market approach. Actors are not atoms in a rational and informationally efficient market as portrayed by the atomistic view of markets found in neo classical economics and neither do purely rational calculations take place. Rather, the calculations that are incorporated in investor frames are deliberations of motivated actors that contextualise and interpret information and take into account other actors in their network, resulting in behavioural and other types of calculations. As an example, investors will monitor certain valuation ratios of particular securities, but more importantly, other calculations revolve around what passive investors will do, what the ECB is intending to do and its effect on prices. Deliberations take form on whether to align strategies with other actors, what likely scenarios may unfold, whether German Chancellor Merkel is going to agree to Greek financing demands, or whether the British Prime Minister is going to take a more nuanced approach to Brexit negotiations and how this in turn will be interpreted by Brussels. Information exchange hence has to cater for a multiplicity of frames, such that plausibility overrides accuracy and sensemaking processes override reliance on simple calculative devices and valuation ratios. 1.4 Methods and Data The thesis utilises what I refer to as holdings-based network analysis based on the concept of corporate interlocks, an analysis of the affiliation networks of the BMCG, semi-structured interviews and qualitative analysis of investment and market reports, interview transcripts and ECB statements. To compensate for some of the limitations of network analysis, specifically the content of ties and the closure question, a mixed-method approach is taken. Creswell and Plano Clark (2017, 41 p. 9) discuss the need for a mixed-method approach when quantitative results require more explanation. In his evaluation of the use of network analysis in sociology, Newman also underlines the need for a researcher to know the social system well which the network analysis is focused on studying.26 The thesis adopts a multi-method approach in which the research questions are answered utilising both qualitative methods (semi-structured interviews, document analysis) and quantitative methods (social and financial network analysis). The theoretical framework was inductively developed during field research. Hypothesis testing was derived from and guided by the qualitative field work. A multitude of data sources were used to triangulate and validate pertinent themes as described by Creswell (2014, pp. 201–202). A rich and thick description is offered on how investment decisions are made in Chapters 2, 6 and 7 to expound the findings and to pinpoint the influence of the ECB onto networks of financial market actors in the asset management industry. Contacts in the field research were used in a form of peer-debriefing and external audit of preliminary research findings to validate and make findings more reliable. Hence, the methods of member-checking and triangulation (Creswell & Creswell, 2014, pp. 211–212) were used regularly throughout the field work. With the theoretical framework anchored in organisational theory and the economic sociology of networks, the research methods employed were focused on three areas. First, I conducted a network analysis of corporate interlocks of the ECB’s asset purchases, specifically the network analysis of the CSPP (chapter four) and PSPP (chapter five). I applied the concept of corporate interlocks developed in economic sociology to the securities included in the ECB’s purchase program to build the ego networks around the ECB’s monetary policy. I also conducted a social network analysis of the affiliation networks of the ECB insiders based on the BMCG membership data (chapter six). Second, I conducted in depth face-to-face interviews in a semi-structured fashion with PMs and senior decision makers in the global financial centres of London, Frankfurt, Munich, Singapore and Hong Kong (chapter two and seven). These were selected based on our sampling strategy discussed in the following section. Third, I performed a qualitative document analysis of official statements, reports, interviews and meeting transcripts of ECB officials, investment banks and AMCs to triangulate findings (chapter six). Document analysis was used as a 26 See interview “In conversation with Mark Newman: The Future of Network Science, 18 Feb 2016” https://www.youtube.com/watch?v=5BQcM-c7-kU [accessed 03.04.2020]. 42 complementary strategy to the interview data employing careful selection as outlined by Flick (2014). I reduced the amount of data by a) focusing on specific periods before and after the policy announcements and b) focusing on the most influential stakeholders as measured by BMCG membership. I elaborate on each of the three phases below. The mixed method approach in this study is best summarised as triangulation (Flick, 2014, pp. 183–191), in that the methods are all linked by the specific context of the thesis research in focusing on the ECB and its interactions with systemically important asset managers. As described in section 1.2.2, the context described is specific in both historical and geographical terms by focusing on the ECB’s PSPP and CSPP programs. The findings of the holdings-based network analysis, the field interviews and observations and document analysis are all integrated in the research approach. 1.4.1 Network analysis and data Bloomberg was chosen as the main data source for collecting securities holdings data.27 As opposed to Thomson-Reuters, Bloomberg has the most comprehensive coverage of ownership data, given the relationship the firm has established in the asset management industry and resources available for data collection. Additionally, the various Bloomberg Barclays European bond indices are used as benchmark indices for asset managers. The data is housed under the HDS function on the Bloomberg terminal.28 Bonds expire at a certain date or are called and subsequently redeemed, thus recording this data every month offers a unique method to study holdings-based networks in the European bond market over time. 27 At the beginning, the ownership data of randomly sampled CSPP securities were cross-checked on Bloomberg and Reuters’ Eikon. Bloomberg had fuller data available with links to the specific funds holding the securities as well as additional fund information. Bloomberg also manages the Bloomberg Barclays Euro Aggregate Corporate Total Return Index, which is a key index for the European corporate bond market and most passive and active investors track this index. The main priority was to ensure a consistent data source over the research period. It also ensures that future studies can be replicated utilising the same method. 28 “The HDS function is built upon the ownership data that Bloomberg directly receives from asset managers and ownership data that Bloomberg sources from public sources and reports. Our preferred method is to reach out to fund managers to see if we can enter into a relationship to receive fund holdings data… In this arrangement, we will have funds send us holdings data on a frequency and lag of their choosing. In most cases, the receipt schedule is monthly holdings sent on a 30 days lag. We standardize the method in which the data is supplied to us so we can develop extraction technologies to ultimately display the data sent in a timely and accurate manner. Outside of this, we also follow semi-annual and annual reports as well as company or exchange releases to obtain holdings information.” E-mail communications with Bloomberg, March 2018. 43 Data access was granted by the Information Centre at the Judge Business School and the Marshall Library, University of Cambridge. The data collection period for the CSPP and PSPP networks covered the period from March 2018 to March 2019. Data were recorded when no field work related travel was done, at the beginning of each month and when data were available. This resulted in the raw data set for March, April, July, September, October, November and December of 2018, and January, February and March of 2019. The value of this raw data is particularly high for bonds, as Bloomberg does not store backdated data of bond holders. In other words, users can access current holders of bonds but cannot retrace details of holders at a date in the past. The data collection was conducted with the following procedure. Securities held under the CSPP and PSPP were recorded from the ECB website at the specified date, including the identifier codes (ISIN codes). The securities for the STOXX 600, Euro Stoxx 50 and iShares €- Corporate Bond Index were also recorded from the respective websites. Following this step, the 20 top holders of each ECB-held security were downloaded from the Bloomberg Terminal, including holding (bonds) or ownership (equities) percentage, investor type, country of domicile, city, and identifier code. In all cases the ECB is the top holder of each security, but given that the ECB does not publish its detailed position sizes, the ECB itself is not listed on Bloomberg. Hence in effect, the data recorded has the top 20 holders of each security, plus the ECB as top holder. This raw data was imported into Microsoft Excel. With the concept of corporate interlocking networks explained in section 1.3.2 and using Pivot Tables to reorganise the data by security, I built an adjacency matrix for each data set separately, i.e. CSPP, PSPP, STOXX 600, Euro Stoxx 50 and iShares €-Corp Bond index. An adjacency matrix is a square reflecting the number of ties between the nodes in a network. With the ECB holding each security in the network with 20 other recorded holders, each security will yield one tie shared with the other 20 holders. The network ties between nodes are weighted by the number of securities shared and result in a weighted network. A weighted network includes not only the ties between node i and j, but also the strength of the tie. For instance, the tie strength between node i and j is 120, if the two nodes share 120 securities holdings. The process then included reformatting this weighted network into a dichotomous network of 0 and 1s. This was to perform analysis and hypothesis testing that requires both weighted and dichotomous networks, as carried out in chapter four and five. 44 The adjacency matrices were complemented by node attribute tables. There were five node attributes used in the analysis: Investor Type, Country, >2%, >5% and >10%. Investor Type refers to the type of institution the node represents, e.g. Hedge Fund, Investment Advisor, Bank, etc. The node attribute data was then coded for use in network software. The node attribute tables and adjacency matrices were imported into UCINET and GEPHI for analysis in chapter four and five. Chapter six includes a social network analysis of the membership affiliation of the BMCG. I first recorded the names, institutions, the year of membership, gender, job title and work location of the individual members of the BMCG. This raw data was manually searched from the BMCG documents on the ECB website. The membership years ranged from 2013 – 2019. A tie is formed between nodes when they share membership during a given year, hence the maximum strength of a tie lies at seven (seven years of shared membership). This resulted in a weighted adjacency matrix of the membership affiliation network of the BMCG. I then used this matrix to create an affiliation network of individuals, cities, and institutions in chapter six. 1.4.2 Semi-structured Interviews, reports and transcripts In the social scientific body of research on the asset management industry, only a few studies have been conducted utilising semi-structured interviews with multiple PMs (Barker et al., 2012; Chong & Tuckett, 2015; Giorgi & Weber, 2015; Kellard et al., 2017; Roberts et al., 2006; Svetlova, 2018; Taffler et al., 2017) due to, in part, the difficulty of access. The study included informal and formal interview settings. All interviews were conducted face-to-face at respective offices, restaurants or cafes. As discussed in section 1.4, key contacts were used for purposes of member-checking and triangulation. Formal and recorded interviews with 19 senior PMs and 2 traders were conducted as part of a wider study into the ECB’s influences on PMs’ decision-making processes in Asia (Hong Kong, Singapore) and Europe (London, Frankfurt, Munich). Interviews were recorded and transcribed where permission to record was given. Conversations lasted around 60 minutes on average and started with a discussion of investment methods and changes therein. Conversations then covered three main topics revolving around information gathering methods (research methods, information and data gathering, meetings, calls, video 45 conference), investment decision-making (methods, influences, awareness of other market participants) and changes in competition with other market actors (ECB, ETFs, regulatory regimes, MiFID II). There was a particular focus on recalling the particular contexts of the implementation of the PSPP and CSPP. Given their seniority, almost all subjects were active prior to the implementation of said programs and were able to offer and contextualise changes brought about by the ECB’s policies. PMs in the fixed income space were primarily targeted given the influence of monetary policy in this sector. However, respondents in equity and cross asset allocation were also consulted to gauge the scope of the policies’ impact on capital markets as a whole. PMs surveyed pursued an active investment strategy. Sampling was based on attributes of the social group, in this case individuals employed in a systemically important AMC; individual seniority in the relevant role, enabling reflections both on their decision-making conduct but also comparisons of pre-ECB policy implementation with the current market context; geographical responsibility in terms of a eurozone focus; and a focus on the fixed income category as ECB policy affects this asset class disproportionately. The key individual criteria for the sampling selection were seniority and prominence of the subject’s firm as measured in assets under management and network centrality. This ensured both width and depth with respect to the richness of relevant information in interviews. Where possible multiple contacts within the larger, most central firms were consulted. The research also included informal meetings and consultations with seven contacts whose functions were hedge fund managers, fixed income and equity PMs, investment strategists, execution traders and financial data provider sales. These informal consultations served for member checking and referrals, as opposed to a more formal, recorded semi-structured interview. Hence the sample selection followed substantive criteria as often applied in qualitative research (Flick, 2014, p. 168) but also formal criteria such as assets under management. The systematic approach to sampling was integrated in the research process of the network analysis and findings of documentary research, identifying key institutions. The documents analysed consisted of interview transcripts, official statements, industry and investment research reports. Interview transcripts and industry reports were all located online. Access to investment research reports was provided by various data bases, including Bloomberg Professional Services, Reuters Eikon, Mergent by FTSE Russell and JP Morgan Research Proquest. 46 During the data collection for the corporate interlocking networks in early to mid 2018, geographically peripheral actors were contacted first for interviews (such as in Hong Kong and Singapore). Interpretation, transcription and consultation of the literature occurred within the mini-cycles of contacting, meeting and conducting interviews and transcribing and interpreting the interview data. This process culminated in meeting the most central actors in late 2018 and early 2019. I have spent 13 years as a PM and investment analyst in the asset management industry. This enabled conversations to flow easily as I was familiar with the methods, practices, language and strategies employed by subjects and the intricacies of the decision-making processes. In the semi-structured interviews conducted, my own biographical context was known and discussed either at the point of introduction or at the beginning of interviews to increase the validity and reliability of the findings, as elaborated by Creswell (2014, p. 202). Given that the subjects’ expertise was focused on a different geographical region and asset class compared to my own previous activities, it provided distance between the researcher and the subject and the opportunity for reflection without employing auto-ethnography (Flick, 2014, p. 517). My professional history thus guided my choice of methods and identifying complex relationships within financial markets and to utilise business connections made to ascertain mixed-method data. Over 10 out of my 13 active years, the focus was on the Asian markets and by conducting research on the eurozone, I stepped into a territory that was handled by other colleagues and peers. This gave additional distance to the research matter to avoid reliance on my own recollections of market events to answer the research questions. Some of my own observations were only used as a complement to the thick descriptions of changes in the asset management industry as a whole in chapter six. I was able to follow up with interview participants via email after the meetings and in some cases, the interviews were better described as on-going conversations. The field work was conducted over an eight months period. Respondents were contacted and located based on personal contacts established in the industry, and to a lesser extent, snowball sampling within specific firms. To protect the anonymity of respondents and their institutions, I have detailed 47 the years of experience and the firm’s Assets under Management in data ranges but have left out city locations in Appendix A. 1.5 Chapter structure In chapter two, I discuss occasions of sensemaking and the process of frame construction with concrete empirical data collected from the semi-structured interviews. The chapter aims to refine previous accounts of framing in the organisation studies and sociological literature applied to financial markets. The changing regulatory and social structure of the asset management industry around the Market in Financial Instruments Directive II (MiFID II) is used to highlight that frame construction can take different shapes, based on the social actors involved in the sensemaking stage. The market environment changes and evolves over time and the chapter highlights the changing role of the investment analyst from frame-maker to a more neutral expert intermediary and data provider that plays a diminished role in the sensemaking process and frame construction. This tendency is also an emblem for the growing independence of PMs and the changing analytical inputs. In chapter three, I integrate the social process of decision-making in the analysis of the interlocking networks in financial markets and develop a holdings-based network analysis approach which aims to quantify material interest in financial markets. The chapter uses the concept of corporate interlocks in social network theory to approximate shared material interest in networks of co-holders of securities the ECB purchases. It is argued that corporate interlocks are a useful tool to examine different interest groups in financial markets. This approach serves as the basis for the network analysis of the ECB’s two most recent asset purchase programs presented in chapter four and five. Chapter four analyses the ECB’s ego network around the CSPP over a period of 12 months. The chapter examines the policy objective of the CSPP in testing whether the portfolio rebalancing effect took place as assumed and how the central nodes influence the overall network behaviour. It is concluded that the central nodes imitate the purchasing behaviour of the ECB and that density is increased as a result. Chapter five examines the ECB’s ego network around the PSPP and analyses geographical homophily in the government bond markets of the eurozone. The network analysis focuses on 48 different countries and blocs in the eurozone and the policy assumption that the program encourages cross-border lending is challenged. The chapter links the sociological notion of homophily with the concept of home bias in behavioural finance. Homophily is analysed in the context of ties between issuers and investors and between investors themselves. The chapter concludes with suggestions to combat fragmentation in the eurozone’s sovereign bond markets. Chapter six analyses the ECB’s communication with financial market participants. I argue that central bank communication is best conceptualised as sensemaking through information exchange in networks of active market participants. It is shown how this communication is a social process led by the ECB in organised meetings of experts in the BMCG and how sensemaking is incorporated in the decision-making by the asset management elite. The chapter outlines how the ECB is involved in what Weick (1995) referred to as the early stages of sensemaking and describes the ways in which co-optation unfolds both from the ECB and its interlocutors. In the context of this study of the ECB, chapter seven explores the process of sensemaking and frame construction through data recorded from interviews with senior PMs in Frankfurt, London, Munich, Singapore and Hong Kong. The interviews focused on how communications from the ECB were interpreted in sensemaking processes and how frame construction and decision-making was shaped by these communications. The hypothesis of the rebalance effect from chapter four and inferences from the BMCG analysis in chapter six is also tested and inductively explored. Chapter eight summarises the findings and its contributions to the literature. 49 2 Sensemaking and Investor Frames 2.1 Introduction PMs are key decision makers in allocating capital for the ~$74tn large asset management industry and it is therefore important to study how decision-making processes are structured, shaped and influenced. Active portfolio management strategies, those that seek excess returns above an index, comprise ~70% of global assets under management (Blackrock, 2018), highlighting the significance of the discretionary decision-making by PMs. In contrast, passive funds replicate a chosen index without active decision-making. As witnessed in the aftermath of the GFC of 2008, investment decisions of a few individuals can affect society as a whole and as a result, new financial regulations were designed. The regulatory framework of the Market in Financial Instrument Directive (MiFID) of the European Union and implemented by the European Securities and Markets Authority (ESMA) is aimed at increasing transparency and fairness in European financial markets. The second phase of the directive, MiFID II, was implemented in January 2018 and is having a ripple effect across the European asset management industry affecting not only PMs and their employers, but also the investment banks servicing them. This chapter examines how PMs make decisions and how these processes can be conceptualised in organisational theory. I introduce the investor frame as a conception that can be used as a theoretical framework for further study in the asset management industry. Accounting and regulation have a real impact on decision-making processes. I seek to highlight the processes of adaptation to accounting changes that are taking place and how this shapes interactions between the major actors in financial markets. PMs combine operational, technical, analytical and mathematical skills in their profession. Typical asset classes of responsibility are in fixed income, foreign exchange or equities with different geographical foci. While responsibilities may vary significantly, ranging from trade execution to financial statement analysis, the PM’s main responsibility is to manage a portfolio of eligible securities in a pre-defined investment universe. PMs receive support from buy-side analysts, analysts employed at the asset manager, and external analysts, strategists and economists (either from the sell-side investment bank research or independent research providers). Traditionally, PMs also visit corporate management of target 50 investments to gauge business trends and discuss industry developments and financial positions. Likewise, sell-side analysts regularly meet companies under coverage, participate in corporate conference calls, build financial models, travel with corporates during investor tours, organise investor trips to see management, write regular research reports, host investor calls and send updates to clients. Sell-side analysts have been portrayed as important financial intermediaries (Brauer & Wiersema, 2018, pp. 221–222), influencing decision- making both at a corporate level (Cohen et al., 2012; Wiersema & Zhang, 2011) as well as at an investor level (Beunza & Garud, 2007; Giorgi & Weber, 2015; Zuckerman, 1999). I seek to highlight the processes of adaptation to recent accounting changes that are taking place and how this shapes interactions between PMs and sell-side analysts employed at investment banks. This chapter aims to contribute to the literature in the following ways. First, the findings challenge the notion that portfolio management decisions are shaped by frame makers in particular situations. A more sustained application of frames in investment decision-making is in the form of an investor frame, a cognitive framework that guides decision-making processes of PMs in day-to-day situations, and that is a direct product of sensemaking processes conducted by PMs themselves. Second, using new empirical data, the chapter shows that the organisation of the asset management industry is changing from how it was depicted in previous empirical accounts, dwelling particularly on the declining influence of major financial intermediaries in the processes of sensemaking. Third, the findings contribute to the investment analyst literature and show that the regulatory and accounting changes of MiFID II alter the role of the investment analysts servicing PMs, from a frame maker to a more neutral expert intermediary. This declining influence then enhances the independence of PMs in sensemaking processes which has not been the focus of previous research on analyst audiences. The chapter is structured as follows. Section 2.2 summarises the contributions and challenges of the literature on framing when contextualising investment-decision-making empirically. Section 2.3 introduces the MiFID II regime and describes the implications of the policy for the asset management industry. Section 2.4 utilises the research findings from semi- 51 structured interviews to describe the current market structure in which decisions are being made and how the accounting changes in the MiFID II regime are impacting decision- making. Section 2.5 summarises the changing role of PMs in financial markets, the occasions of sensemaking in everyday interactions and the resultant process of investor frame construction. I then conclude on the usefulness of further research on the concept of investor frames. 2.2 Framing in the analyst literature Framing and frames have taken different shapes across the literature in management studies, economic sociology and organisation studies.29 Most scholars cite the initial work on frames in Goffman’s work as a starting point, focused on the guidance of behaviour in situations where individuals congregate and interact. Goffman assumes that situations are: Built up in accordance with principles of organization which govern events - at least social ones - and our subjective involvement in them; frame is the word I use to refer to such of these basic elements as I am able to identify (Goffman, 1974, p. 10). These frames organise the experience and involvement of the individual partaking in the situation. Goffman’s (1963) earlier work developed out of the notion that when people congregate, they become part of a particular situation. The basic idea is that frames shape situations by assumed and implicit patterns and frameworks. This is both done through words and what was earlier developed in the concept of tacit understanding (Goffman, 1969; Schelling, 1958) and mutual monitoring (Goffman, 1963). He stated that “we can hardly glance at anything without applying a primary framework, thereby forming conjectures as to what occurred before and expectations of what is likely to happen now” (Goffman, 1974, p. 38). Michel Callon (1998a) famously discusses framing and overflowing, externalities entering the frame, in a market context as a process which brings into existence a system of relationships between agents who reach a harmonious accord through exchanges in said market. Framing also defines the effectiveness as each agent can take into account the viewpoint of others when making a decision. Framing is also the norm, desirable and statistically predominant. In another essay, Callon states that 29 See Cornelissen & Werner (2014) for a review. 52 what the notion of externality shows, in the negative, is all the work that has to be done, all the investments that have to be made in order to make relations visible and calculable in the network. This consists of framing the actors and their relations (Callon, 1998b, p. 17). As expounded in the discussion of the theoretical framework in chapter one, I treat investor frames as cognitive frameworks that guide individual decision-making and are based on past socialisations, experiences and occasions of sensemaking. In the research on accounting and finance, the focus has been centred on the frame makers in financial intermediary industries, most notably investment analysts and the framing of single securities (Beunza & Garud, 2007; Zuckerman, 1999), framing in accounting (Brivot, Himick, & Martinez, 2017; Vollmer, 2007) or framing as an analytical tool to research epistemic communities in accounting (Himick & Brivot, 2018). Of particular importance to the argument on framing is the analyst literature in economic sociology and organisation studies, reviewed below. Zuckerman (1999, 2000) highlights the role and influence of securities analysts on the value of shares under coverage. Most sell-side analysts focus on a specific subset of the market, namely companies that can be categorised into different industries and sectors, on which they make investment recommendations and estimate earnings (Zuckerman, 1999, p. 1407). Securities analysts take an important role of mediating between the share-issuing company and the PM making investment decisions as an institutional investor. Analysts would meet frequently with both PMs and company management or investor relations staff and indirectly convey their views to listeners and participants of the earnings conferences by asking specific and pointed questions (Useem, 1996, pp. 72–76) and are sometimes able to re-frame negative financial results (Useem, 1996, p. 73). In an analysis of Amazon shares, Beunza and Garud (2007) further developed this argument. They noted that sell-side analysts’ framing of companies does not only consist of placing the company in a designated industry, but also involves the application of company specific metrics and ratios to value and aid analysis of the shares, so called ‘calculative frames’. They thereby introduced the notion of analysts as frame-makers in the adoption of analysts’ calculative frames by their audiences. Contradictory frames applied to companies can co-exist at the same time leading to 53 incongruous investment recommendations and resulting in ‘framing controversies’ (Beunza & Garud, 2007, p. 29). Up until then, a polar opposite point to Beunza and Garud’s frame- makers in the literature on analysts, portrayed securities analysts as neutral experts acting as information intermediaries in a monitoring and data collection role (Jensen & Meckling, 1976; Womack, 1996). Beunza and Garud (2007) highlighted the fact that multiple calculative frames can co-exist until an unsuccessful frame is dropped in the framing controversies and that indeed, institutional investors gather diverse perspectives from a multitude of reports. However, the research failed to address actual portfolio management decisions based on the investment recommendations and to what extent analyst reports influence actual investment decisions. Giorgi and Weber (2015) addressed the audience reception of investment reports in their study of analyst reports on the pharmaceutical industry and the resultant analyst rankings. They also highlighted the multiplicity of framings that can occur in the investment advisory business and that individuals will process and evaluate this discourse based on their personal experience (intra-individual) and in relation to framing of others (inter-individual) (Giorgi & Weber, 2015, p. 337). They also contend that investment analysts can become frame producers by being consistent in their recommendations or by contributing novel ideas (Giorgi & Weber, 2015, p. 341). Their findings support the view of the framing literature of analysts as producers of discourse that frame information, particularly when the framing resonates with the audience’s understanding of the investments. 2.3 Asset Management and MiFID II The Council of the European Union’s MiFID I directive of 2004 had its central aims in enhancing integration, competition, efficiency and investor protection in European financial markets, overseen by the European Securities and Market Authority (ESMA). The occurrences of the 2008 GFC gave additional credence to the directive: (4) The financial crisis has exposed weaknesses in the functioning and in the transparency of financial markets. The evolution of financial markets has exposed the need to strengthen the framework for the regulation of markets in financial instruments, including where trading in such markets takes place over-the-counter (OTC), in order to increase transparency, better protect investors, reinforce 54 confidence, address unregulated areas, and ensure that supervisors are granted adequate powers to fulfil their tasks (European Council, 2014). The MiFID II directive was born out of the repercussions of the 2008 financial crisis, the need to protect investors and to improve transparency. It was implemented on 3 Jan 2018.30 The directive spans wide ranging aspects of financial markets instruments including conduct of business standards, client fair dealing, due diligence obligations, best execution, information requirements, pre- and post-trade transparency and so on. The need to account for research expenses separately under MiFID II impacted the interview respondents in this study significantly. Under general principles and information to clients, it was specified in Article 24 (4) (c) of MiFID II that all costs and associated charges must include information “relating to both investment and ancillary services, including the cost of advice, … the cost of the financial instrument recommended or marketed to the client and how the client may pay for it” (European Council, 2014). How to account for research expenses affected respondents significantly, particularly the need to expense and list the cost of investment research separately to clients. Previously, investment research was provided in exchange for institutional investors generating trading revenue (commission) for investment banks. Then, from the enforcement of the MiFID II directive from 2014 onwards in Europe, more ‘unbundling’ was encouraged throughout the industry to slowly prepare for the MiFID II implementation, i.e. trying to account for the cost of trade execution and investment research separately. From the beginning of 2018, funds had to begin listing research expenses separately. Passive asset managers Blackrock (see S. Jones, 2017) and Vanguard, where research expenses are comparatively low given their index replication strategy, led in adopting the expense of investment research cost as opposed to charging the client for it. Some active managers quickly adopted the expense. Understandably, some asset managers with a focus on active strategies were initially reluctant to expense research costs, but in a sense were forced into this accounting change, given the brand awareness and asset size of the passive managers (see e.g. Mooney, 2018b). AMCs can either pay for research expenses through their own profit and loss or establish a Research Payment Account (RPA). As brought up in conversations with interview participants of AM7, the latter option involves establishing an RPA model in which an annual 30 See details at https://www.esma.europa.eu/policy-rules/mifid-ii-and-mifir [accessed 02.04.2020]. 55 research budget is established in advance and through which the cost to each client is disclosed. Given that fixed income investing is not as labour intensive compared to equity research, the ESMA acknowledged themselves, the current lack of established market practices and methods for fixed income research “…currently FICC [Fixed Income, Currency and Commodities] markets do not currently have explicit execution commissions and mechanisms that allow research charges to be deducted alongside transaction fees” (ESMA, 2019, p. 67). ESMA (2019, p. 67) also highlighted the commonalities between macro- economic research reports and fixed income research and that this remains a grey area on where to account for this cost. Potentially, issuer companies could pay for this, in which case the analytical input would be a minor benefit. Hence the research policy formulated by AMCs would need to articulate how research inputs contribute to better investment decisions. This is a particular poignant issue as PMs will have to explicate in what shape certain research inputs affect their decision-making processes. On the one hand, the PM will want to highlight the internal analysis and her investment process which serves as product differentiation to peers, while on the other hand, the PM wants to ensure that she has access to research reports. With the exception of only two notable fund houses with businesses in Europe, the world’s largest asset managers have adopted the expensing of research cost, rather than charge the client from the Net Asset Value. Even Ashmore, the largest active Emerging Market AMC in Europe, went ahead to adopt the research expense in their own profit and loss.31 Most notably, towards the end of 2017, eight major asset managers were still going to charge research cost to the client but have since reversed their position, leaving Deka and Carmignac as the sole funds that charge clients for research related costs (Mooney, 2018c). While relatively innocuous from an outsider perspective, this accounting change has spurred massive redundancies at investment bank research divisions and outsourcing of research from investment banks to external providers. The other aspect of investment practice that has changed as a result are company-fund manager and fund manager-analyst meetings. The way investment banks were able to push their material interests in the past, their ‘book’ and investment ideas, has been reversed. 31 Active fund managers usually have the highest research expense given the need to outperform on a relative as well as absolute basis in order to attract clients, see Mooney (2018a). 56 On the one hand, investment banks are facing falling revenues due to a variety of reasons: the inability to take balance sheet risk and regulatory restrictions in conducting business. On the other hand, institutional investors, especially PMs surveyed, face difficulty in justifying higher expense ratios to account for the investment research cost amid falling average management fees. This has been spurred by lower management fees in the passive fund management sector, especially ETFs (see e.g. Balchunas, 2016). Interview respondents at AM7, AM8 and AM9 highlighted, that MiFID II spurred the formation of a separate team dedicated to addressing this directive. A budget for equities and fixed income is decided separately and senior managers allocate resources to which investment banks or independent research provider they prefer. 2.4 Investor meetings and the changing landscape of investment research The framing literature discussed in section 2.2 requires frame makers to co-ordinate market participants’ behaviour and expectations to influence investment decisions. In recent conversations with PMs, the physical aspect of coordination in the framing literature has become increasingly complex as less face-to-face interaction takes place given the cost of arranging meetings. Investment banks, providing investment advice to PMs, used to host investor conferences, where speakers, investment themes and corporate management were selected by the banks themselves, but this seems to be slowly eroding in Europe as clients under the new MiFID II regime would have to pay up to $1000 per day to participate.32 Corporates host investor calls and press conferences live online and with the increasing transparency of information and availability of data, market participants find new ways to follow and monitor these events. For PMs surveyed, MiFID II equates going ‘back to basics’, i.e. doing more analysis in-house and doing financial statement analysis themselves. The industry has seen significant voluntary and involuntary departures from investment banks’ research divisions, as well as outsourcing investment research services to external research providers.33 32 Low paying clients (e.g. under $100,000 annually) would be able to attend group meetings with corporates and attend the main conference track in the past, both in Europe and Asia. New ‘price lists’ are still variable at the moment according to interview participants and banks would first get a firm commitment before hosting an investor conference in London or Frankfurt. 33 See e.g. UBS Evidencelab. 57 I: So it means you don’t see that many analysts anymore? So it's all internal now? HF1a: Yeah it's about going on the website, looking at the annual report, presentations, speaking to the company, back to basics. I: ...but I suppose you do have access to broker research right? HF1a: You do, but the quality of the research has become useless. The good ones left the industry and gone to the buy-side. So, the point I was making about MiFID II is you end up with more under-researched stocks, where you probably going to add value. (HF1a interview, June 8, 2018) This fall in analyst coverage and under-researched stocks is something that was noted elsewhere (Hannah Murphy, 2018). In this context, it is interesting to see that PMs are increasing their internal analysis to fill the void of investment bank analysts, who were previously in a position of acting as influential frame makers (Beunza & Garud, 2007; Giorgi & Weber, 2015; Leins, 2018; Zuckerman, 1999). Indeed, PMs I spoke to consider the main analytical input of investment bank research as helping to gather ‘what consensus thinks’ rather than to gain specific insights into a situation. That also means it is less important which analysts you meet or where you are based to construct an investment thesis. AM1: I still get all the research. It's too much information, actually, so the internal analysis helps. We have internal calls and also internal risk management who tell you what impact various scenarios have. (AM1 interview, August 16, 2018) AM2: When I started in the 1990s there was only CEIC, and without internet, everything was physical. If you didn't have it in the library, you didn’t have it. We had to get the physical annual reports. Now I can get everything through Bloomberg and it’s easy to model companies or manipulate the data. Take the reserves [we discussed his analysis of foreign currency reserves of various Asian countries] for example, I just use the info to gauge consensus. (AM2 interview, August 17, 2018) 58 HF2: Now it’s just a 30 minute meeting on Monday morning, and one call with the economist during the week. Otherwise we are left to our own devices. (HF2 interview, June 13, 2018) During the early 2000s, when a PM passed a trade to a sales trader or salesperson, it was often done in conjunction with a lunch in the city or connected with a conversation on a recorded line. It was also done from person to person with the knowledge that business was given from one to the other. This was important as the PM may have to, in the future, rely on the salesperson for other favours such as help with a job search. Towards 2006/2007, the Financial Information Exchange (FIX) gained in popularity and the Bloomberg terminal was able to include this add-on software. Prime brokers, including Morgan Stanley were encouraging usage of it. With FIX, PMs and traders were able to directly pass trades to the broker and monitor the trade execution in real time.34 Previously, one would have to get manual updates via either the Bloomberg messaging system or email. In my conversation with a sales trader from a large French Bank, it emerged that almost all equity clients are using FIX now. It removed a lot of the inefficiencies of either calling or meeting up to give instructions on the trade.35 It also transpired that FIX, alongside the use of Blackberries, helped facilitate the move to cut out the human, soft conversations associated with placing trades. Even the account review, which used to be an occasion to treat the PM to a lunch, has been rationalised. 36 Furthermore, the introduction of MiFID II has cut these social links between PMs and their brokers in other ways significantly. Participants described the awkwardness under MiFID for PMs of trying to avoid getting invited for lunch given the need to declare this to their compliance officer, and likewise a reciprocal awkwardness by the investment banker not wanting to appear rude. The new generation of PMs having started their careers after the GFC of 2008 are thus far removed from these old practices. The introduction of MiFID II also brought about changes for the analyst calls and research trips. Analyst calls used to be a combination of a pure courtesy call and an update on various developments in the sector and stocks under coverage. They are now billed by the hour. For research trips in the late 1990s and early 2000s, the broker would travel with the PM to the country of destination and attend the same company-fund manager meetings to both extract 34 For more details see McEachern Gibbs (2010). 35 It was still very common to get boasting rights in the 2000’s when writing a ‘ticket’ with the client’s trade after a long lunch with said client. 36 A review in which the sales-person and PM discussed service needs and also review commissions paid and potential commissions to be generated in the future. 59 information that may be relevant to other clients, but also aid in sensemaking processes of the attending PM, show face with the corporate and strengthen the relationship as intermediary. In other cases, the covering analyst would join the meeting and aid in the technical analysis of issues at hand. The implicit role of the broker was to aid in sensemaking processes, help the conversation with management when the PM ran out of questions, pitch the analysts’ recommendation in casual conversations in the elevator or in the car. Later in the mid 2000s this slowly changed and the local salesperson in the country of destination would take the PM to meet corporates. More recently it has become common for the broker to only arrange the company-fund manager meeting and the transportation as sales desks have been systematically reduced since 2008. Under the new MiFID II regime, PMs increasingly arrange their own research trips as they would otherwise be charged exorbitant fees. In a discussion about corporate access, a PM made the following deliberation. HF1a: Yeah.. so that is a big business cause at the moment brokers are charging $1000 a day to join a trip. You say, hold on I pay for my ticket, my hotel, how is that worth $1000? Well, break down the cost. You know it’s not like I'm having lobster every lunch time whatever. It's not like I'm going in a limo to these meetings. Uber can do that for me. So, you break it down and then, and by the way I'm paying you the salesperson to come with me, you don't need to be there. So you're just someone to call up to arrange a meeting. It's like 20 pounds. 5 a day that’s 100 pounds. If at all and you wanna charge me a thousand? These guys are coming in and saying 200. We'll do it all for you for 200. and arrange the drivers, just call an UBER. Anyway, it’s assuming 500$ and if you are in Africa or Brazil, 20$ is enough. Come one you need to adjust it right? So, the whole industry is changing. I think the Europeans haven’t woken up to this. Because they're the ones that are really going to suffer. They're so used to getting information. (HF1a interview, June 8, 2018) AM8 clarified that if a client generated annual commissions of $200,000 in the past, the investment banks are now trying to generate the same commission on the various services they provided in the past by breaking down this revenue stream into the various services, including arranging conference calls, company-fund manager meetings, fund manager- analyst calls and financial model updates. PMs surveyed are now increasingly arranging their 60 own research trips and analyst contact has been reduced commensurately. Thus, the dynamics have seen a significant alteration in that the investment bank now has less opportunities to shape PMs’ decision-making and research becomes a PM-controlled sensemaking process. Fixed income analysis by its nature is arguably less research intensive and does not require such extensive travel as traditionally in the equity space. Surprisingly, trading in fixed income markets is still largely manually operated. This is partially due to the fact that as opposed to equity markets at exchanges, investment banks often have to warehouse bonds on their balance sheet and actively make markets.37 Zoeb Sachee (2016), Citi Group’s head of European Government Bond trading, also noted a decline of electronic trading in European fixed income. This was echoed by respondents: AM7a: Maybe there's a difference to the equity market, because credit markets are more illiquid and not traded on the exchange, so it really makes sense to be a fundamental or active manager in corporate bonds because I think it’s much more illiquid and information inefficient, in comparison to the equity market, just because we have so many bonds outstanding, different capital structures, very illiquid instruments, there is always some under valuation and overvaluation in the market that can be traded. (AM7a interview, October 5, 2018) The way that investment banks used to have power to shape markets was both through their analysts and through their proprietary trading unit. Recent developments with the introduction of MiFID II in January 2018 have brought with it explicit costs for advisory calls with analysts as discussed. PMs surveyed felt a decline in the quality of research and questioned the change in business model. AM6: Before you had SocGen, Deutsche Bank, HSBC, Barclays, you had Citi, you had UBS, you had BOAML, Morgan Stanley, Exane, you know them all yeah… In the end, if somebody asked you what is the important part, you would say all of them. 37 Goldman Sachs introduced GSessions in an attempt to aid automated electronic trading to emulate what had happened in equity markets, but they later pulled out given slow adoption (see Alloway & Mackenzie, 2014; Burne & Turner, 2015). Bloomberg started giving an idea of bid offer sizes in their electronic bond trading platform, but the specific counterparty still needs to be contacted first and fixed income trading remains highly inefficient, according to our respondents. 61 You need all of them. Then once you pay, you ask yourself, what am I actually reading, what is important for me, and why do I need it. In a way it helps you to be more focused, be more cutting edge, and not, like, keep all the garbage there. So have I seen a significant increase in quality of research? No, not really. (AM6 interview, October 5, 2018) Experienced asset allocators we spoke to did not pay too much attention to investment bank research anymore as the quality and time horizon deteriorated and shortened. A significant portion of the heads of research and senior analysts have left the sell-side for other opportunities. A former star-rated analyst by the Institutional Investor magazine put it this way in our conversation. I: Do you feel they [investment banks] still have the power to shape the narrative? AM5: That is mostly gone now. They have become very reactionary. There are also no views on the long end of the yield curve at all and longer-term outlook of how things are going to be in the eurozone. There are less people around as well. The time of them trying to shape the narrative is over, I mean the idea of top-rated analyst on the street is gone, another reason why I left. (AM5 interview, October 1, 2018) When PMs surveyed conducted physical fund manager-analyst meetings or conference calls, it was done as part of a fact-finding mission or confirmation of the investment thesis. Given that the PM will have to carefully consider which questions to ask and which sets of data to get advice on, the meeting has acquired more purpose and a clear structure in terms of time and regularity. Decision-making then relies on the level of understanding of a certain situation. A diverse number of actors can be contacted in the sensemaking and data collection phase of the PM. In other words, PMs contact experts to understand a situation better, rather than to get advice on what decisions to make. According to AM7, out of a $70m research budget for 2019, the fixed income proportion was only 10% of the total budget, while equity research makes up the rest. It indicates that fixed income PMs have to rely more on their own methods and analysis. This is also illustrated by the comments of AM9 on investment methods. 62 I: But, ehm, do you see that as a big change now? AM9: Changes drastically, yeah MiFID has changed the structure a lot. Personally for me it doesn’t matter so much because…the current institution where I work have good agreements in place, which means I am not losing out on much of the research and I have access to and I don’t consume a large amounts of research, because as you say in fixed income, I have my own models, I do my own homework. If I don’t form my own view then I cannot check how it’s wrong or how it develops, right? If it’s somebody else’s. So personally, for me has it changed a lot? No, other than it’s created some administrative headaches, you know where I have to log a call and do this and do that…For a lot of buy-side places it created the idea that you need more in-house research. (AM9 interview, January 18, 2019) This may be elaborated in a simple illustration. A PM investing across different asset classes in Europe contacts a previous ECB policy maker for conversations. The conversations are about the Greek debt crises, what certain politicians are planning to do, what their interests are and what shape QE could take. It transpires that the planned QE program will be back- end loaded and more impactful than previous QE programs. This conversation does not serve to obtain decision-making advice, or actionable data points, but serves as what investors would know as a ‘channel check’. These are occasions of sensemaking in which the PM acquires new information about the investing context and which she brings up at an internal investment committee meeting. These new data points are then reconciled with other data points in the group and made sense of in the committee meeting, after which she then potentially decides to adjust her overall investor frame to being long Greek government bonds. Such changes would then be communicated to clients in the monthly newsletter or market views detailing changes in asset allocation. As Barker et al. (2012) concluded, information cannot be categorically defined as actionable and non-actionable, but information and data forms a tacit understanding of a specific context. Hence, the information gathered at meetings needs to be put into the context of the investor frame and can result in a multitude of decisions. A key tenet of tacit understanding is that individual market participants carry their own respective frames and base decisions within this frame. Data sources and interlocutors are hence not only analysts from investment 63 banks but can be derived from expert networks outside the investment community. Another respondent elaborated on company-fund manager meetings and fund manager-analyst meetings: HF1a: It's more an understanding of what's going on. A lot of it is getting questions from them about the industry they're in. You wanna know who's your biggest competitor, unlisted… So there is a lot more inferring questions to the company about things that can be about our business model, so is this gonna be the 5, 10 year winning company with the right business? Well the only way you're going to know that is not by looking at their model but within the industry as well. And if it's unlisted, you know… so it's not insider information just sensible questions to try and understand ok…Who is it, who runs it, what's the pricing strategy, how do you differentiate yourself against the other people.. (HF1a interview, June 8, 2018) Investments and cues relating to such investments are viewed through the prism of the institution and from the investor’s point of view rather than a focus on the investment object itself. It follows that the reputation of an analyst is not the key deciding factor on the importance of a stock recommendation (situational), but each portfolio holding within the overall investor frame. That is the reason why single investments, such as Amazon shares, can have a diverse investor base as the security falls into divergent investor frames, ticking different boxes at the same time, rather than the calculative frames and categories proposed by the frame makers as Beunza and Garud’s (2007) suggest. The point for a company-fund manager is not to gain insider information, but PMs surveyed used company meetings as a method to confirm different tenets of their investor frames. The findings confirm the points made by Barker et al. (2012) and Taffler et al. (2017), that company meetings act to confirm the investment thesis of the PM and as a result, can help reduce anxiety. Company-fund manager meetings help to contribute to a tacit understanding of the context in which a particular company operates in. Conversations are rich and put into the context of specific industries, sectors and macro trends. Interview participants elaborated on the ways in which investment banks used to frame the discourse through research analysts representing ‘their own book’ and also by reflecting the 64 investment bank’s own positions in proprietary trading. To avoid a conflict of interest, investment banks set up firewalls to prevent analyst recommendations to be influenced by the investment bank’s positions, but respondents underlined the duplicity of the investment banking business model. According to respondents, the investment bank was actively competing with their clients. HF1b: So in the old days, let's contrast pre-crisis with post, before the financial crisis up to 08/09 particularly before Dodd Frank, investment banks could take large prop positions right? So you could have a good market maker in JP Morgan or Barclays or any other big bank, could go to his head of risk, and say look, here I am going to trade the Brazilian election, say January 2018. Election's in October, 6th of October. This guy is a very good trader, proven track record, bank trusts him and he says, give me a billion dollar balance sheet…on the view that after the elections, the market is going to rally again. He would be able to take 6 months or a year view. This has completely changed. (HF1b interview, September 25, 2018) As HF1b put it, investment research has become short term as the economic processes and profitability had been adjusted through regulation, both from a balance sheet perspective and also what sales traders can actually tell clients even in the FX and Rates space. IB1a: There have been increased regulations that hit the market, like MiFID II, the Volcker Rule, and so on, due to regulation based changes there are a lot of restrictions of what we can tell clients. Clients don't trade bid away [bid at a higher price], compared to 3 to 4 years ago. They couldn't buy any paper above par, which they only later changed. (IB1a interview, August 20, 2018) HF1b elaborates on the connection between material interest and actual shaping of research question tackled by sell-side analysts. This did not only include single securities by corporate issuers, but also government securities under the coverage of rating agencies. HF1b: So, cause the traders in the investment banks wanted those longer term views, the research was just paid from trading profits, very important to understand that: 65 Research is paid by trading profits. So research must do what traders want. Whatever helps traders make money is what the research will be instructed to do. So, because of that, cause the trading could be long term, so the research could be long term and therefore it was relevant to us. So we would happily use investment bank research, but because it was inadequate in its coverage, we have always done our research inhouse in addition to that. Which basically means that all our traders, we don't… we embed the research function into each portfolio manager [emphasis added]. (HF1b interview, September 25, 2018) This deliberation shows how AMCs are now shifting the decision and analytical frame of their investment decisions to the PMs themselves. For the sell-side, the profitability of research is not being driven by the investment bank balance sheets anymore, but more so on conference calls with clients. As a result, the analyst turned into a consultant on a paid one hour call, rather than a frame maker, expounding on calculative frames from which the firm benefits. The PM’s view of this change in motivation and the content of the fund manager-analyst calls becomes less pertinent and framing is displaced by information gathering. I: …So most of the analysts you talked to, the strategists? HF1a: Gone. And a lot of stuff is, you know how it works right? The quant screen, everyone can do the quant screen. Data availability is not a problem anymore. They're not adding value to that data. So, since the MiFID started I just speak to companies directly. (HF1a interview, June 8, 2018) PMs surveyed are becoming increasingly sophisticated in how to make sense of certain analyst calls, the content of information emanating from specific analysts, and how to contextualise this in asset allocation decisions. In a discussion about taking active risk vs passive investing AM9 elaborated on the position of the analyst trying to make an ‘active’ recommendation. AM9: I would be personally more careful how I would invest my own money in this environment…you saw that I think with a lot of the equity performance from October 66 to the end of December last year…if you are an analyst and you take the active decision of underweighting Apple, you taking a big tracking error risk. 8% is a large position, but that’s the most you can do probably if you benchmark. So, if you are a passive fund you don’t do that right? And then all of a sudden if your signals, if you are systematic, and your signals go and then trigger selling and that begets more selling, and then the passive funds come in because flows start to ebb and come out. So, it just amplifies reactions in markets which is not a healthy dynamic to me, that’s how I would classify. (AM9 interview, January 18, 2019) The contextualisation of individual sell-side recommendations into the proposed conception of the investor frames enables the researcher to move on from specific situations in which analysts apply frames to their securities or industries under coverage and allow for divergence and coherence across the PM’s complete enacted environment, the portfolio of securities. It allows for coherence and sensemaking for the PM herself in an increasingly complex environment. 2.5 Portfolio Managers’ sensemaking and investor frame construction Building on the research data in section 2.4 and refining the literature discussed in section 2.2, the conception of the investor frame introduced in this chapter focuses on the PM (the decision maker) rather than the analyst servicing the PM. Interview data in this chapter showed that frame construction is actually performed by the PMs themselves and that analysts are becoming less important in shaping investment decisions. An onlooker can observe a situation and devise her own framework as to the context, likely possible courses of actions and future outcomes of the situation. Akin to the onlooker, PMs observe market developments and enter specific situations with deliberation and intent driven by their own cognitive framework of decision-making. Hence what is of relevance to the investors’ frame is not necessarily situational, e.g. particular calculative frameworks of specific companies, but embedded in the individual and institution’s historical context. In this study, the ‘inter- individual’ aspect of framing becomes a matter of PMs wanting to gauge consensus views and I contend that frame construction unfolds from the decision makers’ point of view ‘intra- individually’, not from the securities analyst’s view as in the prior literature. Analysts are just 67 one counterparty in the PM’s activities. Occasions of sensemaking help PMs develop decision systems and reduce ambiguity. Weick (1995) stipulated that meaning is created by weaving past experiences of socialisation together in the form of frames; an accumulation of past socialisation experiences. Starbuck and Milliken (1988, p. 59) also contended that antithetical processes are inherent in sensemaking frameworks, oftentimes contradictory, and frames can be adjusted to the situation where needed. Frames are thus adjusted, from time to time. A main determinant of the longevity of the investor frame in the asset management industry depends largely on its economic success, PMs surveyed made clear that validation of the investor frame requires economic profitability. PMs are often in a position in which they exercise discretion in managing their portfolio within an investment mandate. This is often not limited to a single investment, but decisions are made in the context of a portfolio of investments and a broader view on the overall market. For example, the PMs surveyed can ‘sit out’ certain situations promoted by investment banks. The interview data in this chapter link the investor frames with the material interest of market participants. People not only tend to ‘talk to their own book’, but also regularly fine-tune methods. A more quantitatively oriented and highly numerate PM, HF3, in this study elaborates on his investor frame. HF3: Quant funds now operate like a casino, we have 15 strategies running at the same time…some strategies work for one month, some for six months, then usually an algorithm captures the outperformance and you have to move on. (HF3 interview, August 16, 2018) HF3 mentions that he does not understand any ‘fundamentals’, but what he does is build strategies inductively and then test deductively. He elaborates that one of out of the many simultaneous strategies follows such principles by playing a merger and acquisition event. After the computer monitors the deal flow and news from a Dow Jones data provider, the computer quickly executes a buy order in the acquired company and a sell order in the acquirer. He mentioned that this worked for a few months, however then new algorithms 68 have taken advantage of this strategy and it stopped working.38 Another such strategy was the index inclusion of a security, specifically, the point and day at which passive funds would need to purchase said security in their portfolio. HF3 hence sees investing and financial markets as a series of ‘games’. At a dinner, he mentions to me that writing code is easy but you have to be creative to come up with new games. Various forms of the short volatility trade worked very well for him in 2018. In other words, the trend of global central banks removing the risk of a market collapse brings opportunities across different securities, such as through the VIX39 directly, or stock options, warrants etc. The computer would monitor volatility spikes across thousands of instruments and would sell when unusually high and square the position after the spike to book gains. Another example of a different investor frame would be an interview participant who specialises as a so-called Macro PM. Fluent in multiple languages in the region he invests in, he reads all local newspapers in the mornings. Frequent trips across Europe include meetings with various regulators, government representatives, political journalists or attending conferences. The meetings extend normal work hours and the office, but also take place at dinners or luncheons. From this context he forms views of his investing universe and tries to find ways to express these through multiple instruments or securities. For instance, looking at the austerity measures across Europe, he gauges which governments are performing better and starting to pay off their debt. While certainly monitoring levels of spreads or economic data series, the main analytical input revolves around the ways pieces in a complex puzzle fit together and how he interprets those. How different themes are traded is a creative expression in numerous instruments of the universe under coverage. For instance, being short Italy and long Ireland is one way to express the continuing trend of higher trending Purchasing Managers’ Indexes40 in the Western periphery compared to the Southern Eurozone. Regular new developments are discussed in internal committee meetings and made sense of with colleagues with different areas of expertise. Each PM goes away to draw their own conclusions from the meeting, expressing ideas in taking positions in different instruments by geography or asset class (e.g. spread trades in Korean rates, relative trades in Eurozone government bonds, long short equity positions, etc). 38 HF3 also mentioned that some so-called bots issued false takeover rumours on Twitter to capture resulting price moves from algorithms acting on this news. 39 The VIX is the index of the volatility of the S&P 500 equity index in the US. 40 See glossary. 69 Yet another example for an investor frame would be AM7a. A rigorous analysis for inherent value based on a handful of valuation metrics is complemented by the analysis of the particular business cycle of a company, such as a European telecoms operator, which is seeing good growth in data revenues given the increasing usage of smartphones. His investor frame is really complemented by a philosophical view of not ‘overpaying’ for any positions. Even during the quantitative easing phase, and given the philosophical outlook on capital markets, he tried to justify purchases with creative adaptation to the changing market environment where all assets seemed overpriced. Hence, the investor frame is a cognitive framework that guides information processing and the decision-making processes of PMs and is fine-tuned after occasions of sensemaking such as fund manager-analyst meetings, company-fund manager meetings, internal committee meetings or indeed for fixed income PMs, in gatherings with significant market actors such as central banks. Data points are placed into frameworks that make sense of events, new data and cues, and redress portfolio management decisions. “Managers literally must wade into the ocean of events that surround the organization and actively try to make sense of them” (Daft & Weick, 1984, p. 286). The discretionary aspect of the decision-making leads for competing interpretations to unfold in occasions of sensemaking, in which the PM questions her hypotheses and interpretations, and the investor frame is then constructed in her mind, using various selected data points, cues, analytical tools, rather than a discussion of competing arguments in a committee meeting. The investor frame’s uses are manifold. First, it indicates boundaries in which markets developments are seen, not limited to a particular situation but the overall market context. For instance, surveyed PMs in the corporate bond space would not buy a security if they cannot make a valuation argument to justify the purchase. Second, it justifies and legitimises decision-making processes by providing accountability. AM8, for instance, follow a REDACTION: Personal data removed for confidentiality reasons process to which each security selection has to adhere to, often making fast growing technology shares or related analyst recommendations frame-incompatible. The investor frame is also used in marketing material to existing and prospective clients to highlight the perceived robustness of the firm’s investment process and to differentiate the investment product from other competitors. Hence the investor frame takes precedence as a key marketing tool for AM8, even over company- 70 fund manager meetings which were used to demonstrate knowledge of companies to clients in Barker et al.’s (2012) study. Third, it helps market participants order and process an increasing amount of information by reducing the relevance and importance of data. It enables market participants to categorise and process the ongoing information flow and integrate diverse and conflicting sources of analysis. Fourth, it reduces perceived uncertainty and ambivalence. The findings showed that the context of the investor meetings is changing and that PMs increasingly pay for analyst meetings given the regulatory change of investment research under MiFID II. Thereby, the fund manager-analyst meetings and information exchanges become more controlled, as PMs direct the topics and form of the paid conference calls as sensemaking leaders. This authoring in investment decisions and the enactment of the PMs environment through portfolio implementation is the result of the implementation of an investor frame. As discussed in chapter one, the changing character of the fund manager-analyst meetings is best summarised with Maitlis’ (2005, p. 33) notion of guided organizational sensemaking, as the PM controls both the regularity, time, date, context, content and the level of input of analysts in the meetings. Senior actors such as e.g. AM7d and AM8 would have direct input in the percentage of the overall research budget to be allocated to a particular investment bank employing the analysts. The increased cost of the call focuses the conversation on specific topics. These exchanges aid the construction and adaptation of the PM’s investor frame and data is gathered to confirm and refine investment theses. Hence investor frame construction becomes a process central to each decision maker within the AMC. Asset managers in the study have organised the decision-making processes around one PM responsible for a given portfolio. On occasions of sensemaking with company management and internal/external analysts, the PM constructs and refines the investor frame used to execute and justify investment decisions. Using this concept in financial markets, portfolios of securities are expressions and artefacts of the investor’s enacted environment and are a product of the decision-making that has been guided by an investor frame. As Weick (1995) pointed out, as the individual acts and enacts it receives feedback from its environment. He states that “the enacted world is also a subjective, punctuated, bracketed world because it has its ‘origin’ in mental models of causally 71 connected categories that were part of the strategizing that carved out artifacts [sic] in the first place” (Weick, 1995, p. 37). The feedback provided in financial markets are the daily changes in security prices in the portfolio and the information circulating in expert networks in which PMs are socially embedded. As opposed to the CEOs and executives in Starbuck and Milliken’s (1988) examples, PMs receive instant feedback on their decisions and on the effectiveness of their perceptions through daily price changes in their portfolio. Hence using Starbuck and Milliken’s (1988) metaphor, PMs are more akin to batters in baseball who see the immediate consequences of their action. Figure 2.1 depicts this circular process that market participants undergo in their investment decisions. As discussed, the investor frame guides the portfolio implementation, investment decisions and the execution of trades. It is constituted of idiosyncratic measures and factors that derive from the sensemaking process. Through the purchases of portfolio holdings, such as bonds or stocks, the investor enacts her environment and creates interlocks with co- investors. Prices and information pertaining to the issuer (corporations) of these securities develop on a daily basis. These developments or sequence of events are then analysed in expert networks and sensemaking exercises. There are certain rhythms to these sensemaking exercises but as Weick noted, retrospection happens in certain intervals. Both price and information cues then lead to an adaptation or revision of the investor frame. These revisions, then, require the portfolio to be adjusted and changes to be implemented. 72 Figure 2.1 Decision-making processes in investor networks. Source: Author’s own elaboration. Fundamental investors across all asset classes believe in an inherent distinction between market prices and the intrinsic value of the underlying asset (Zuckerman, 2012, p. 229) and attempt to profit from an eventual convergence of market prices with the intrinsic value of the underlying asset (Graham, 1959; Whitman & Shubik, 1979). From a market perspective, fundamental investors should aid price discovery, liquidity and aid market efficiency: if an asset is mispriced, fundamental investors would attempt to arbitrage the profit. In stark contrast, passive strategies mimic the underlying assets of an index. Thus, no attention is paid to the underlying economic reality of the asset but it is bought on the premise of its membership to an index. Most active PMs will also have to ask themselves the question whether eventually, other participants will agree with the rationale of investments based on their investor frame as this will eventually aid the realisation of profit. The active manager inevitably relies on other market participants adapting their investor frame regarding a portfolio of securities to prove that she is ‘right’ about any mispricing. This simple illustration is of importance to frame construction. The passive investor is following what the creator of the index previously selected, thus the investor frame for the passive investor is based on the index constituents. An active investor, such as a fundamental Adaptive Frame Construction Portfolio Implementation Enacted Environment, Interlocking Network Formation Cues, price changes, Issuer developments, Information Flow Sensemaking Processes and Occasions of Sensemaking 73 investor, will exercise discretionary security selection. The PM is then tasked to form a strategy and implement it with the relevant instruments. Institutional investors use common industry matrix and calculations, such as e.g. Yield to Maturity (YTM), current yield, credit rating for bonds, or Price to Earnings, Price to Book ratios, Discounted Cash Flows for equity investors. Most of these ratios form an initial case for security selection for the portfolio to be built on and can easily be screened for. A main selection criterion is formulated and explicated in concepts, guiding the analysis and selection process. This selection criteria defines brackets within the universe of what is to be considered and thus investment targets are narrowed down. The selection criteria differ from PM to PM, it could be as simple as looking at low PE ratios or securities that fit into a certain macro theme. PMs at AM7 elaborated on the investor frame construction process in which data about market positioning is collected in order to contextualise the fundamental analysis. AM7a: I wouldn't say that. Just to explain how we work, I mean, we are bottom up fundamental credit investors, most research is on the issuers, the companies and business models and financial policies and things like that. On the other hand, we try to get the market beta, trying to understand whether we are in a widening or tightening path. That analysis is more looking at things like monetary policy maybe also technicals in the market, like the dealer access. That is also quite interesting because through the CSPP that is distorted, so you can see how the dealers are positioned. If you aggregate the runs they are sending out in the market. I: More qualitatively right, or not? AM7a: No, you can really look into every single bond and see if the brokers on aggregate are looking for bonds or are offering. Then you can aggregate this on a ticker and sector, currency, and then you get quite a good picture. What the technical situation in the market is. It’s very tight, so, traders are really looking for bonds. That’s a good sign the spreads are going to tighten. (AM7a interview, October 5, 2018) 74 Hence the inter-individual analysis is aimed at integrating the fundamental analysis in the market context. As AM7a elaborated, collecting dealer quotes, indicative positions and aggregate axes, PMs can ascertain liquidity conditions in the bond market. The axe is the interest a trader shows regarding a position that is on their books. She may be interested to hedge her position in case of unforeseen events. Hence this liquidity analysis depends on information from and on other market actors and their material interest. In her PM study of financial model usage, Svetlova (2018, p. 89) observes a similar inter-individual analysis in the ‘plausibility checks of consensus’. AM4: We have to worry about single issue exposure, single position exposure, concentration risk. Active managers can't justify if your positions are in the benchmark. With all these constraints I have to have way over 30 holdings. I: Do you analyse competitor positioning? AM4: I do analyse the fund flows, I do the demand side analysis and use domestic sources, the supply side is more about fiscal hedging. (AM4 interview, August 21, 2018) In order to observe market developments methodically, PMs use other market monitoring tools that help address ambiguity in the form of data series on a current market position. Market positioning can be understood as the momentary portfolio allocation of active market participants in aggregate. One such example is the Commodities Futures Trading Commission (CFTC) report which summarises futures positioning. Figure 2.2 gives an example of how this would look for any given week. The Z-Score gives an indication how significant the positioning is compared to the average. The data in Figure 2.2 shows that investors are bearishly (short) positioned in the British Pound, seen in the negative figure, and bullish (long) on the Yen, as seen in the positive figure. The CFTC positioning can be obtained weekly. These are data series that give a snapshot to contextualise current positioning and what the market is currently betting on. As an illustration, should the BOE unexpectedly cut rates, assumptions from work on expectation formation would indicate that the Pound would depreciate. However, given the positioning in Figure 2.2, the Pound may strengthen as PMs could take profits on their positions. Thus, the link between expectations and investment decisions is not a linear connection and information remains ambiguous. 75 Figure 2.2 CFTC Net Speculative Positioning (Futures & Options report). Source: Data from interview participant. One of the equity HF managers elaborated on this monitoring of market context. HF4: A rate cut can be seen as positive or negative by the market depending on where we are in the cycle, so we don't look at the press conference and statements live, but we use 'consensus economics' [talks about subscription and cost of the service], use this tool to gather what consensus is. A rate cut can not only be seen as a positive but also as a sign of desperation. Mostly the market will view it based on the cycle, not based on the increased transparency of central bank communication of the recent years. The BoJ had this going for a long time, for the ECB it's relatively new, from 2011 onwards. (HF4 interview, August 21, 2018) 2.6 Conclusions The chapter discussed how regulatory and accounting changes have affected decision-making processes in the asset management industry. The findings also offered empirical evidence on how the political economy of financial markets is altered by these changes. Investment banks and their analysts are losing significant influence on PMs’ investment decision-making processes. Hence, in addressing the research on analysts literature, the findings underline a change in the role of sell-side analysts from frame-maker to a more neutral role of information provider. PMs see individual recommendations within a larger context of market conditions which reduces the level of analytical input to portfolio management decisions. More research into which market actors influence investment decision-making at a PM level is thus of importance and this area remains under-researched (see Brauer & Wiersema, 2018). 76 Given that both fund manager-analyst meetings and company-fund manager meetings are becoming increasingly animated and highly controlled, PMs are commensurately more independent of situational framing and calculative devices. Instead, the investor frame has become a sophisticated cognitive product of social processes of sensemaking and how the PM sees the world based on past socialisations. The conception of the investor frame helps to theoretically bridge the divide between micro level behavioural patterns and the networks they are executed in. The fact that investment banks are losing significant clout in shaping financial markets exemplifies that financial markets remain a place for struggles between parties with different material interest. PMs can be seen as key actors in this environment and are actively adapting to the circumstances they are facing and are shaping these environments as a result. 77 3 Corporate Interlocks and Embeddedness: Towards a Holdings-Based Model of Financial Networks “Actors do not exist as social atoms. As a member of society, an actor exists within a system of actors and evaluates alternative actions within that context…actors within a system have interdependent interests” (Burt, 1982, p. 5). 3.1 Introduction This chapter elaborates on the network approach to study financial markets discussed in chapter one. The chapter introduces the use of corporate interlocking networks as applied to securities holdings in the networks the ECB is actively trading in. The corporate interlocks thereby introduce Granovetter’s (1985) embeddedness to financial networks, an embedded market approach. Chapter one discussed that economic actors are affected by the ties with other actors in the network. It is the focus of the next three chapters to uncover how the ECB changes the social structures through its policies, and more specifically, the enacted environment for market actors. Utilising a holdings-based model of corporate interlocking networks, the research questions focus on homophily, network density and imitation (uniformity of investment decisions). Financial economists tend to measure correlations between asset prices, trading volume and volatility. Indeed, the application of network analysis in financial economics usually revolves around price correlations between sets of indices. Sociologists tend to ask questions concerning how and why decisions are made. Ultimately, it is important to look at who creates financial networks as a product of interactions between these agents. As introduced in chapter one, this chapter discusses further the measurement of medium-term ties in corporate interlocking networks between institutions that co-own assets, rather than focusing on simple short transactions between buyers and sellers. These longer term ties are similar to what Baker (1990, pp. 594–597) referred to as the ‘relationship interface’, while the transactions between buyers and sellers resemble the ‘transaction interface’. While Baker introduced a hybrid model of these, the proposed ties through cross holdings in this chapter are medium term ties that are instituted through transactions. In other words, an AMC can form a tie with the ECB by purchasing a common bond that the ECB holds. Early studies in economic sociology have already noted the rising role of AMCs, or institutional investors, in shaping 78 financial markets (Mintz & Schwartz, 1985; Useem, 1980). Based on the empirical data from chapter two, investment banks, the facilitators of trade in financial markets, are losing significant influence to shape financial markets. Indeed, a large portion of research studies of economic sociologists in financial markets involved investment banks (Burt, 2000; Mintz & Schwartz, 1985; Podolny, 1993; Podolny & Phillips, 1996). Hence, the focus on ownership ties between AMCs and the ECB, through the analysis of the corporate interlocking networks of portfolio holdings, forms a novel approach to refine and re-apply previous research in economic sociology to current financial markets. The method aims to deepen the understanding of markets beyond the traditional focus in monetary economics on prices, but instead to see prices as a product of decisions being made between market participants in a given financial network, examine how these decisions are made and how centrality of institutions influences network behaviour. Whereas previous network studies of financial markets have focused on the social characteristics of the human protagonists, a holdings-based network focuses on empirically quantifiable positions that form connections between financial institutions. The key implication of Granovetter’s embeddedness model discussed in chapter one, is that material interest in financial markets is not atomistic and independently conceived of by investors, but conceptualised and embedded within an interlocking network structure of co-holders. As Baker (1984:804) contends, the relationships of the social structure of a given market influence prices directly. This approach to investing as action in financial networks comes close to what Ronald Burt (1982, pp. 8–16) developed as a structural perspective to action of interrelated actors in networks. Actors realise individual goals by conceiving of these within a given social context. “Actors find themselves in a social structure. That social structure defines their similarities, which in turn pattern their perceptions of the advantages to be had by taking each of several actions” (Burt, 1982, p. 9). Thus, as investors purchase securities embedded in a network structure, they are both constrained by and at the same time modify the structure itself. The structure shapes the context of action (such as cues to investment decisions), actor interest (portfolio holdings), and action (trading and adjusting portfolios) itself. Holdings-based network analysis offers an alternative method of analysis to price studies. As a result, this model treats securities (bonds and stocks) as entities embedded in social networks of owners, forming connections between economic agents in a given network. Market participants in a given financial network are herewith referred to as nodes. Nodes in 79 financial market networks are the institutions owning the securities, which to a large extent, are AMCs. The representatives and underlying decision makers are PMs of the funds holding these assets. The proposed network membership of including only holders of securities in the network, those that have physically transacted, makes practical sense, as, by definition, a security only exists if it has holders. Material interest will be defined as the holdings of securities by an individual node, which are, by definition, shared with other holders. The notion of centrality will be defined in detail below. This chapter first reviews relevant financial network research by both sociologists and economists in section 3.2. The notion of centrality is then dealt with in more depth in section 3.3. Section 3.4 discusses in what ways social scientists can address research questions in the study of financial networks and how approaches in sociology and economics could be combined. The proposed interlocking holdings-based analysis is then described in more detail both from a methodological and conceptual angle in section 3.5. 3.2 Financial Networks As detailed in chapter one, Granovetter’s (1985) embeddedness idea and the social network approach to markets sparked significant research by sociologists into financial markets. Early network studies of financial markets were developed by Baker (1984) who examined a stock option exchange in Chicago. At the time, trading activity was still very much physically linked to the exchange and the study was conducted before the advent of electronic trading. For Baker, “each location is thus the observable market place for specific options” (Baker, 1984, p. 776). Baker (1984) showed that the social structure of the network influences behaviour of actors within the network and applied this to an option market. Baker and Iyer (1992) also examined how information is disseminated in financial networks and how this in turn affects investment and trading decisions. Podolny (1994) focused on the concept of markets, constituted of social relations between participants and on decisions being shaped based on status within the network. He (1994) contended that in his status-based model of market competition, status is a currency through which single actors have influence over others. In an earlier paper, Podolny (1993) highlighted the categorial distinction between the actor and her role/status. Status can emit different signals depending on both present and past expressed opinions and actions 80 (Podolny, 1993, p. 831). As discussed in chapter one, if a central bank communicates a certain policy to market participants, the new information is interpreted by contextualising it in the current and past behaviour of the central bank. As introduced in chapter one, the corporate interlock literature looked at the relationship between class, elites and social cohesion in the US corporates of the 1980s. While not referencing or building on this body of research, recent related research was conducted on the ECB’s board of governor constellation (Lebaron, 2010; Lebaron & Dogan, 2016). This research focuses on the ECB as an institution rather than on its role as market actor and highlights the social and educational attributes of individuals in positions of authority at said institution. Economic studies of financial networks, in contrast, focus on empirically observable price changes in securities to make inferences on correlations and regressions (see e.g. Kumar & Deo, 2012; Namaki, Shirazi, Raei, & Jafari, 2011; Roy & Sarkar, 2013). Indeed, the wide availability of price data and the concurrent emergence of computing power over the past 20 years give feasibility to this kind of analysis. However, these network studies have some limitations in following the general assumptions of Fama’s (1970) Efficient Market Hypothesis (EMH), that market participants act in a rational way, using perfect information to make investment decisions. The discussion on the content of ties in chapter one summarises this. The EMH stipulates that prices reflect all available information to market participants and thus reflect economic reality (Malkiel & Fama, 1970). However, during bubbles, financial markets exhibit distinct supply and demand dynamics. As financial market prices increase, demand for these assets rises commensurately. Demand does not decrease as securities become more expensive, but as price rises signal buying behaviour, a phenomenon known as Keynes’ ‘beauty contest’ ensues (see e.g. Allen, Morris, & Shin, 2006; Scheinkman & Xiong, 2003). This stands in direct contrast to the EMH. Keynes’ ‘beauty contest’ likens investing in financial markets to participating in newspaper competitions where competitors have to pick the six prettiest faces out of hundred photos and where the winner is she who picks those that most closely correspond to the average participants’ choice (Keynes, 1936). In order to win, participants in Keynes’ beauty contest must make second and third order assumptions on not only which faces they deem the most beautiful, but also what they assume other market participants consider the most beautiful faces to be, and what others assume that others assume the most beautiful faces to be. This introduces a second or third 81 order of inference as market participants not only make investment decisions based on the cash flows or other fundamental principles but, indeed, what market perceptions are of the same and likely will be. “Thus the professional investor is forced to concern himself with the anticipation of impending changes, in the news or in the atmosphere, of the kind by which experience shows that the mass psychology of the market is most influenced” (Keynes, 1936, p. 155). Furthermore, Podolny’s (1993) finds that perceptions of status and reputation guide market participants’ actions in a network. Hence, the basic EMH network perspective of action can be summarised as the atomistic perspective, in assuming “that alternative actions are evaluated independently by separate actors so that evaluations are made without reference to other actors” (Burt, 1982, p. 5). With the recent growth in available financial data and utilising the concept of corporate interlocks, a holdings-based network analysis offers a useful tool to analyse shared material interest in financial ties for each node in the network. The nature of cross-holdings in a financial network is dealt with in more detail in the following section. 3.3 Centrality in financial networks Wherever an economic agent A has influence on a financial network, Agent A has to carefully manoeuvre within a web of material interests so as to not harm her own endeavours. Agent B will take actions, communications and perceived interests of Agent A into consideration and the resultant conclusions will shape a variable in the frame construction of Agent B. However, Agent A is arguably not able to determine interests for Agent B. Unless the situation occurs where Agent A benefits from the interests of agents in the network to an extent where the interests of Agent C,D,..,n are significantly distorted, Agent B is able to enter and exit the network by means of trade and withdraw herself from the situation and social context. However, such occurrences are rare. One example was the listing of the Japan Post IPO in 2015. In conjunction with the BOJ’s QE, the government planned to list its mail operation and the associated bank on the stock exchange but wanted to ensure that prices rise in value after listing in order to benefit individual retail investors that participated in the issue. In this way, a central actor distorted actions by under-pricing the security in order for those economic benefits to be shared by many small investors in the after-market. The fact that the issue would be big enough to be included in various stock indices, for example the 82 TOPIX and the Nikkei,41 ensured that individual investors would be able to sell this on to price insensitive passive funds. Passive funds would buy Japan Post at whatever price at the date of the inclusion into the index. Given the potential discrepancies between actions and perceived reputation as in Podolny’s work, it is difficult for central actors in financial networks to dictate decision-making processes for other market participants. Small investors have often suffered losses on the back of what governments had promised. If a central actor signals an intention to act in a certain way, market participants take this into consideration when choosing a course of action but also act on the basis of perceptions of past events. It is clear that the success and profitability of these endeavours plays a part in whether these agents can continue doing this in the long run and stay in business. Given that market participants are constrained by competitors in a network (Burt, 1983, p. 16), clients may redeem their money with a PM should she underperform, thereby reducing her ability to implement affairs as she sees fit in the future and thereby reducing her influence as market participant. Given the reflexive character of financial market actors, in that other people’s actions and expressed opinions are taken into consideration in both investor frame construction and decision-making, a central bank such as the ECB, would need to back up communication with physical trading in order for economic agents B,C,D,…,n to perceive a situation in the way this was communicated. Very often, control of a situation can easily be lost, if communication and economic action of a central bank is incongruent. Signals emerge out of day to day activities, for instance sales traders informing counter- parties what other clients are buying or selling. Likewise, a personnel change or news of a change in levels of assets under management for Actor B may also signal a change in material interest and holdings for that actor. A simple example was the steep sell off in bond and equity markets, when previous bond manager Bill Gross left Pacific Investment Management (PIMCO) in Sep 2014 (see e.g. Domm, 2014). Bill Gross was co-founder of PIMCO, one of the then largest asset managers globally, and managed one of the world’s largest bond funds, the PIMCO Total Return Fund. His positions were unwound after his 41 Two of the major Japanese stock market indices. 83 departure and led to steep moves in capital markets. Given a confluence of other factors,42 market participants went to assume a disorderly unwind of a prominent Gross trade, namely that of the unwind of the ‘short volatility trade’ which he openly propagated over the previous year as his biggest position. Another example was the ECB’s announcement of the sovereign debt purchase program in March 2015. While a particularly novel and also contentious policy, the ECB hinted at this in earlier investor meetings and press conferences from the end of 2014, before announcing the program proper (see chapter six). Once the policy was announced, an information cascade unfolded leading to a particularly strong positive impact on equity markets, indicating that this was indeed unexpected, despite it having been previously introduced into the central bank’s discourse. Equity managers who did not traditionally follow central bank monetary policy in great detail needed to join the information cascade and some translation work was done by experts from the investment banks. These examples illustrate that the notion of an anonymous and informationally efficient market is not appropriate in actuality. The extent market participants consider positions of counterparties is higher than given due consideration. In summary, networks are not deterministic structures, but central nodes can influence and impact distinct courses of action for other market participants. Measuring how influential a given Actor A is, leads to the measure of centrality, developed and refined by various sociologists (Bonacich, 1987; Freeman, 1978; Marsden, 2002). The aim is to include in the holdings-based network analysis measures of not only the pure number of connections of a given node (degree centrality), but also aim to measure second and third degree connections. For instance, the simple quantity of friends does not necessarily reflect the influence an individual node has compared to other nodes with the same quantity of friends, in a linear fashion. If a network is split into clusters, there exist what Burt (1992) referred to as ‘structural holes’. Nodes that can fill and mediate between clusters are able to profit from such structural holes, either by facilitating exchange or access to multiple information sources. This involves active engagement by agents across divergent networks. 42 See e.g. Hedge Fund Moore Capital’s take (Frieda, 2015). 84 Status-based models of centrality looked at network centrality as a source of influence (Bonacich, 1987; Podolny, 1993). The degree of influence an Actor A can exert in financial markets depends on her position in such a financial network. This incorporates the ability to ‘move markets’ or more specifically, to move prices. Such ability is embedded in the networks of material interest at that particular time, whereby material interest is understood to be the shared securities in said market. If Actor A’s position in a network is central, her actions, communications and changes in her individual situation will be monitored by the network of actors and used as cues to the investment decision-making process. One limitation of this approach may be that the counterfactual remains unaccounted for. In other words, this approach does not account for actions that have not been taken. However, given the focus and endeavour to measure relations in an agency-based network, the aim is to analyse shared material interest as defined above. 3.4 Empirical usage Taking the network approach to financial markets, the consociation during a trade will cause one market participant to own an asset and another to reduce her holding of it. In the proposed model, examining the holdings can serve as method to analyse relationships between market agents. Concomitantly, it becomes one empirical expression of material interest in financial markets. Rising prices are a strong signal sparking social imitation in the network. Expanding ownership and thus increases in the number of holders of a security traded sees an expansion in the network of holders. As securities are grouped in different market segments, different networks of holders of these securities can be analysed. These holders then form a financial network with defined membership parameters. Inevitably, economic fortunes are intrinsically linked between these actors. This network effect on financial markets becomes obvious in the context of studying contagion risk. For example, when the Greek government was about to default on their bonds in 2010, market participants across all capital markets paid specific attention on which holders were most exposed to the bonds, which in turn impacted the price developments of those exposed securities (see e.g. Chambers, 2010). Exposed agents were then perceived as risky counter parties. Considering other market participants in a network also plays a role in day to day asset management. If a central bank enters a specific segment of a capital market, 85 e.g. the ECB’s entry into the European corporate bond market in 2016, it provides a stimulus for prices as actors in the network consider the impetus the entrance of the central bank will have. Such cues can take different forms and are a continuous input to decision-making processes. In active portfolio management, where funds aim to outperform relevant benchmarks, PMs are key actors in shaping the interest of AMCs as they select the investments the firms will be exposed to. In a network analysis of financial markets, the material interests of PMs and the companies they represent are thus intrinsically linked. Any list of top owners of a security reflects the decisions made by a PM managing active funds for an AMC. Conversely, passive funds would own securities purely on the basis of those securities being constituents of the index. The network approach to financial markets thus shows that material interests are not an independent pursuit such as in the atomistic perspective, but shared with other market participants in financial networks. I therefore propose to examine these interests methodologically through network analysis. 3.5 Notations and Overview In a corporate interlocking holdings-based model, the financial market to be researched is constituted of securities 𝑠(1,2, … 𝑛). Each security 𝑠(𝑛) incorporates a n-list of owners with varying percentage ownership levels. Hence nodes are the institutions and investors holding and owning the securities that form a financial market. For purposes of this thesis research and to keep the data manageable, nodes of only up to the 20th holder are included as the level of influence is deemed too small beyond this level. So 𝑁(𝑔) = 𝑠(𝑛)1,2,..,20. This results in 𝑁(max) = 𝑠(𝑛) × 20 𝑎𝑛𝑑 𝑁(𝑚𝑖𝑛) = 𝑠(𝑛). This threshold ensures that the maximum 20 number of holders for a given 𝑠(𝑛) will only have a maximum level of ownership for the 20th largest holder of 5%. Let there be four ordinal k-core levels of ownership with 𝑘1 > 0%, 𝑘2 ≥ 2%, 𝑘3 ≥ 5% 𝑎𝑛𝑑 𝑘4 ≥ 10%. As discussed in previous sociological accounts (Mintz & Schwartz, 1985, pp. 98–99; Useem, 1980, p. 42), institutional investors are able to exert significant influence through their portfolio holdings. This both impacts prices of securities when trading is conducted, but also proxy voting. Consistent with these accounts, the assumption taken here is that with greater ownership levels comes greater influence and the different ownership thresholds K1 – K4 yield multiple graphs 𝑔𝑘𝑛 at k-levels of concentration. 86 The resultant network (𝑔) is an undirected network as investors share securities holdings which define the network boundary. Only nodes that are holders of a security are included in the network, so network (𝑔) consists of vertices 𝑔𝑖𝑗 = 𝑔𝑗𝑖 and 𝑔𝑖𝑗 ≠ 0. The number of degrees of a node 𝑖 is simply 𝑁𝑖(𝑔) and the maximum degrees a node (𝑁𝑖) can have is thus 𝑖(𝑁−1). When looking at ECB purchases as part of the PSPP and CSPP, the network is essentially an ego-network (𝑔)in which the central bank is, by definition, connected to everyone holding the securities included in the program, the so called eligible bonds. This is not the case for an equity index, for instance, in which multiple actors may own all the securities in the index if they are so-called passive investors tracking the index. Given that purchase programs of central banks are not based on a known index created by an index provider, the purchase program network of a central bank is mostly an ego-network.43 Financial markets constitute networks of agents that are a) linked through trading activities at a specific time, but even more so through b) sharing ownership in certain securities. Networks are not static structures, but agents reflexively engage and interact with other agents and thereby create and reshape networks themselves. If a financial market is defined by the traded securities it incorporates, these securities can shape relationships between the buyer and seller. Securities are hence embedded in these networks of market participants. The interlocking ties between these institutions are defined by shared securities holdings. The extent and strength of the relationship can be measured by the k-level of holding percentages. This shared and common economic situation requires market participants to be aware of other people’s actions and signals. There are two separate network relationships: A short term relation between buyer and seller at the point of transaction and a more medium to longer term relationship between co-holders of securities for a certain amount of time. This distinction, again, is akin to what Baker (1990) refers to as transaction and relationship interface, in which the relationship interface is a more longer term market tie. The holdings- based network assumes that the material interest of co-owners of an asset is more strongly tied and aligned than with a buyer and seller of an asset, where by definition, the interest of 43 However, the Fed recently bought a corporate bond ETF directly in its last quantitative easing initiative during the Covid-19 pandemic. 87 the two market participants changes in an inverse direction, by one party reducing and one party increasing the exposure to a certain asset. Co-owners of assets are likely to have more influence on each other’s behaviour given their shared economic position. In contrast to previous sociological work into networks, the financial market is here defined as the securities it constitutes, not the physical setting of the trading conduct as in Baker’s (1984) study. The network approach suggested in this chapter does not focus on status-based preferences of trade as I deem the actual execution of the trade, the transaction interface, a means to a tie formation and hence a secondary consideration. Rather, decision-making is formed as a complex rationalisation of assumed courses of actions by taking other nodes into consideration. I make the same assumption as Podolny (1993, p. 830) in acknowledging that the structure of a market is socially formed, particularly by market participants’ perceptions. Marrying this principle with Keynes’ ‘beauty contest’, the resultant financial network is shaped by perceptions of market participants, but also by assumptions of other market participants’ perceptions, second and third order principles (Allen et al., 2006; Barberis, Shleifer, & Vishny, 1998). Smaller single name investments, with short-term holding periods, could be influenced by collaborative networks but whole portfolios and medium-term asset allocation decisions are not. The medium-term asset allocation decisions are ultimately contributing to the formation of these financial networks. The measure of degree centrality in the model is a direct measure of the number of connections a node has by holding select securities in a given financial market. For example, taking the largest holders 𝑁1,2,…𝑛 of the constituent securities of the S&P 500 index and shared holdings among these 1 to 𝑛𝑡ℎ holders can be recorded as connections in an adjacency matrix. It is of course highly likely that most passive funds own the entire index constituents, but it is not clear whether the particular passive fund is among the top 𝑛𝑡ℎ holder of each security. Given that this is a two mode network, the edges, or ties, are weighted by the number of shared connections between two nodes. Hence the measure of strength of ties is confined to the top holders to clearly delineate measures of influence. Once the analysis moves beyond indices that are widely tracked by passive funds, such as the ECB’s bond holdings, inferences on central node influence can be made. 88 Derived from the conception of the investor frame in chapter two, node attributes are primarily geographical and institutional types. Taking the investor type of specific nodes, ‘pension funds’ act on specific parameters agreed on by plan fiduciaries, with the legal framework adding a significant input to frame construction. Likewise, taking geographical origin, German investors, for instance, may have a mandate to invest in German bonds only, which shows some country bias in the selection process of the investor frame. Information cascades could be beneficial to certain actors, geographically close to a central node, who may be able to extract a profit quicker than those further away in the network through front- running expected investment behaviour. Indeed, Coval and Moskowitz (2001) found geographical proximity to have an outsized effect on investment returns. Any subgraph (𝑔′) will display clusters by node attributes such as Country or Investor Type. This discussion shows that when an actor such as a central bank enters a certain financial market segment with its asset purchase program, it can be complicated to align the central bank’s monetary policy with incentives, communication within a network and the intended market reaction. As a central node in the network, the central bank would have to consider the direct beneficiaries of purchasing programs in these networks based on node attributes. Analysing homophily in the context of financial markets can give clues to the type of decisions made in the past and the potential preference of a particular type of market actor. As homophily assumes that similarities in node attributes bring about connections, a regulator looking at a particular financial network would want to know what attributes shape the network. This will give detail on why a certain financial market exhibits a particular network ownership structure and characteristic. Analysing homophily will thus aid in examining herding behaviour. Are there home biases for security selection and where are those concentrated? Looking back at the example of a Greek default on their government bonds, who holds these bonds and why? What decision-making processes have shaped these ownership networks? These are some questions that could be addressed by the network graph. 3.6 Conclusions This chapter elaborated on the outline of a holdings-based network analysis introduced in chapter one and emphasised, that if a central bank such as the ECB chooses to enter a particular financial market, it can look at the network constitution of certain eligible 89 securities and anticipate which actors will be impacted by its potential policy decision. Choosing a network with predominantly passive funds for instance could reinforce the ECB’s policy measures as passive funds continue to blindly buy those securities that are growing in value thereby mimicking and amplifying the buying behaviour of the central bank. Chapter four will look into this in detail. I have attempted to show that an analysis of financial markets from a social scientific perspective needs to address questions of material interest and the embeddedness of securities in financial networks. Financial markets are networks of nodes that share ownership in common securities. Securities, or portfolio holdings, form both an expression of material interest of a node as well as a connection between nodes. This perspective aims to offer a deeper understanding of why prices develop the way they do and attempts to paint a picture of the social structure of a given financial network. The chapter introduced a model to analyse exactly this through applying the methods developed in the corporate interlock network research onto securities holdings. Strength of ties, homophily, network centrality and home bias are key aspects of the financial network that are to be researched in chapters four and five. In relation to the findings of chapter two, central nodes in financial networks should not be seen as frame makers that determine the structure and behaviour of nodes in the network, but as agents that exert influence on individual PM frame construction and adjustment. The key consideration for central nodes is the influence on price formation and potential signals it can send through its actions and communications. The central bank was introduced as a potential actor, but networks can be drawn for any actor, large and small asset managers alike. Awareness of other market participants’ can be strengthened as a result. By synthesising sociologists’ advancements into the study of financial markets with the literature addressing limitations to the EMH, this research portrays the network structure of financial markets as a product of interactions between the group of market participants, rather than as a static structure that determines behaviour in the network. The idea of choice and optionality within the network is a key distinction to other network approaches. Financial networks are thus a complex system of relations that are dynamic over time. Dynamic networks are an essential part of financial markets in which relations shift with price changes and actors adapt positions. Sociology acknowledges that actors in a network take their role 90 and other actors into consideration, and financial network analysis needs to address this second and third order of assumptions made when asking research questions. In the next chapter I will present specific holdings-based financial networks that have been targeted by the ECB in their asset purchase programs. 91 4 Corporate connections and social imitation in the European Corporate Bond Market 4.1 Introduction Financial networks are significant to society at large. Contagion spread through financial networks of cross-holdings during pervious episodes of crises such as in Greece and peripheral Europe between 2010 and 2015. Portugal, Ireland, Greece, and Spain contributed to severe financial duress in European capital markets in the years following 2010. In response, the ECB launched large scale asset purchases, Europe’s version of QE, with their PSPP and CSPP to combat both market turmoil and disinflationary trends. With the increased globalisation of capital, financial networks are both growing in scale and complexity. Financial shocks spread through networks and impact society at large creating the need for an approach to study financial networks, not based on asset price, trading volume or volatility correlations such as in financial economics,44 but, as Schweitzer et al. (2009) call for, novel approaches that can combine the study of centrality with individual agent-based decision- making processes. The objective of this chapter is to utilise the holdings-based network analysis introduced in chapter three and apply this to one of the ECB’s asset purchase programs, the CSPP, and analyse how the ECB shapes financial networks as a central actor. More specifically, the chapter examines the micro-structure of the European corporate bond market and analyses how the ECB shapes investment behaviour of the network. The CSPP is predominantly there to establish the ECB as a major investor in €-denominated non-bank corporate bonds by continuous purchases in both primary (buying directly from the issuing bank) and secondary (on the secondary market) bond issues. This process aims to inflate prices and reduce yields of said bonds and encourage other holders to sell and move to riskier bonds or those with a longer duration, in turn, improving financial conditions for the system and borrowing conditions for riskier and smaller companies. This migration to riskier assets is referred to as portfolio rebalancing effect. 44 See e.g. (Brida, Matesanz, & Seijas, 2016; Dimitrios & Vasileios, 2015; Roy & Sarkar, 2011, 2013). 92 This chapter offers an overview of the actual holders of the assets the ECB is purchasing and thereby examines the resultant beneficiaries and the observed network behaviour. European corporates have not faced such easy monetary conditions before and were encouraged to increase leverage. Likewise, bond investors benefitted from a free lift in prices of their portfolios. Given the relatively recent history of European QE, there are few studies of the purchase programs, let alone those utilising social network analysis. QE had reached an unprecedented scale in late 2018 with the ECB’s balance sheet alone amounting to almost €5tn.45 This has caused discomfort with those European central bankers who pursue a more Germanic and hawkish philosophy, but also with political bodies such as the Dutch parliament intensely questioning and criticising Mario Draghi, the then ECB president.46 A hawkish stance reflects a focus on fighting inflation rather than economic growth. Now, since net purchases for the CSPP were halted in December 2018, it is necessary to analyse the dynamics of the CSPP network and potential distortions an unwind of positions could bring, should the policy reverse. The chapter aims address the conundrum of corporate interlocks and influence discussed in chapter one and three, It also addresses the question of interlocking networks and profitability raised by Mizruchi (1996, p. 275) and tackled in recent research (Cohen et al., 2008, 2012). Using descriptive data, Section 4.4 discusses the decision methodology and the resultant mechanism of herding of central nodes in the network and indicates beneficial alignment with the central node. Analysing the ownership structure of financial markets in which the ECB is conducting monetary policy is important so as to see whether dominant actors in these markets are willing to leave their habitat for riskier assets as the portfolio rebalancing effect and central bankers suggest. Lastly, the overview of the holdings-based network of the European corporate bond market and the dynamics therein may contribute to the understanding of potential contagion and financial market risks a regulator has to take into consideration. With potentially concentrated and dense networks, financial shocks could be exacerbated with the need for central banks or other authorities to bail out individual market 45https://www.ecb.europa.eu/pub/pdf/other/ecb.eurosystembalancesheet2018~5b51d1aefe.en.pdf?eea517404936 d72c65611c1bb3553ee6 [accessed 03.04.2020]. 46 German sentiment towards QE had been resentful from the likes of Jens Weidmann, Otmar Issing, Jürgen Stark, but also of politicians across Germany and the Netherlands questioning the legality of sovereign bond purchases. 93 participants exposed to certain assets.47 This is of significance to society as a whole as corporate bonds have been sold to retail investors, particularly in Spain and Italy through the local banks (European Commission, 2017, p. 40) The chapter is structured as follows. Section 4.2 introduces the background and the two hypotheses that I aim to test. Section 4.3 offers an overview of the background to the study, a description of the data collection and a brief literature review. Section 4.4 presents the results and analysis. The concluding discussion puts the findings into a broader context and elaborates on implications for the literature. 4.2 Background and Hypotheses Whereas Baker (1984) saw the financial market as a physical property in the options exchange he was studying, he (1984, p. 806) also acknowledged that electronic trading could change some of the findings. Indeed over the past two decades, financial markets have become increasingly de-materialised (Knorr Cetina & Bruegger, 2002). Even Over-the- Counter (OTC) options, agreed between two parties rather than cleared by the exchange, are mostly traded through electronic communication tools – listed options, are traded electronically on the exchange itself. Amid the proliferation of electronic trading and physical mobility, nodes in financial networks exchange and filter an increasing amount of information that is incorporated in investment decision-making. Within a network, nodes are increasingly aware of each other’s positions, communications and constraints. Early research into interlocking directorates contended that greater number of ties bring about increased intercorporate influence and economic interdependency, leading to common action (Mintz & Schwartz, 1981, p. 852). More recent research in corporate interlocks finds that higher intercorporate connections to reduce diversity in decision-making and enhance social imitation in the network. Hence dense networks should result in stronger ‘Granovetterian’ ties and also similar investment decisions (Fracassi, 2016). Communications and actions from a central node in a given network can amplify certain actions and mechanisms also known as systemic feedbacks. Concentrated networks for instance can bear systemic risks, risks cascading from one node to the whole system, as seen in previous financial crises (Banerjee, 47 The prime example would be the bail out of AIG in mortgage-backed securities during the Lehman crisis in 2008/2009. 94 1992; Schweitzer et al., 2009, p. 424f). This can lead to herding and imitation in networks through information cascades (see e.g. Banerjee, 1992; Banerjee, Chandrasekhar, Duflo, & Jackson, 2013). Crises tended to emanate from unknown events, in which people are caught unaware. Hence imitation is not necessarily mindless, but it involves drawing rational inferences from limited information (Easley & Kleinberg, 2010, p. 484). Monetary economics assumes that the portfolio rebalancing effect helps the transmission mechanism of the monetary policy. As a means to empirically examine this claim, economists look at asset price movements, in this case yield contraction, in asset markets ensuing announcements of monetary policy (Zaghini, 2019). Only a few studies capture changes in ownership (Nederlandsche Bank, Boermans, & Keshkov, 2018) and I am unaware of research, looking at the network of CSPP securities holders directly, specifically during the time prior to the announcement of the end of the program. A network of eligible assets should as a result become less dense with central market participants, usually owning these assets, migrating to riskier securities with different maturities and credit rating. In the context of this study, when the ECB becomes a significant investor in the market itself, does this lead to a dense network and imitation? Are investors following the ECB or migrating into riskier parts of the capital structure as the ECB president suggested? To assess such behavioural similarity is to measure the number of shared ties in the corporate interlocking network. This leads to the first Hypothesis. H1: The behaviour and positioning of the ECB leads to a comparatively dense network encouraging imitation and herding in eligible assets. I aim to test this hypothesis in section 4.4.4 by examining whether the strength of a connection in form of shared portfolio securities leads to an overlapping neighbourhood in the form of increased number of shared ties in the network. Secondly, data gathering was conducted from March 2018 – March 2019, which spanned a time when the CSPP was both well-known to bond market participants and during which, the ECB also tapered (reduced monthly purchases), announced the future end of the program (December 2018) and then halted net purchases for said program (January 2019). The fact 95 that the study was conducted almost 2 years into the purchasing program has advantages in that the network structure had time to form. However, it is difficult to make inferences on how the network was before the ECB started buying. To circumnavigate this limitation, I have included comparative measures through the Bloomberg Barclays European Corporate Bond index and the STOXX 600 Europe Equity index. Other central banks such as the BoJ actually conduct purchases of equities as part of their asset purchases so that it is an apt comparative measure. As a result, the Euro Stoxx 50 and the STOXX 600 equity indices were also analysed as part of testing Hypothesis 1. Since the network analysis does not include the time before the ECB entered these securities in 2016, it is interesting to observe the change in the strength of ties during the exit phase, ‘on the way out’. If the network indicates strong tie characteristics as tested in H1, does the level of this bandwagon effect taper off with a fall in ECB purchasing activity? This leads to Hypothesis 2. H2: The level of strength of ties across the network decreases in line with the falling purchasing levels of the ECB. 4.3 The case study of the European Central Bank’s QE 4.3.1 ECB as market participant node As the central bank for the eurozone, the ECB regulates money supply and interest rates and has a single mandate of price stability. Acting within this mandate, the ECB implemented unprecedented market-based monetary policy through its large-scale asset purchases, using the central bank’s balance sheet to accumulate eligible assets. Eligible assets are defined by the policy and are included in the Eurosystem collateral framework (ESCF) which currently consists of over 24,000 securities.48 The Fed had launched its QE in 2008. In a similar vein, the ECB launched its own version of QE, mainly with the PSPP. It was discussed in press conferences and other presentations in late 2014, announced in January 2015 and implemented in March 2015. The PSPP was part of the Extended Asset Purchase Program (EAPP) as the ECB was already buying covered 48 The collateral framework is updated regularly and detailed positions are available under: https://www.ecb.europa.eu/paym/coll/html/index.en.html [accessed 03.04.2020]. 96 bonds and Asset Backed Securities (ABS). The CSPP was a further extension of the Asset Purchase Program and was introduced in March 2016 and initially ended in December 2018. It aimed to include IG non-bank Corporate Bonds denominated in €. After only a month of buying in July 2016, the ECB had already acquired around €7bn across 600 securities.49 The number of issues included in the sample ranged from 1100 to 1200 for the study. Holdings were posted bi-weekly on the ECB website. After only two months of buying, the size of the CSPP amounted to €13.2bn,50 eclipsing the then largest investment vehicle in the European corporate bond market, the iShares Eur Corp Bond ETF of €9.3bn. These purchase programs were born out of the need to raise inflation expectations for the eurozone during 2014 and were unprecedented. A central bank in Europe had never before purchased private debt securities on such a large scale, directly impacting idiosyncratic corporate risk and becoming a central node in the market itself. The justification for such a program was based on the portfolio rebalancing effect. As Draghi put it, the purchases: Not only alter the price of risk-free securities…They also generate scarcity in the market in which we buy, which encourages investors to shift holdings into other asset classes – e.g. from sovereign to corporate bonds, from debt to equity, and across jurisdictions” (Draghi, 2015b). The example of the CSPP gives ample room to investigate how market participants interpret and incorporate new information from institutions into their decision-making processes. Through forward guidance, also known as the signalling process, the ECB reassures and gives a continuous detailed assessment of future purchases, duration and credit quality of the selected targets in a predefined list of bonds, available to the public. Given that the minimum size of direct investments in corporate bonds is beyond the threshold available for regular retail investors, the market participants are mainly institutional investors managing investment vehicles such as mutual funds, hedge funds, pension funds or ETFs. PMs are tasked to invest funds according to the specific investment parameters laid out by their mandate. 49 See Jezek (2016) for details. 50 https://www.ecb.europa.eu/mopo/implement/omt/html/index.en.html#cspp [accessed 03.04.2020]. 97 Herding in the context of institutional investors has been studied in various context (see e.g. Chang, 2009; Hirshleifer & Hong Teoh, 2003; Kellard et al., 2017; Nofsinger & Sias, 1999; Sias, 2004; Spyrou, 2013). The bifurcation of investment choices into eligible and non- eligible corporate bonds at the announcement of the CSPP shaped and framed the credit strategy significantly over the duration of the program, as can be observed in many analyst notes either calling to buy CSPP eligible or non-eligible bonds (de Zeeuw, 2016; Jezek, 2016; Suttard, Kini, & Edwards, 2016). Once tapering of the CSPP unfolded in the reduction of monthly purchases, the question that market participants considered was the eventual end of the program itself. The question of how central banks influence financial markets with their monetary policy has been answered predominantly from an asset price, trading volume or volatility correlation perspective in investment research, monetary economics and central banking research. After the era of QE ensued, many sociologists and political economists have raised questions about the benefits of QE and large-scale asset purchases. This is important when evaluating the consequences of QE, be it moral hazard or overreliance and dependence on central banks. 4.3.2 Node considerations – Active and Passive Active vs Passive Investing is a wide ranging debate in the literature (see Fama & French, 2010; Sharpe, 1991). In the context of the CSPP, figure 4.1 below lists the largest passive ETFs in the European corporate bond market. Index constituent securities are bought in line with any new fund inflows and according to the index calculation elaborated below. Active funds on the other hand, would consider which names to buy and when to dispose of positions based on proprietary analysis. If the ECB signals iterative buying in certain securities, active funds would face the dilemma of whether to a) buy the same securities and in a sense front-run buying behaviour of the ECB or b) move into riskier assets with lower credit rating or longer dated maturities. The former option would result in herding behaviour, the latter would be a portfolio rebalancing mechanism in that market participants move up the risk curve. In the bond market, one could either go into junk bonds or higher duration which bears higher risk in terms of interest rate and duration risk. In practitioners’ terms the rationale for portfolio rebalancing would be a spread contraction trade, i.e. the riskier bonds would benefit more than the IG bonds from improvements in financial conditions. 98 Only the active investor is in the position to consider this additional information, the forward guidance of the central bank and its intended course of action for market participants. In order for the active manager to make a return above the benchmark, she could either consider buying the same securities quickly at lower prices with the ECB driving up prices thereafter through its anticipated purchases or assume that the contraction in spreads would be proportionately higher for lower quality and higher beta issues. Thus, from a network perspective, I would anticipate for the CSPP network to become denser over time as passive funds imitate the iterative buying behaviour of the ECB. The denser a network becomes over time, the more it indicates that herding takes place as formulated in Hypothesis 1 above. Holders in a certain security of an index have a very specific reason for holding this. Given that for instance the ECB chooses the securities included in their purchasing program with the help of the collateral framework and its eligible list of securities, this is by definition shaping the network. Consequently, the ECB’s centrality in the network is not the main interest, but more so which market actors are very close in the network and what kind of reasons stand behind this. 4.3.3 Data Collection Data gathering was conducted from March 2018 to March 2019, which spanned a time when the CSPP was both well-known to bond market participants and during which the ECB also tapered, announced the end of the program and then halted net purchases for said program. The network analysis does not include the time before the ECB entered these securities. The ISIN code for these securities is available on a bi-weekly basis from the ECB website and can easily be downloaded into Excel. Although the ECB publishes securities it had purchased online (CSPP holdings are made available for securities lending by the national banks), the amount purchased in each individual security is not given. There is a 70% issue limit. Given that the ECB is the largest single investor in the securities, I equated the % ownership with the given highest holder for the 6th, 7th and 8th node attribute (see further below). Thus, in each ego network the ECB at least shares as many edges (connections) as the largest other node. 99 The nodes in the network are organisations holding the bonds in the CSPP and are mainly Investment Advisors (a broad category including mutual funds, ETFs, active Long Only Funds, passive funds), Banks, Pension Funds, Insurance Companies, Sovereign Wealth Funds, Governments, Individuals, Corporations. The edges represent the number of shared securities held between two nodes. By definition, the networks presented are both ego- networks, dichotomous and edge-weighted networks of the ECB’s CSPP program recorded over a one-year period. Over the past year, the maximum recorded N was 535, and there are on average 520 – 540 nodes in the network. Nine node attributes were recorded of which five were selected for the network analysis (Investor Type, Country, >2%, >5%, >10% ownership). I coded and reformatted the data using macros in Excel, for use in the Gephi network visualisation tool (see Bastian, Heymann, & Jacomy, 2009) and UCINET Software (Borgatti, Everett, & Freeman, 2002). I split the continuous % ownership data of securities of each individual node into a) the count of % ownership in Security 𝑆𝑛 into a Pivot Table for each Node 𝑁: 𝑆𝑛1 ≠ 0 = 1 𝑎𝑛𝑑 𝑆𝑛2 = 0 and b) the count of all non-zero % ownership in Security 𝑆𝑛into three further 𝑘 − 𝑙𝑒𝑣𝑒𝑙𝑠, > 2%, > 5%, > 10% . The former was used to construct an edge- weighted Adjacency Matrix (𝑀) with the weight of the edges representing the 𝑆𝑢𝑚 𝑜𝑓 𝑡ℎ𝑒 𝐶𝑜𝑢𝑛𝑡 𝑜𝑓 𝑒𝑑𝑔𝑒𝑠 𝐸𝑎𝑏 or 𝐸𝑎𝑏 = ∑ 𝑆𝑎𝑏 𝑛𝑛 𝑖=1 . The latter was used as the 6th, 7th and 8th Node Attribute in form of the 𝑐𝑜𝑢𝑛𝑡 𝑜𝑓 𝑛𝑜𝑛 − 𝑧𝑒𝑟𝑜 𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝𝑠 𝑖𝑛 𝑙𝑒𝑣𝑒𝑙𝑠 𝑘1, 𝑘2, 𝑘3. I defined the network boundaries with a nominalist approach, as described by Marsden (1990), and limited membership to the top 20 holders of the securities held by the ECB as part of the CSPP. 4.4 Results 4.4.1 Network demographics and statistics As of March 2019, the network of the top 20 holders of CSPP securities consists of n = 527 holders. Out of the 527 holders, 77% are investment advisors (mutual funds, ETFs), 13% banks, 6% insurance companies and the balance is split between governments, hedge funds, pension funds, corporations, foundations, holding companies, brokerages and other. Around 17% of holders are domiciled in Luxembourg, 11% in Spain, 10% in France, 10% in the US, 100 7% in the UK, 6% in Germany and 4.5% in Switzerland. The remaining funds are domiciled in Italy, Sweden, Austria, Canada, and a few other small representations. 4.4.2 Network Structure It is necessary in network analysis to work with comparisons. The STOXX Europe 600 index works well as comparison for the CSPP network structure, as it is the broadest equity index to measure the European stock market. It could also be a possible monetary policy target and thus a potential policy measure.51 The STOXX Europe 600 has a fixed number of 600 component stocks, incorporating stocks from 17 European Countries and is one of the broadest indices tracking European equities.52 The number of constituent securities of the index comes a bit closer to those of the CSPP, rather than the often used Euro Stoxx 50. Hence I take the STOXX 600 as a best way to reflect European Stocks and in turn the Bloomberg Barclays European Corp Bond index and the CSPP itself as reflection of European credit markets. The CSPP is referred to as a 2-mode network (Borgatti & Everett, 1997) with institutional holders and bonds. To analyse the network structure, I first imported the above data using UCINET software (Borgatti et al., 2002). Table 4.1 depict the descriptive network statistics of the CSPP in relation to comparable networks. At a level of >0% ownership, all financial networks show a similar density. However, the network density of the CSPP rises the most from 0.21 at >0% to 0.76 at a >2% threshold. It is surprising to see how dispersed the European stock market ownership structure is on all the measured variables, Investor Type and Country of domicile. It can also be concluded that the network of European corporate bonds in the CSPP is significantly smaller than that of equities and dominated by passive funds, in particularly ETFs. This is surprising as the CSPP consists of 1,201 securities as of 15th Feb 2019 whereas the broadest European equity index has only half the number of constituents. Likewise, a universe of thousands of single bond issues is essentially controlled by around 99 nodes at the 2% ownership level (see table 4.1). In short, the ECB’s targeted corporate bond market, as measured by the CSPP network, is 51 The BOJ, for instance, chose to include equity ETFs in their QE. 52 See https://www.stoxx.com/index-details?symbol=SXXP for more details [accessed 03.04.2020]. 101 notably denser than the European equity market in both the STOXX 600 and the Euro Stoxx 50. Table 4.1 Comparable network statistics at >0% and >2% By definition, the network diameter for the CSPP and PSPP is two, as the ECB holds all the bonds in the networks. In a different market structure, the STOXX 600 has a relatively wide network diameter at 3. Figure 4.1 depicts how the network density outlined in table 4.1 change. As the k-level threshold for membership is increased, the CSPP (blue line) is becoming increasingly dense, rising continuously from 21% at k1 to 89% at k4. The ECB’s sovereign bond purchase program (orange line) on the other hand rises initially as the threshold is increased but plateaus and then declines to 62% at k4, indicating that the largest holders are less connected compared to those at k4 in the CSPP. Likewise, and despite only consisting of 50 index constituent securities, the Euro Stoxx 50 equity index (yellow line) shows high disbursement at an ownership level of >10% with only one tie between 10 nodes. With the broadest European equity index, the STOXX 600 index (grey line) density is consistently between 25% and 35% across all k-levels indicating less holder concentration and density. 102 Figure 4.1 Network density at 𝑘1−4 ownership levels. The following network visualisation illustrates the difference in density of the CSPP and the STOXX 600. The graphs were constructed using the Fruchterman-Reingold algorithm (Fruchterman & Reingold, 1991) and Gephi software (Bastian et al., 2009). Figure 4.2 shows the STOXX 600 top 20 holdings-based network visualisation for March 2019. The size of nodes is based on Freeman’s (1978) closeness centrality and the colour is set by node attribute Investor Type. The network shows a core of Investment Advisors (pink) and Banks (dark pink) but very mixed clusters of nodes around the core. This stands in stark contrast to the CSPP ownership network in March 2019. In Figure 4.3, the ECB is the central node (dark pink), classified under Investor Type ‘Government’. Immediate neighbours are predominantly Investment Advisors (pink), Banks (green) and Insurance Companies (light blue). Whereas European equities seem to have a very broad ownership network by Investor Type as seen in Figure 4.2, the CSPP network is dominated by asset managers and banks. That also implies that the benefits of the ECB’s QE are more concentrated among these interest groups whereas equities would be much broader. 0 0.2 0.4 0.6 0.8 1 >0% >2% >5% >10% CSPP PSPP STOXX Europe 600 Eurostoxx 50 103 Figure 4.2 STOXX 600 top 20 holdings-based network Figure 4.3 CSPP top 20 holdings-based network Given that the CSPP network is much more homogenous, this leads to a few inferences. At a >5% ownership level, the resultant concentration of nodes in the CSPP network is found in 104 Figure 4.4 using the same technique as above, but using colours to mark node attribute Country. The results show networks in April, July, October of 2018 and February 2019. Figure 4.4 CSPP Network with >5% partition and weighted edge size (Apr 2018 - Feb 2019) April 2018 July 2018 105 October 2018 February 2019 106 Over the observation period, the network graphs show the central positions of the US-based (dark grey) nodes Blackrock and Vanguard alongside the ECB. Both investment groups focus on predominantly passive investment strategies and the largest investment vehicles in this network are ETFs. The graphs also show the central German nodes (in orange April, July, Oct 2018 and in pink Feb 2019) Allianz, Deka, Union Investment and Deutsche Bank. While Deutsche Bank benefits from market making in bonds and syndication, the largest investment vehicles have become ETFs under the DB X-Trackers. Union Investment and Deka are the only central active asset managers in this network, while Pictet and Schroders are slightly away from the core. All other nodes in the core of the network graphs are following index rules or are essentially holding the bonds on their book for market making purposes. This is a significant finding which will be explored in section 4.4.4. 4.4.3 Centrality Linton Freeman (1978) developed the concept of closeness centrality. This is particularly useful for ego-networks in which every node is connected through at least one node. In this network, every node is connected to the ECB by at least one connection. Closeness centrality takes into account not only the number of connections but also the number of shared connections with central nodes by incorporating the geodesic (shortest distance between given nodes i and j). Closeness Centrality takes the inverse of the geodesics of a node and in that way measures the efficiency and cost of exchanging information; short distance means faster and lower cost of transmissions (Brandes, Borgatti, & Freeman, 2016; Freeman, 1978). From this research point of view, I argue that this also raises the risk of contagion if central nodes on this matrix are following each other. The formula used for this is found in Freeman (1978, p. 225) as the equivalent of: 𝐶𝐶(𝑖) = [∑ 𝑑(𝑖, 𝑗) 𝑁 𝑗 ] −1 . Using UCINET to calculate Closeness Centrality, Figure 4.5 shows a simple chart of the distribution of the values. The chart exhibits a very high concentration in the top 20 nodes after which the centrality scores drop quickly. Table 4.6 ranks the top 20 nodes by Closeness Centrality score. Out of the 20 nodes, there are only two very large active fund managers, German Deka and Union Investment. 19th ranked Hedge Fund Manager GAM also pursues an active mandate but is comparatively small in asset size. Indeed, eight out of the nine most central nodes are heavily focused on ETFs and these are the main holders of CSPP securities. 107 Figure 4.6 corresponds well with the central nodes in Table 4.1. iShares belongs to Blackrock, X-Trackers to Deutsche Bank and SPDR to State Street. Figure 4.5 CSPP Node Chart listed by Closeness Centrality Table 4.2 Nodes table ranked by Closeness Centrality Score Figure 4.6 Snapshot of largest ETFs tracking the Bloomberg Barclays Euro Corporate Bond Index Source: Bloomberg. 108 4.4.4 Herding and Imitation The descriptive statistics presented so far have shown how much denser the ego network of the ECB’s CSPP is than comparable networks. Accounting for a >2%, >5% and >10% partition, this is even more exaggerated. The larger the holding size of securities, the larger the impact on prices of securities in the network of said nodes. The consequences of that are analysed below. The Bloomberg Barclays € Corp Bond index, a widely followed benchmark for European corporate bonds, rebalances monthly to reflect price changes during the month which enables both active and passive investors to anticipate rebalancing with the following formula.53 𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒𝐵𝑜𝑛𝑑 = (𝑃𝑟𝑖𝑐𝑒𝐵𝑜𝑛𝑑 + 𝐴𝑐𝑐𝑟𝑢𝑒𝑑 𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡𝐵𝑜𝑛𝑑) ∗ 𝑃𝑎𝑟 𝐴𝑚𝑜𝑢𝑛𝑡 𝑂𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔𝐵𝑜𝑛𝑑 As the ECB buys eligible securities, the 𝑃𝑟𝑖𝑐𝑒𝐵𝑜𝑛𝑑 rises and in turn increases 𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒𝐵𝑜𝑛𝑑 and thereby spurs further purchases of nodes tracking the index. As the ECB was buying bond issues every month, this would affect the index value as for bond indices there is a far higher turnover of components, whereas e.g. for the STOXX Europe 600 there is usually an annual rebalancing and only a few components change. As bonds approach maturity, they leave the index a year before maturity, thereby causing a higher need for bond portfolios to churn compared to equities. Given that the network is comparatively denser and structured around the dominant passive ETF asset managers, the CSPP program will inevitably distort the market structure more as would the same program into equities. The Closeness Centrality scores are clustered around the top nodes, with twenty nodes recording over 0.8. The central nodes hold the securities 53 Here is an excerpt from the index methodology: “The Projected (Statistics) Universe is a dynamic set of bonds that changes daily to reflect the latest set of index-eligible securities. As an up-to-date projection of the next month’s Returns Universe, the Projected Universe assists active managers by providing them with the necessary insight to modify their portfolios ahead of any index changes and assists passive managers by preparing them for any executions needed ahead of monthly rebalancing. Indicative changes to securities are reflected daily in both the Projected and Returns Universes of the index and may cause bonds to enter or fall out of the Projected Universe, but will affect the composition of the Returns Universe only at month-end. The examples below illustrate how several transactions are treated in the Returns and Projected Universes over the course of a month” (Bloomberg Professional Service, 2017, p. 49). 109 predominantly in passive investment vehicles, thereby imitating the buying behaviour of the ECB on a large scale causing the herding of nodes 1 through 20. Another way to test for connectedness in the network is to test for strength of ties by using both the dichotomous and the valued CSPP network data collected. Mark Granovetter’s (1973, 1977) strength of weak ties idea suggests that nodes with a strong connection (in terms of the number of shared securities) are also increasingly connected to the same nodes, resulting, in this specific context, in herding. This follows the principle that strong ties resemble close-knit networks in which ‘a friend of my friend is also my friend’ (Goyal, 2007, p. 127). In the CSPP network, having the same neighbourhood simply means that the group of nodes made the same investment decision to purchase ECB securities and indicates similarity in decision-making or imitation. Borgatti and Feld (1994) first operationalised the strength of weak ties hypothesis in their UCINET software by utilising Hubert and Schulz’s (1976) Quadratic Assignment Procedure (QAP) correlation. Their assumption is that the stronger the tie between two nodes, the more their neighbourhoods should overlap. In this procedure, the adjacency matrix of the valued network is correlated with the matrix depicting the number of people to which each node in a pair is connected to. In simple terms, the adjacency matrix of the valued network is dichotomised into a 0,1 adjacency matrix. This simple matrix A is then post-multiplied by its transpose A’ using UCINET software (Borgatti, Everett, & Johnson, 2013). This is the overlap which counts the number of times each pair of rows has a 1 in the same column. A correlation between the valued network and the overlap with 5,000 permutations yields a Pearson’s r value which can be tested for significance. This procedure compares the observed correlation with a distribution of random correlations. If the Pearson’s r is significant compared to the p-value, the findings indicate strong support for a dense and strong tie network and is another way to look at the ‘bandwagon’ effect described above. H1: The behaviour and positioning of the ECB leads to a comparatively dense network encouraging imitation and herding in eligible assets. Table 4.3 shows the QAP dyadic correlation for the CSPP network, the network of the European corporate bond benchmark, as measured by the Bloomberg Barclays index, and the broad European equity index. The results show that the nodes in the CSPP network share 110 more nodes in common the stronger they are connected. While correlations for all asset classes are significant, the CSPP network scores the highest Pearson’s r value compared to measures of the European corporate bond and equity market. The Pearson’s r of 0.4834 for the CSPP network is a moderately strong and significant measure to indicate an imitation and bandwagon effect of well-connected nodes in the network. In other words, there is a moderately strong relationship between the number of corporate connections nodes have in common and the extent of shared financial interest. Table 4.3 Pearson Correlation CSPP, iShares Bloomberg Barc Euro Corp, STOXX 600 February 2019 Since the data collection period does not include holdings data for 2016, before the implementation of the CSPP, I look at the effect of the ECB reducing asset purchases during the data collection phase from March 2018 – March 2019. This would imply that with falling purchases of the ECB, nodes should start to anticipate selling by the ECB and the bandwagon effect should taper off. Figure 4.7 shows the extent of the reduction in the ECB’s net purchases of CSPP-eligible bonds, particularly from March 2018 until Jan 2019, which falls into the data observation period. Figure 4.7 Diagram of ECB monthly net purchases by asset purchase program Source: ECB. 111 H2: The reduction of net purchases of the ECB leads to a reduction in the bandwagon effect. To test this hypothesis, I ran the QAP Correlation at different time points during the observation period. The signalling of an exit from QE by the ECB described in section 4.2, as well as the reduction of purchase behaviour in Figure 4.7 has led to a decrease in the measure of strong ties across the network in column 1 of table 4.4 below. While the Pearson’s r remains significant indicating a strong-tie network, the correlation steadily decreased over the observation period in line with the reduction in purchasing behaviour of the ECB. Table 4.4 Strength of Ties test over the observation period 4.4.5 Analysis Both the descriptive statistics of the overall network structure of the ECB’s ego network in CSPP securities and the related correlation analysis has shown that the network is comparatively dense and has tendencies to encourage imitation and herding of nodes around the bonds purchased by the ECB. Nodes unaware of the passive herding in the CSPP network will be impacted significantly as liquidity constraints continue to surface given the operationally difficult way to implement the monthly rebalances discussed in section 4.4.4. This network mechanism may cause longer term liquidity problems as central active managers are sparse and usually fulfil the role to arbitrage mispricing in the market and make it more efficient and thus liquid (see e.g. European Union, 2017, p. 22). The dense network structure coupled with this herding mechanism will enable contagion to spread easily throughout the whole network and will be amplified should a node amongst the top 20 face selling/redemption pressure. In addition, there are only a few active managers to balance this risk and buy at distressed levels to act as contrarians. Previously, the European Commission Expert Group on European Corporate Bonds (2017) pointed out the prevalence of active managers in European corporate bond markets, but this is 112 not reflected in the ECB’s CSPP ego network. Indeed, the comparative analysis of the CSPP with sovereign purchases of the ECB in the PSPP and the generic equity indices in Europe show that given the structure of the European corporate bond market, the ECB has a disproportionate network effect. 4.5 Conclusions This chapter contended that by publishing the eligible bonds, announcing the anticipated buying amounts in their policy and becoming the largest investor in that market, the corporate interlocking network of the ECB results in herding behaviour based on an endorsement effect. This network mechanism enables market participants to both imitate and adjust portfolios, either in mechanical ways such as the index trackers described in section 4.4.4, or active managers anticipating ECB behaviour in the network. As shown, this also applies to when the ECB is exiting the network incrementally, in the form of the reduction of purchase amounts. By comparing the European corporate bond market to equities, it also transpired that the network structure of the ECB’s chosen policy tool enhances imitation based on the technical investment methodology of the most central nodes. This network structure inevitably leads to distortionary conditions as the ECB’s behaviour is imitated by central nodes in the network causing cascades and a bandwagon effect. It will also make an exit for the ECB out of this network difficult without significant ripple effects. As of April 2019, the ECB is replenishing maturing bonds maintaining a static balance sheet size. Once the policy would require reduction of the balance sheet, i.e. net selling of eligible bonds, it would be difficult for any node to take up that selling pressure given the network imitation of central nodes. It will likely amplify the negative effects of exiting QE. The ECB has become a gigantic node in the European corporate bond networks but is not able to implement what the policy set out to do, namely, to push central nodes out of eligible securities into riskier ones as intended by the portfolio rebalancing effect. By imitating the behaviour of the ECB, central nodes benefit from bond price increases. Interlock centrality in this network results in economic benefits by central nodes profiting from price increases of bonds included in the ECB purchases. Hence, this chapter explored yet another way in which nodes can benefit from corporate interlocks, addressed in other network research (Cohen et al., 2008, 2012; Fracassi, 2016). 113 Competition in the market, defined here as divergent behaviour by economic agents, in the market is reduced. The denser network structure then facilitates a domination of the ECB through the mirroring of positions by close central nodes. This cannot be seen as a healthy development and is an unintended consequence of QE. As discussed in section 4.3.2, the role of active asset managers such as Deka and Union Investment becomes ever more important to balance and arbitrage price inefficiencies and distortions. The European Commission’s Expert Group on European corporate bond markets recommended the European Commission and the ESMA to encourage corporate bond ETF trading given the arguably positive impact on price discovery and liquidity (European Commission, 2017, p. 51). In light of the findings in this chapter, ETFs that follow a formulaic investment methodology described in section 4.4.4 will exacerbate contagion in the network. As ETF redemptions have to be met with actual sales of the underlying assets can cause a negative feedback loop. Given the centrality of the ETF instruments in the CSPP holdings-based interlocking network, ETF redemptions and underlying bond sales could also cause concerns for retail investors holding single issues that are included in the ETFs. 114 5 Homophily and Home Bias in the Northern and Southern blocs of the eurozone 5.1 Introduction Chapter four analysed the social factors influencing decision-making processes in European corporate bond networks resulting in herding around corporate bond securities included in the ECB’s CSPP. Chapter five will analyse social factors in the network constitution of the PSPP using the holdings-based network model and analysing homophily based on security selection. Given that the PSPP includes sovereign bonds from both governments and government entities, the chapter will test whether home bias can be found in the PSPP network. This is significant as the ECB aims for the eurozone to share risks and encourages cross-border holdings of government debt. According to former ECB Vice President Vitor Constâncio: Financial integration provides risk sharing mechanisms which can reduce the impact of country-specific shocks and contributes to macroeconomic stability. Internationally diversified portfolios - cross-regional and cross-border asset holdings, including firm ownership claims - are more resilient to global and local shocks and can mitigate the impact of such adverse scenarios (Constâncio, 2018, p. 2). This chapter examines whether the ECB is able to enact an environment in which market participants increase cross-border holdings and share private sector risk or whether home bias prevails. It will also examine the case of Spain in detail, as the reduction of financial stress in Spain and Italy was specifically targeted by the ECB’s monetary policy. The study compares holder networks of German, French, Dutch, Italian and Spanish PSPP bonds and looks at the whole network of European sovereign debt. In general, interbank and cross-border sharing of debt ownership is important in times of crisis for a number of reasons, primarily as governments are able to tap into a more diverse pool of investors to raise debt. The influence of a single node during a crisis on a network can be mitigated by connections with other nodes. France, for instance, was particularly exposed to the Greek debt crisis as its banks, BNP Paribas amongst others, had large chunks of Greek government bonds on the balance sheet. In order to achieve financial stability in the eurozone, the ECB 115 encourages diversified cross-border holdings in portfolios as expounded by Constâncio above. This holds true for both the NCBs but also for other market participants. Hence it is the aim of this chapter to test for homophily in a geographical context for the PSPP networks: Are there clusters of market participants that have shared country or regional attributes? Are there home or regional biases in the eurozone? This chapter aims to contribute to the existing debate around risk sharing in the eurozone by analysing the network structure in the five most important sovereign bond markets of the eurozone (Germany, France, Italy, Spain and Netherlands) and make inferences on the social conduct in said markets. Empirical evidence is presented to underline the importance of financial network research conducted into home bias and homophily, and it is argued that sociology and behavioural finance find common ground in the exploration of these notions. It is shown that homophily in a geographical context, expressed as home bias, plays a pivotal role in the construction of financial networks of holders of European sovereign bond markets and takes precedence over assumptions of diversification in modern portfolio theory. Home bias, homophily and a domestic orientation of ownership in a market brings a lack of diversity in bond ownership and could lead to an amplification of market shocks. The chapter is structured as follows. Section 5.1.1 discusses the theoretical background to sociological studies of homophily in networks and links this conceptually to the notion of home bias in financial markets. It contextualises and presents the four hypotheses to be tested. The following sections deal with the PSPP program and the presentation of empirical results. Section 5.1.2 provides the policy and operational background to the PSPP. Section 5.1.3 discusses the relevant literature. Based on the presentation of the holdings-based model in chapter three, section 5.1.4 discusses the data collection for this chapter in more detail. Section 5.2 presents the results of the network study and section 5.3 forms the conclusion of the findings with its broader implications for European capital markets. 5.1.1 Homophily and Home Bias I introduced the concept of homophily in chapters one and three. This chapter will flesh out homophily in the context of geographic proximity. The main point of reference on homophily in the sociological literature is Lazarsfeld and Merton’s (1954) study of friendship formation in two US communities, Hilltown and Craftown. Homophily specifies that individuals make 116 connections with others that are similar in some respect, based on social characteristics such as ethnicity, gender, religious affiliation or age: “a tendency for friendships to form between those who are alike in some designated respect” (Lazarsfeld & Merton, 1954, p. 23). In the two communities under study, Lazarsfeld and Merton (1954) found that the degree of similarity between individuals in intimate relationships heavily depended on attributes such as ethnicity and gender, while educational status was less significant. The main assumption of homophily is that the social distance equates to the distance in the network position (McPherson et al., 2001). McPherson et al. (2001) also underline that people with network structural similarities are more likely to have increased communications and interactions, attend to each other’s positions and influence each other’s behaviour. In other words, nodes that share certain attributes are more likely to be densely connected in a network that is based on decision-making. The corporate interlocking networks of the PSPP are an apt study of this phenomenon as it includes bonds issued by eurozone governments. The node attribute of country, expressed as the institutions’ domicile, could be an important determinant of decision-making. For instance, if you are a PM in Frankfurt, you are more likely to regularly encounter local peers at investor meetings or in the financial centre compared to investors from Milan, Madrid or Paris. Geography, geographic proximity, is hence an important attribute in the study of homophily. In his analysis of the 1985 General Social Survey, Marsden (1988) contended that the degree of homophily in a network has implications for the network density and its stratification. Homophily in networks infers a high degree of local and low global integration. Likewise, individuals with offices in the same country may choose to increase interactions when meeting abroad, sharing information on restaurants, hotels or transport. If strong ties occur between nodes with similar attributes in weighted financial networks, certain reaction functions can be amplified as seen in chapter four. When analysing the monetary policy of the ECB, it would be interesting to explore to what extent sovereign bond markets, as measured by its constituent securities, in different countries included in the PSPP, experience home bias and how well they are regionally integrated. On the other hand, it is important to analyse behaviour of investors themselves, as measured by the network’s nodes, both when investing in local markets but also when making decisions in other countries of the eurozone. Portfolio diversification has been a known concept for many years both in financial economics as well as in practices of the asset management industry (see e.g. Levy & Sarnat, 117 1970; Markowitz, 1952). Modern portfolio theory proposes the benefits of diversification to portfolio returns, diversification in terms of asset classes but also country and sector exposure. The main idea could be summarised in the saying ‘don’t put all your eggs in one basket’. Despite evidence of the benefits of portfolio diversification, empirical research continues to record social factors influencing investment decisions in the phenomenon of the home bias - investors’ preference and bias to predominantly invest in domestic assets – geographically proximate - or those that are more familiar (see e.g. Coval & Moskowitz, 1999; French & Poterba, 1991; Huberman, 2001). In the empirical research this pertained to both individuals holding their employer’s shares as well as a higher asset allocation to the home country for investment funds. In their second study of home bias, Coval and Moskowitz (2001) show that mutual funds investing in companies that are geographically close are able to extract abnormally high returns as a result of more informed investment decisions. In their study of business school students, Kilka and Weber (2000) show how respondents in the US and Germany expected a higher return for their respective domestic equities. These findings corroborated the respondents’ own assessment of their competence, in other words familiarity with German and American equity markets respectively. Grinblatt and Keloharju (2001) document the preference amongst Finnish investors to hold domestic stocks or those from culturally similar geographies. More recently, the concept of home bias has also been addressed in studies of European sovereign bond ratings. Fuchs and Gehring (2017) contend that sovereign bonds from a culturally similar country receive consistently higher ratings than economic fundamentals would justify. It is thus important to understand the social influences shaping the European bond market when analysing the ECB’s sovereign bond purchase program. If there is a geographic stratification in the market, this home bias can act against what the ECB deems to be a positive outcome of risk sharing and diversified holder bases in the eurozone. As argued earlier, eurozone governments that depend on a small number of funding sources are more vulnerable to financial distress. Sociologists also studied homophily in financial networks and its effect on investment decision-making, such as in the US Venture Capital industry (Sorenson & Stuart, 2001). Sorenson and Stuart (2001) contend that the common thread between geographic and social proximity increases the likelihood of chance encounters and with it the opportunity to form links in financial networks, in that “differences in the influence of propinquity and homophily in economic exchange systems stem from variation across actors in their opportunities to trade" (Sorenson & Stuart, 2001, p. 1547). Hence, the phenomenon of homophily spans 118 social distance, while the notion of home bias spans geographic distance. More specifically, homophily represents social propinquity in the network (McPherson et al., 2001) and for home bias, geographic distance equates network distance. While these two notions have been developed in separate academic disciplines, they have the same principal assumption, in that node attributes, defined by geographic or social factors, shape decision-making processes and thus the network formation and structure. The idea of ‘home’ does not necessarily have to be connected entirely to geography but can merely represent the ‘familiar’. Thus, findings in sociology and behavioural finance indicate that ‘cultural similarity’ plays a role in network formation and decision-making. In that sense and despite disciplinary boundaries, the two notions of homophily and home bias are very closely linked. Simply put, home bias is an expression of homophily in a geographical context. Drawing from this prior research, it is interesting to analyse home bias not only in the countries of domicile, but also in different geographical blocs of the eurozone. Shared securities form the edges in this network. The question arises whether the membership to the securities under study has been made to align portfolios with other nodes (imitation or social influence) or whether the choices had been made to reflect nodes that are more alike (selection). 54 In this particular affiliation network, this question revolves around whether networks were formed on the basis of imitation (either of the ECB or other central nodes) or based on the investor frame of the node that guides decision-making processes. In the former case, the example in chapter four illustrates that nodes tend to follow central nodes in the selection process. In the latter case, the shared attributes, for instance the domicile country, may infer shared investment practices of nodes in the same country. Hence if there are homophilic tendencies in these networks, they either indicate that there have been a) shared criteria that led to the same decision of actors A and B to buy asset C, b) a central actor A owns asset C and due to its social influence makes actor B buy the same asset or c) a group of nodes D, E, F,…,n all hold asset C and exert social influence on actor B. The first process is referred to as focal closure, while the latter two processes are referred to as membership closure .55 54 For a discussion see Easley and Kleinberg (2010) p. 82 – 83. 55 The shared institution is referred to as focus, thus focal closure. If B buys the same asset that A already holds, this would be referred to as membership closure. 119 5.1.2 European Quantitative Easing and the Public Sector Purchase Program Under Ben Bernanke’s leadership, the US Fed embarked on QE through large-scale asset purchases with the Troubled Asset Relief Program (TARP) at the height of the GFC in 2008. Concurrently, under the previous president Jean-Claude Trichet, the ECB actually raised rates, a quantitative tightening measure. In 2011, Mario Draghi replaced Trichet as ECB president and quickly adopted a more dovish policy stance, signalling for the first potential asset purchases akin to those in Bernanke’s US-style QE in 2012 during his ‘bumblebee speech’.56 In 2014, the idea of sovereign bond purchases was floated at various opportunities, such as ECB speeches late in the year and at the Jackson Hole Economic Policy Symposium (Draghi, 2014b; Liesman, 2014). The idea was to control the widening of peripheral European sovereign spreads by large scale asset purchases, as well as combat disinflation at the end of 2014. The program faced significant political resistance, especially from Germany, where there was still a feeling of resentment about having had to finance the lion’s share of the European Financial Stability Facility (EFSF) at the beginning of the eurozone crisis in 2010. The Dutch government also voiced criticism of the program in their meeting with Mario Draghi in the autumn of 2017.57 Given that within its mandate, the ECB is unable to finance government budgets, a case was brought forward to the European Court of Justice (ECJ) by German Eurosceptics as to the question of the program’s legality.58 The ECJ however found that the ECB would be acting within its mandate to pursue such a policy. In response to a question by the press about the legality of the program, Mario Draghi stated Evidently we are convinced that a QE programme which could include sovereign bonds falls within our mandate, or better, is an eligible instrument that we could use in the pursuit of our mandate. Not to pursue our mandate would be illegal (Draghi, 2014b). The GC raised the initial PSPP issue share limit of 25% to 33% that the ECB can hold of a given PSPP-eligible bond (European Central Bank, 2015b). The initial issuer limit was, 56 Transcript available at: https://www.ecb.europa.eu/press/key/date/2012/html/sp120726.en.html [accessed 03.04.2020]. 57 The meeting was recorded https://www.youtube.com/watch?v=B4l1vL2uxew [accessed 03.04.2020]. 58 See details in the link http://curia.europa.eu/juris/celex.jsf?celex=62014CC0062&lang1=en&type=TXT&ancre= [accessed 03.04.2020]. 120 in their words, to avoid the ECB becoming a dominant creditor of Euro- area governments.59 After initial procedural revisions in 2016, the PSPP purchases would be €70bn per month, 20 percent of securities would be held by the ECB: 10 percent would be directly purchased by the ECB, purely in supranational securities, and the other 10 percent in domestic sovereign issuers transferred by respective NCBs. 10 percent of total purchases was targeted at supranational securities while 90 percent of purchases would be targeted at domestic issuers. That way each NCB focused on its respective domestic market for sovereign securities, see Figure 5.1. This, in itself, is quite interesting for the study of homophily in networks. If the domestic NCB is getting more active in the domestic bond market from a pure execution perspective, this could amplify local specificity in the respective bond market. For instance, the outline of the PSPP specifies in what ways the executing NCBs and the ECB are to communicate with the market and in what ways intermediaries are able to communicate trade positions and activities to other market participants (see chapter six). When participating in actual trade, monetary policy is not conducted in a vacuum and positions, activities and intentions are communicated to market makers and other market participants. Monetary policy and the estimated amounts of the purchase program are based on a priori assumptions of quantities of bonds, but cannot take into account how governments and issuers react to the increasing demand for bonds that the ECB’s policy itself creates. Figure 5.1 details these procedures. The NCBs conduct purchases of their government securities in line with the capital key. As of January 2019, the five largest countries in the capital key with eurozone membership are Germany, France, Italy, Spain and the Netherlands.60 These five largest country allocations in the PSPP will be analysed in detail below. Given the execution activity of the NCBs detailed in Figure 5.1, there already exists some home bias in that purchases are confined to respective NCB territories. The question arises how this behavioural organisation of the ECB’s monetary policy impacts market actors in the respective country networks. During the various phases of the APP, total monthly amounts varied from €60bn between March 2015 and March 2016, €80bn between April 2016 and March 2017, €60bn between April 2017 and December 2017, €30bn between January 2018 and September 2018 and 59 See question 3 https://www.ecb.europa.eu/mopo/implement/omt/html/pspp-qa.en.html [accessed 03.04.2020]. 60 For a full breakdown of the capital key, see https://www.ecb.europa.eu/ecb/orga/capital/html/index.en.html [accessed 03.04.2020] 121 €15bn between October 2018 and December 2018. The methodology was changed but in the first steps consisted of processes detailed in Figure 5.1. Figure 5.1 Monthly net purchases by agent at the start of PSPP and expansion of policy, based on ECB illustration (European Central Bank 2016) March 2015: After expansion, April 2016: During the 2012-episode of the eurozone crisis the sovereign bonds of the PIGS countries (Portugal, Ireland, Greece and Spain) were under heightened scrutiny and faced a widening of government bond spreads. During the 2018-episode of the eurozone crisis, Italy had come increasingly under attack from markets as populist fiscal budgets cautioned investors against Total PSPP €60 NCBs buy €55.2bn of their respective domestic sovereigns ECB buys €4.8bn Total PSPP €70bn National Securities €63bn NCBs buy €56bn according to capital key ECB buys €7bn according to capital key Supranational Securities ECB buys €7bn 122 taking on new Italian sovereign debt. Both Italy and Spain stand in stark contrast to Germany, whose sovereign bonds are seen as haven assets globally, and particularly in the eurozone. Papadia and Välimäki (2018) described the phenomenon as “…just too large not to be noticed, as about 10 per cent of total banking assets in Italy and Spain in 2015 (Affinito, Albareto, and Santioni, 2016) was invested in public sector debt” (Papadia & Välimäki, 2018, p. 253). A further debate evolved around Hans-Werner Sinn’s discovery of the unusually high Trans- European Automated Real-Time Gross Settlement Express Transfer System (TARGET2) balances on the German Bundesbank’s balance sheet (Sinn, 2011; Sinn & Wollmershäuser, 2012; Wolf, 2011). 61 TARGET2 entries in the balance sheet represent outstanding transfer payments owed by other central banks. At the height of the eurozone crisis of 2012, these balances were extremely high with Germany accumulating a significant surplus and Greece and other southern geographies a deficit. Market participants had linked these imbalances with liquidity issues and financial duress of members states in deficit. These imbalances widened again over the past few years coinciding with the implementation of the PSPP and some market observers have linked the policy with the rise in TARGET2 balances (Castillo & Varela, 2017). If the Bundesbank purchases German sovereign bonds from counterparties in other member states, this would contribute to the imbalances. It is beyond the scope of this chapter to tackle the issue of TARGET2 in great detail but this debate highlights the complex process of balancing individual countries’ cross-border monetary positions and capital flows within the eurozone and that the issue of jurisdictions is contentious. This underlines the need to research individual home biases below. A comparative analysis of the five largest country allocations, Germany, France, Italy, Spain and the Netherlands will thus serve as an interesting starting point to explore the notions of homophily and home bias in European sovereign bond markets. As detailed in section 5.2.4, Germany and the Netherlands will be included in the northern bloc, while France, Italy and Spain in the southern bloc. The ECB’s Financial Stability Review from March 2018 made obvious the need for Italy and Spain, as the two governments most dependent on debt markets, to roll over existing debt, illustrated in Figure 5.2. It showed, that as of March 2018, 61 More details under https://www.ecb.europa.eu/paym/target/target2/html/index.en.html [accessed 03.04.2020]. 123 Italy (IT), with around €35bn, and Spain (ES), with around €28bn, as the two governments with the most immediate debt servicing needs. Over the past 8 years, these two bond markets have been under repeated attack by markets and have seen spreads spiral high. It is thus particularly interesting to see whether Spain and Italy have a significantly different holder base to the bond markets of Germany and Netherlands and whether this structure is shaped by homophily or home bias. Figure 5.2: Total debt servicing needs due in 2 years Source: Financial Stability Review, ECB, March 2018. 5.1.3 Data Collection Weighted data was used in chapter four, while the results of chapter five below were adjusted into binary networks for the analysis of home bias and homophily. Data from April, September, December 2018 and March 2019 was used to study the historical development of the E-I and HHI to test Hypothesis 1. Data from March 2019 was analysed in detail to test Hypothesis 2 to 4 as it best resembled the time in which the effects of the halting of purchases could be observed. The PSPP program was the largest sovereign bond purchase program to be conducted to this date, so that the bonds included in the program should be highly representative of the sovereign bond markets in the countries included. Judging from the extremely large purchase sizes (cumulative net purchases of >€2tln as of May 2019) the ECB is by far the largest holder. For reference, the total outstanding sovereign €-denominated debt stands at €10.06tln as of q2 2019, so that the ECB and its NCBs hold around ~20% of total outstanding debt. 124 For the purposes of this study, the ECB was categorised as 0 (supranational) for its node attribute ‘Country’ and 3 (Government) for its node attribute ‘Investor Type’. This was done to avoid distorting the findings for home bias and homophily studies. While the ECB is based in Frankfurt, it holds a supranational mandate and should be seen as such in the context of this particular study. All other holders are categorised by domicile of the ultimate beneficiary of the vehicle holding the securities. For instance, if the iShares € Government Bond 1 – 3 years ETF holds the securities, the node will be Blackrock US, unless the vehicle is held under a separate entity, domiciled in a different country, e.g. Blackrock Ireland. 5.2 Results 5.2.1 Network demographics and statistics As of 31 March 2019, the PSPP consisted of 844 securities across 18 distinct issuer countries. According to Decision (EU) 2015/774, the 19th country Greece was excluded from the PSPP given its financial assistance program.62 There were ~530 nodes among the top 20 holders of the 844 securities, which was relatively stable over the prior 12 months of the study. It is noteworthy to compare the 530 nodes with the number of possible nodes of over 16,880 (844 securities x 20 nodes). Among the top 20 holders of securities in the whole-network, there are on average 25 holder countries of domicile represented. The largest exposure by asset size in € of the PSPP was Germany, France, Italy, Spain and the Netherlands in descending order. The study examines both country of domicile for the nodes – the holders of securities – but also the issuer country of the sovereign bonds itself. For instance, we can look at Spanish nodes and their activities in the whole-PSPP network, spanning sovereign bonds of all countries included in the PSPP, or solely in the network of Spanish sovereign bonds. Hence for clarity, there are holder countries, countries from nodes that invest in the PSPP securities and issuer countries, countries of government, sovereign and supranational institutions that issue the debt securities. Table 5.1 shows the network statistics for the five largest Sovereign bond networks within the PSPP whole-network. Spain exhibits a comparatively dense network, especially when 62 The decision is outlined in this document: https://www.ecb.europa.eu/ecb/legal/pdf/oj_jol_2015_121_r_0007_en_txt.pdf [accessed 03.04.2020] 125 adjusting for a threshold of >2% holdings of constituent bonds. Interestingly, the Netherlands seems to have the most divergent network structure, even when taking into account the edges at a >2% holder threshold. As the largest market, Germany is comparatively less dense, indicating a more diversified holder base. Table 5.1 PSPP network statistics by country at >0% and >2% threshold 5.2.2 Whole network homophily measures As discussed below, the actual execution of the program is conducted by the respective NCBs in their own domestic markets, so the Deutsche Bundesbank buys German Bunds, the Banque de France buys French sovereign bonds and so on. Homophily in the local group of nodes in the respective local market is one aspect of home bias. Krackhardt (1988) found that there is an increased commitment to local nodes in times of crisis. Given a certain level of home bias, local nodes are more likely to have homophilic tendencies in their home market of PSPP securities. Hypothesis 1: There is a divergence between expected and actual levels of in-country connections for nodes in the PSPP bipartite network. Connections in the PSPP network are a result of past securities selection by the AMC’s PMs. There are numerous opportunities to form links with market participants in financial markets. As argued above, the decisions contributing to the financial network formation can either be a) at random, b) an imitation of nodes B, C,…,n holding the security before A or c) adherence and cohesion with decisions by a central agent Z. If randomness can be eliminated, 126 there are either shared decision-making processes that derive at the same conclusion, or imitation based on certain attributes. In Hypothesis 1, I test for homophily in the Country attributes. Krackhardt and Stern’s (1988) E-I Index is a measure of homophily and is calculated by subtracting the ingroup links (IL) from the external group links (EL) and dividing that number by the total number of connections. 𝐸 − 𝐼 𝐼𝑛𝑑𝑒𝑥 = 𝐸𝐿 − 𝐼𝐿 𝐸𝐿 + 𝐼𝐿 The theoretical range is thus from -1 (all links are internal) to +1 (all links are external), but if the Node 1’s country group only has 2 members and the whole-network 530, Node 1 will likely have a high E-I index value. A permutation test calculates the expected value of the E-I Index given the amount of nodes and edges in the network as well as a range of likely values given a certain network structure (Borgatti et al., 2013). The initial idea of the index was based on the assumption that conflict leads to increased commitment to the home group and decreased commitment to external groups. Krackhardt and Stern’s (1988) assumption was also that adaptation to a crisis requires increased co-operation, thus a need for more external group connections. The E-I index is used to measure homophily and social cohesion in that it indicates the behaviour of nodes with the same country origin across the eurozone, so behaviour at home and away from home. Given the relative density of the network, the high number of bonds and monetary size at €10.06tn, the number of nodes and issuer countries, these networks should in theory display almost pure heterophily. The fact that there are usually at least 25 holder countries present, indicates a high likelihood of external links exceeding internal links. The results in table 5.2 analyse homophily based on the node’s country. A high E-I index value thus indicates a dominance of external over local connections. However, the larger a network gets, the more likely there are external connections as the number of nodes from other countries increases. Based on the E-I index values over the past year, the findings show the whole PSPP network as slightly more homophilic than the expected value. To further examine homophily in the sovereign bond markets of member countries, I present an analysis of the holder concentration below by country. The HHI is a measure of 127 concentration in a given population which has been widely used in management studies and sociology (see e.g. Barro & Mccleary, 2003; Iannaccone, 1991; Podolny, 1993). It is closely related to the sociologist’s Blau Diversity index, which subtracts the sum of squared constituent proportions from 1 (Blau, 1977, p. 78). Diversity in a market is hereby taken to be a low reading in the HHI. Given the European Commission’s usage of the HHI, this index will be used below. The HHI is defined as the sum of squares of constituent portions of given groups of a population and that value can be used as a comparative measure (Hirschman, 1964). 𝐻𝐻𝐼 = 𝑀𝑆1 2 + 𝑀𝑆2 2 + ⋯ + 𝑀𝑆𝑛 2 A value of >1000 in Europe usually indicates a moderately consolidated market. Values can range from 0 to 10,000 with 0 indicating diversity and 10,000 a monopoly. Table 5.2 shows HHI values for selected countries and for the network as a whole on four data collection dates of April, September, December of 2018 and March of 2019. Notably, the overall HHI slowly rises from 751 to 792 over this observation period of the PSPP’s decline in net purchases, indicating rising consolidation and concentration of holders. The absolute HHI value is moderate, however Spain seems to have a particularly strong influence compared to the larger economies of the eurozone and its overall weighting in the capital key. It is also interesting that nodes based out of Luxembourg, have the highest HHI value after the US. The mismatch between the capital key and the level of influence of the nodes’ domiciles demands more detailed research into the PSPP sovereign debt networks of specific countries below. 128 Table 5.2 Whole-network PSPP E-I Index and HHI, April 2018 – March 2019 Overall, values indicate some homophilic tendencies and divergence from expected values on a whole-network basis but more pronounced tendencies when looking at a specific group of nodes within the network, notably Spain with an E-I index ranging from 0.44 to 0.47. Taking 129 the HHI values over the past year, it can also be concluded that the reduction of ECB purchases and the NCBs has led to a slight consolidation in the PSPP-holders network and further holder concentration. In the example of Spain, for instance, this is negative as its E-I index value is particularly low. More divergence of expected homophily values will be dealt with on a country basis in the following section. 5.2.3 Homophily in nodes and home bias in local bond markets Hypothesis 2: The home country group of nodes exhibit more homophily and higher influence in their home market compared to other country group of nodes, displaying home bias. Home bias is hereby defined as the ratio of the number of connections by the group of local country nodes (LNC) divided by the total number of connections (TC) in a given network (i). 𝐻𝐵𝑖 = 𝐿𝑁𝐶𝑖 𝑇𝐶𝑖 This measure gives a sense of the level of influence local nodes take over nodes from other countries on the local network structure as a whole. Table 5.3 shows the empirical results for selected country networks of sovereign bonds. This is a representation of the government bond market in selected countries. The Netherlands and Germany present the markets with the lowest level of home bias at 6% and 11% respectively. France, Italy and Spain on the other hand have higher levels of home bias in their markets at 14%, 15% and 25%. Using the individual country HHI values, Spain is moderately consolidated as measured by the European Commission guidelines (see section 5.2.5). In combination with the elevated levels of home bias, Spain seems highly vulnerable to an ECB withdrawal from said bond market. Based on the amalgamation of these measures, the five countries studied could be boxed into two different categories. As I argued in Section 5.1.1 in relation to Marsden’s (1988) findings, moderate homophily in a network infers local integration and global disintegration. Hence in this study, globally integrated and oriented sovereign bond markets are those of the Netherlands and Germany. France, Italy and Spain are locally integrated and domestically oriented bond markets with a higher level of home bias. Additionally, France, Italy and Spain 130 are more concentrated and denser bond markets compared to their neighbours based on HHI values. Table 5.3 Statistics of Sovereign bond markets in selected countries Table 5.4 exhibits the in-country connection frequencies compared to the population’s country breakdown of nodes. If the PSPP network formation would be perfectly random, the in-country connection frequencies should exactly match those of the % breakdown of nodes. As seen in previous studies of religious affiliation and friendships such as Baerveldt et al. (2004), these values give the expected connection frequency. Hence in table 5.4, I define the extent of homophily as the % of in-country connections over the relative % of countries in the population. Below, Spain, Malta, Austria and Italy have the highest nominal divergence of eurozone countries and indicate a higher level of inbreeding homophily among these groups of country-nodes. This table also indicates that overall, the Southern Bloc of eurozone countries has a higher level of inbreeding than the Northern bloc given a higher divergence from the expected proportion of in-country connections. Interestingly, as one of the largest asset management hubs in the eurozone, Luxembourg actually has less in-country connections than the population distribution would indicate – a small geography with lots of fund managers. Likewise, Ireland has become a domicile for a plethora of funds and investment vehicles over the past 15 years, given a lower tax environment and a population of able administrative staff. Like Luxembourg, the lower than expected in-country connections infer the domicile harbouring a heterogenous population of nodes. This indicates the diversity of market participants based out of Luxembourg and Ireland. 131 Table 5.4 Inbreeding homophily: Frequency of in-country connections by country breakdown (%) (n=530) The results in Table 5.5 below are adjusted for eurozone membership and expressed as a ratio of actual vs expected in-country connections. With the overall larger size as network constituents, Spain and Italy remain significantly homophilic with around double the expected in-country connections. Luxembourg and Ireland remain the heterophilous stalwarts of eurozone domiciles. Greece is a bit of an outlier since the country is not included in the PSPP program and domiciles only a small host of nodes so that, by definition, external connections are more likely. PSPP Connection Frequencies (%) compared to Population % by country (March 2019) Expected % Actual % Country of Domicile Percentage of total Population Percentage of in-country connections Inbreeding Homophily Spain 10.94% 26.9% 16.0% Malta 0.75% 14.3% 13.5% Austria 3.40% 9.4% 6.0% Italy 6.42% 12.2% 5.8% Japan 2.45% 6.7% 4.3% Germany 6.79% 9.1% 2.4% Slovenia 1.32% 3.4% 2.1% France 8.30% 10.2% 1.9% Netherlands 2.45% 4.1% 1.7% United States 12.83% 14.4% 1.5% Poland 0.75% 2.3% 1.5% Switzerland 4.53% 5.9% 1.4% Canada 2.83% 3.7% 0.8% Finland 1.51% 2.1% 0.6% Belgium 2.26% 2.6% 0.3% Australia 1.13% 1.4% 0.2% Portugal 1.70% 1.9% 0.2% United Kingdom 4.15% 4.2% 0.1% Supranational 0.19% 0.0% -0.2% Czech Republic 0.19% 0.0% -0.2% South Africa 0.19% 0.0% -0.2% Cyprus 0.19% 0.0% -0.2% Isle of Man 0.19% 0.0% -0.2% Hungary 0.19% 0.0% -0.2% Other 0.19% 0.0% -0.2% Taiwan 0.19% 0.0% -0.2% Bermuda 0.38% 0.0% -0.4% South Korea 0.38% 0.0% -0.4% Thailand 0.38% 0.0% -0.4% Greece 1.51% 1.0% -0.5% Denmark 0.57% 0.0% -0.6% Liechtenstein 1.32% 0.6% -0.7% Norway 0.75% 0.0% -0.8% Sweden 1.89% 1.1% -0.8% Ireland 1.89% 0.9% -1.0% Luxembourg 14.91% 12.4% -2.5% 132 Table 5.5 PSPP in-country connections (actual vs expected) for eurozone members The network data presented in this section have shown that both on a country and whole- network basis, nodes domiciled in countries of the southern eurozone have both a home bias in local markets and homophilic tendencies in the whole-network. In other words, this shows that AMCs in France, Spain and Italy tend to hold the same securities both locally, as well as across the eurozone. On a local market level, these measures vary significantly insofar as Italy and Spain, and to a lesser extent France, can be categorised as locally integrated, inward-looking and concentrated markets with a significant home bias. While nodes domiciled in Germany and the Netherlands show only slight homophily compared to their expected values, the local bond markets in these countries are globally oriented and diverse. Having established the categories of the northern and southern bloc for our five countries under study, I now turn to the home bias and homophily in culturally similar blocs. PSPP Eurozone Member Connection Frequencies (%) compared to Population % by country (March 2019) Expected % Actual % Country of Domicile Percentage of total Population Percentage of in-country connections Inbreeding Homophily Ratio Malta 0.75% 14.3% 18.93 Austria 3.40% 9.4% 2.76 Slovenia 1.32% 3.4% 2.57 Spain 10.94% 26.9% 2.46 Italy 6.42% 12.2% 1.91 Netherlands 2.45% 4.1% 1.68 Finland 1.51% 2.1% 1.41 Germany 6.79% 9.1% 1.35 France 8.30% 10.2% 1.22 Belgium 2.26% 2.6% 1.14 Portugal 1.70% 1.9% 1.11 Luxembourg 14.91% 12.4% 0.83 Greece 1.51% 1.0% 0.66 Ireland 1.89% 0.9% 0.46 Cyprus 0.19% 0.0% 0.00 133 5.2.4 Regional Homophily, Southern Bias Homophily can occur both by individual country but also by blocs of countries that are either geographically close or share certain attributes. This leads to the following hypothesis. Hypothesis 3: Nodes based in either the northern or southern bloc display a higher level of influence in their respective blocs and preference for sovereign debt of similar countries to the home country. The southern periphery has been depicted as a closed market, while northern markets are portrayed as more outward looking. Another defining characteristic has been support for QE in the Southern bloc, while the Northern bloc has been more hawkish and opposed to QE. Should a country be more opposed to QE overall, investor preference would gravitate to ‘safer’ northern bloc debt, rather than that of the Southern periphery. This leads to the last hypothesis. Hypothesis 4: More inward-looking southern markets of the eurozone exhibit a higher level of home bias when compared with the northern markets. Strong home bias in the network would imply low cross-border risk sharing and inter-country relationships. This would work against the ECB’s aims and policy objectives to increase cross-border and private sector risk sharing and cross-border portfolio flows. These tendencies would both hinder the transmission mechanism of monetary policy to the economy in that capital will not flow indiscriminately throughout the eurozone but is fragmented into different territories. For purposes of robustness countries are added to the Northern and Southern bloc to balance the numbers. Austria, Finland and Belgium are added to the Northern bloc and Portugal and Malta to the Southern bloc. These countries have been in various contexts connected to the existing members of the two blocs respectively. Both are either geographically proximate or linguistically closer and aligned with the existing bloc members in the attitude towards the ECB’s purchase programs. After filing a statement of a shared vision of the future for the EMU, some of the countries in our defined Northern bloc were referred to as constituting the Hanseatic league (see e.g. Acton & Brunsden, 2017; Khan, 2018; The Economist, 2018). 134 The findings in Table 5.6 show that the Northern bloc shapes 23% and 24% of the total edges in the networks of the German and Dutch sovereign bond market. Even more strikingly, the Southern bloc claims 40% and 44% of total edges in the networks of the Italian and Spanish sovereign bond market. The contrasting levels of bloc influences are of equal significance. The Northern bloc constitutes only 14% and 15% of total connections in the networks of the Italian and Spanish government bond market respectively, while the Southern bloc makes up only 18% and 16% of total edges in the networks of the German and Dutch government bond market. At almost half the proportion of total edges (44% and 40%), both Spanish and Italian sovereign bond networks seem to be significantly shaped by nodes domiciled in the Southern Bloc. In other words, nodes from only five domicile countries shape almost half of the network of sovereign bonds in Spain and Italy. The Northern bloc influence of 23% and 24% of edges in the networks of German and Dutch sovereign bonds is more in line with an equally weighted % distribution of country nodes. The German sovereign bond network has 27 constituent countries and a simple average distribution of five constituent countries would yield 18%, not too far from the 23% recorded below. This further indicates that the level of bloc bias is more pronounced in Italy and Spain, while France is somewhere in the middle. Table 5.6 PSPP Individual Country Network Statistics: Bloc influence These results underline homophily and home bias in the holdings-based networks of both country group of nodes and local markets. These network characteristics extend to blocs of countries suggesting either a geographic or cultural fragmentation within the eurozone. This is particularly pronounced in the southern bloc. 135 5.2.5 Notable national sovereign bond market characteristics While it is plausible to expect divergences from the predicted values of home bias, the findings underline the outliers and extremes. As outlined in section 5.2.4, the southern bloc of nodes consists of those domiciled in France, Italy, Spain, Portugal and Malta. France is chosen as a member of the Southern Bloc as the network statistics found it to align more with other southern markets. The Banque de France, the central bank of France, has also been more supportive of expansionary monetary policy rather than the more hawkish stance of the Bundesbank or members of the Hanseatic league. Looking at table 5.6 and the holder network of Spanish sovereign securities shows a strong 25% home bias. Spanish based investors are also the second largest holders of Italian PSPP securities. Should there be financial distress in Spain, this can easily spread to Italy, and from Italy through e.g. Luxembourg based investors further afield. The Spanish asset management industry is relatively small in comparison with its peers, which could be a concern. As a reference, the ten largest Spanish asset managers are only 30% the size of the largest French asset manager or roughly equal to the second largest German asset manager.63 With the Spanish nodes representing such a central role in the network and the Spanish sovereign bond market being dependent on comparatively small local market participants, the country is more vulnerable in crises. Based on the European Commission guidelines on horizontal mergers, there are concerns should there be a merger resulting in a change of 250 HHI values in an industry with a HHI between 1000 and 2000.64 Industry consolidation has a significantly higher threshold of regulatory intervention than looking at simple biases in financial networks of bond holders. Hence, given that the Spanish PSPP network of holders reach the threshold of the European Commission, this should be seen as a significant indication that the Spanish sovereign bond market is highly concentrated. Table 5.7 shows the HHI values for the Spanish PSPP network. 63 For reference see IPE Hub Research https://hub.ipe.com/top-400/spanish-asset-manager-tables- 2018/10007228.article [accessed 03.04.2020]. 64 See point 20, page 3 https://eur-lex.europa.eu/legal- content/EN/TXT/PDF/?uri=CELEX:52004XC0205(02)&from=EN [accessed 03.04.2020]. 136 Table 5.7 Herfindahl Index Values, March 2019 US-based market participants display a higher level of homophily in all eurozone networks under study. Additionally, these are the largest market participants by asset size. What characterises these nodes was also seen in chapter four, in that they mostly pursue a passive investment strategy which may explain the consistently higher level of homophily as they track a common index. Another assumption here could be that a ‘when abroad, we stick together’ mentality is pursued or that there are other factors influencing security selection for nodes based in that geography. 5.2.6 Analysis In Krackhardt’s (1988) original work and in recent research across various disciplines (see e.g. Constâncio, 2014; Gabrieli & Salakhova, 2019; Tortoriello & Krackhardt, 2010), the benefits of external links were highlighted. A dense network with low outgroup linkages is ill-equipped to deal with crises. However, increased global integration requires increased private sector risk sharing. During the data collection period of the study, the ECB first reduced the net purchases under the PSPP, then announced the likely end of purchases and finally halted all net purchases in December 2018. During this time, the Spanish sovereign bond network increasingly consolidated. Unlike the CSPP analysed in chapter four, in which different NCBs conduct the purchases, the NCBs corner the domestic market for PSPP purchases. This behaviour perhaps amplifies the segregation and home bias present in the network structure. The ECB’s withdrawal of net purchases also shows that the networks of the Southern blocs, particularly Spain, are increasingly dependent on the ECBs monetary policy for outgroup connections. The findings also show that the ECB’s monetary policy does not lead to greater cross-border investments and thereby increases the dependency of vulnerable countries discussed in section 5.1.2 (figure 5.2), Italy and Spain, on the ECB. H e rf in d ah l I n d e x V al u e s, S p ai n P SP P N e tw o rk (M ar ch 2 0 1 9 ) S u p ra -N a ti o n a l 4 .1 A u st ri a 6 .1 B e lg iu m 2 .1 B e rm u d a 0 .1 C a n a d a 2 .5 D e n m a rk 0 .2 F in la n d 0 .6 F ra n ce 1 4 6 .8 G e rm a n y 7 0 .2 G re e ce 0 .0 Ir e la n d 0 .3 It a ly 3 4 .1 Ja p a n 6 .7 Li e ch te n st e in 0 .1 Lu xe m b o u rg 5 2 .9 N e th e rl a n d s 3 .6 P o rt u ga l 0 .9 Sp ai n 6 3 2 .0 S w e d e n 2 .9 S w it ze rl a n d 2 4 .1 U K 2 9 .0 U S 1 8 8 .3 to ta l 1 2 0 7 .8 4 137 One limitation of using the E-I index for this research is that it works best for equally sized groups. To address this limitation, the findings included longitudinal data in order to be able to compare values across time and thus reduce concerns about unequal group sizes. I have also looked at overall HHI values to complement the measures of concentration and measures of diversity. Northern sovereign bond markets are more diverse on this measure compared to those of the southern bloc. Home bias was defined and introduced as the ratio of connections of local nodes to the total number of connections of the network as a comparative measure. This showed a heightened bias of local nodes in the southern bloc. A common explanation for home bias in the economics literature is that higher transaction cost and foreign currency exposure deters investors to venture beyond national boundaries. In this study, the ECB’s monetary policy has only one base currency and bid-offer spreads are on par for liquid sovereign bonds, independent of the country of the market maker. Given the findings on homophily in the PSPP network, these network characteristics demand more research and underline the inherent problems with both the tenets of a) the portfolio diversification theory and b) the ECB’s monetary policy objective of higher integration of and cross-border investments in the eurozone markets. Home bias and homophily can be characterised as ‘group-think’. It is encouraging to see diverse networks in Luxembourg and Ireland. Unsurprisingly, these are the rising domiciles of the asset management industry. Despite being the second largest asset management domicile in the eurozone, AMCs based in Germany exhibit a surprisingly high value of homophily. These network structures across European sovereign bond markets are perhaps a result of the decision-making structure and practices in said locations and will be discussed in chapter six and seven, both in connection with Frankfurt as a central location in the social networks of ECB experts and also in the investment practices gathered from interviews with senior PMs in Frankfurt. 5.3 Conclusions In this chapter, homophily and home bias was studied in the sovereign bond markets the ECB is active in. Capital markets in the eurozone are often conceptualised from a top-down approach in terms of supranational indices such as the Euro Stoxx 50 or sovereign spreads 138 over German Bunds, the quasi risk-free rate. The amalgamation of national securities into a supranational structure as seen in this study of the PSPP program interweaves domestic markets with distinct social characteristics and preferences shaped by different social attributes. As currently practised and designed, monetary policy conducted in supranational indices inadvertently impacts distinct financial networks and habitats, rather than the sum of constituent country networks. Modern portfolio theory and the concept of diversification does not seem to take precedence over social influences, such as in which country the bond market is and where the investors are from, shaping decision-making processes and practices. Familiarity, broadly conceived, plays a crucial part in blocs of geographically and culturally close bond markets. As central bankers attempt to shape government bond markets – by creating conditions through monetary policy that encourage cross-border risk sharing - the structure of the government bond markets in the eurozone remains idiosyncratic, fragmented and socially shaped. The ECB needs to take these factors into consideration when conducting monetary policy as an active market participant. The current approach and strict geographical separation of execution duties of the ECB and the NCBs in figure 5.1 may have amplified the social fragmentation in the financial network studied. The networks around the PSPP securities show home bias and homophily, particularly in the southern bloc of countries. To reduce the level of home bias in domestically oriented sovereign bond markets of the Southern eurozone, one could introduce a different execution in which NCBs of the northern eurozone could conduct the purchases in southern parts of the eurozone and vice versa. In the CSPP, there is some more flexibility where purchases are spread across six NCBs and where four factors are used to determine the country category, namely management location, country of primary listing, country of revenue and reporting currency of the issuer.65 As a result, this may encourage non-eurozone based nodes to decrease their own levels of homophily such as in the case of the US-based nodes in southern European sovereign bond markets. If a diverse ownership network is the goal, the NCBs would also have to be active across national boundaries. 65 See Question Q2.4 Who conducts the purchases? https://www.ecb.europa.eu/mopo/implement/omt/html/cspp- qa.en.html [accessed 03.04.2020]. 139 As discussed in Krackhardt’s (1988) work, in a time of crisis, external links help to mitigate a crisis. Whether nodes from the same country of domicile have a similar decision-making process to arrive at the same conclusion or whether there are social ties between nodes of the same country will be further dealt with in chapter six. However, social and cultural preferences enter financial networks in various ways, and with the qualitative interview data presented in chapter seven, the aim is to explore the social networks within the holdings- based networks of the PSPP. Network analysis has shed light on the network structure of the corporate and sovereign bonds holders in the eurozone from one standpoint. As the results indicate, social networks in national markets alongside the PSPP holdings-based networks contribute to each other’s network formation. The following two chapters will explore the social networks of individuals within the institutions examined, and will shed light on the shared social attributes, interactions, information, knowledge or practices. How does the ECB communicate its courses of actions and how are these interpreted by PMs and asset allocators? Do PMs meet and share information which influences investment decisions? Chapters six and seven will tackle these questions in detail. 140 6 Expert Networks and Sensemaking: The case of the Bond Market Contact Group “Unconventional policy prescriptions and ruminations about the longer-term outlook for economic and financial market developments might never be surfaced at meetings, for fear of igniting a speculative reaction when the discussion was disclosed” (Greenspan, 1993, p. 4). “The competitive advantage of brokerage does not come to people who passively wait for the network to deliver it. The advantage provided by network brokerage depends on personal engagement with conflicting opinion and practice” (Burt, 2010, p. ix) 6.1 Introduction Central bank communication has become a well-studied phenomenon across the social sciences with central banks transforming from secretive to increasingly transparent institutions. Blinder referred to this as the ‘quiet revolution’ (Blinder, 1998, 2004). As seen in chapters four and five, the ECB’s monetary policy has evolved since the GFC and communication has grown commensurately, both in terms of the amount of data published by central banks, but also in the way communication is assumed to shape expectations of market participants. In addition, communication itself has become a policy tool for central bankers to amplify and implement policy (Asmussen, 2012; Blinder et al., 2008; Ehrmann & Fratzscher, 2007; Papadia & Välimäki, 2018). Forward guidance is the practice of the central bank’s commenting and evaluation of the current economic situation and giving guidance on likely courses of policy action. Over the past ten years, forward guidance has included a very detailed path of future asset purchases of the QE programs thereby aiming to harmonise expectations and resultant behaviour. If the market reacted unexpectedly to certain communications, central bankers concluded that the communication was unclear. The central tenet of forward guidance is greater transparency, the assumption that markets work better and that policies are reinforced with forward guidance (Kohn, 2018). In the recent past, central banks, including the ECB, have regularly hosted conferences dedicated to communication and its analysis.66 Not only has the role of central bank communication 66https://www.ecb.europa.eu/pub/conferences/html/20171114_communications_challenges_policy_effectivenes s.en.html [accessed. 03.04.2020]. 141 changed, but given the backdrop of unconventional monetary policy of the past decade, the demand for more information for market participants’ sensemaking has increased. The highly cited Alan Blinder defined central bank communication as "…the provision of information by the central bank to the public regarding such matters as the objectives of monetary policy, the monetary policy strategy, the economic outlook, and the outlook for future policy decisions” (Blinder et al., 2008, p. 913). Podolny (2001) contended that information and resources flow through ‘pipes’ between nodes in networks which are embedded in markets. While this PhD thesis has used the corporate interlocking networks of chapter four and five to analyse the financial resources that constitute such pipes, this chapter provides a re-focus on communication as information exchange with PMs in the asset management industry. As discussed in chapter one (see section 1.2.1 and 1.2.3), central bankers are reliant on active market participants from the asset management industry in helping policy transmission. Furthermore, central banks rely on capital markets for feedback on their policies and capital markets are in turn, largely shaped by decision-making of the asset management industry. In this chapter, central bank communication will be dealt with as the continuous information exchange in expert networks of the investing elite. It will be elaborated in what ways the ECB uses different channels of information exchange and in what ways this resembles co-optation. The focus of this analysis centres on the variability of and harmonisation in interpretations for social conduct in financial networks based on information that is exchanged. This is thus a qualitative documentary exploration of themes explored in chapters four and five such as uniformity of decisions and social influence. By focusing on the case of the BMCG it is the aim of this and the subsequent chapter to highlight the asymmetric and non-neutral nature of information exchange in financial networks and the contextualisation of communication in investment decisions. Central banks are seen as just one actor in the information exchange of financial networks. The application of the economic sociology of networks to information exchange adds focus on the influence actors with decision-making responsibilities have on shaping consensus and on the ECB’s own evaluation of market situations. As discussed in chapter one, the area of research that has been de-emphasised thus far is the exploration of the ways in which central bank communication is interpreted, conceptualised and incorporated in sensemaking processes and investment decisions. It addresses the missing link when analysing central bank communication, namely actual investment decisions and asset allocations resulting from 142 these information exchanges. It also explores Weick’s (1988, 1995) process of sensemaking in different expert networks of the ECB derived from collaborations and face-to-face meetings. It is concluded that members that can mediate between expert networks can arbitrage knowledge in the vein of Seabrooke’s (2014) concept of epistemic arbitrage and Burt’s (1992) notion of structural holes. Structural holes exist between clusters of nodes in a network and experts who can forge connections between divergent networks can extract value as information brokers. The value here lies in information that can help create excess returns by correctly anticipating behaviour of big market actors such as the ECB and results in more informed investment decision-making. The chapter contends that there is a need to study central bank communication with an inductive approach and focus on the context, as words derive from a context of agency. Monetary policies are interdependently and socially shaped and re-shaped in information exchanges of expert networks. This is shown in both the analysis of information exchange in contact groups of the ECB, and in the ability of a central bank leader to instigate paradigmatic and policy change by gauging social and political will for such change. The chapter is structured as follows. Section 6.2 briefly details the evolution of central bank communication from secrecy to transparency. Section 6.3 looks at how the ECB actually communicates with markets and highlights recent developments to that effect. It analyses the BMCG in detail using network data, industry reports, ECB transcripts and details the diffusion of a new ECB policy, the PSPP in 2014/2015. Section 6.4 details the mechanism of central bank communication as information exchange in networks of investment management professionals, with insights based on documentary research. As discussed in section 1.4.2 of this thesis, I do not rely on auto-ethnographic recollections of events but find evidence in the documents and texts in Section 6.4. 6.2 A Background to Central Bank Communication: From Secrecy to Transparency To put into context the current state of central bank communication at the ECB in the era of unconventional monetary policy, the following section briefly describes the well-documented 143 trend towards greater transparency in central bank communication through the lens of the US Fed. This is necessary given the longer history of the Fed and the relatively recent practices of the ECB. Prior to the 1990s during the phase of the ‘monetary mystique’, central bank communication was seen in terms of its ability to surprise market participants, i.e. the opposite of what constitutes transparency where there are no surprises. Market participants would have to look at bond yields to deduct the Fed’s policy shifts. This paradigm assumed that cryptic communication constituted information that is both non-public and material.67 This is an important fact that an observer should keep in mind when analysing central bank communication. Early accounts of central bank communication derived from central bankers themselves, highlighting the secrecy and mystique in which the Fed shrouded itself. The hitherto longest serving Fed Chair Bill Martin (1951 – 1970) was a market practitioner before joining the US central bank. Former Fed Director of Staff Stephen Axilrod found that Martin’s “influence on the substance of policy was grounded largely in his colleagues' belief that his sensitivity to market psychology (that is, to the evolving attitudes of key participants in credit markets and businesses) was unusually apt" (Axilrod, 2009, p. 25). Axilrod (2009, p. 4) contended that a great central bank leader is not differentiated by her or his economic sophistication, but more by the ability to gauge when social and political boundaries can be pushed to introduce a paradigmatic shift in policy and execute such a shift. Here, Mario Draghi’s shift in central bank communication discussed in the section 6.3.1 is an apt illustration for this. Goodfriend (1986) summarised the state of central bank communication at the time in a much-quoted article. He contended that there was political dissatisfaction with the level of secrecy around the Fed and that given the value of the information the Fed communicates, large number of Fed watchers were employed to decipher said communications. Up until 1994, monetary policy decisions were not publicly announced as is current practice but market participants were to deduce the decisions based on movements in interest rates. In his 67 If information is non-public, i.e. not known to the public at large, it is either derived from superior information or based on the internal modalities of the decision-making process in the central bank. If information is material, it can and should move market prices. 144 1984 letter to the Subcommittee on Domestic Monetary Policy, previous Fed Chair Paul Volcker (1979 – 1987) pointed out that secrecy was a necessity to ensure functioning markets and a level playing field (Goodfriend, 1986, pp. 76–77). The heart of the problem, as I see it, is that markets constantly are trying to anticipate what might happen in the future. They would like the Fed to in effect ‘tell’ them. But, by the nature of things, we cannot. Our own operations in the market from day to day are dependent upon future events - some technical, some not - that we cannot reliably forecast with accuracy. One danger in immediate release of the directive is that certain assumptions might be made that we are committed to certain operations that are, in fact, dependent on future events, and these interpretations and expectations would tend to diminish our needed operational flexibility [emphasis added] (Volcker quoted in Goodfriend, 1986, p. 76). This was echoed by Axilrod in that depending on how information is disseminated, some market participants, mostly investment banks at that time, would have an unfair informational advantage, would trade and build positions based on such information (see Goodfriend, 1986, pp. 71–77). In addressing central bank communication with trading counterparties, Paul Volcker’s own account of working for the New York Fed on the trading floor in the 1950s is particularly illuminating. Literally months went by before I was permitted to talk to a dealer and then to actually make a trade. The concern was that the words we used or the tone of our voice might inadvertently tip off our counterparties to a change in the direction of monetary policy (Volcker & Harper, 2018, p. 36). During his tenure, the chief trading officer of one of the Fed’s prime brokers coached Volcker on how he “should precisely, or perhaps, more obscurely communicate with the market” (Volcker & Harper, 2018, p. 36). Volcker went on to state that “the method of conducting and communicating monetary policy at that time gave the trading desk in New York more leeway to react to the market” (Volcker & Harper, 2018, p. 36). The overarching concern was equality in information dissemination in order to circumvent profitable trading based on informational advantages obtained by certain market participants, but also, to not let 145 the Fed be beholden to market participant positioning.68 Volcker’s conclusions still rang true during former Fed Chair Janet Yellen’s tenure (2014 – 2018). In a 2017 panel discussion, Yellen mentioned that she herself and her colleagues were caught entirely off-guard when the ‘taper tantrum’ unfolded in 2013, and when the Fed decided to back-paddle its rate hike projections in 2015 (Draghi, Yellen, Carney, Kuroda, & Wessel, 2017). Likewise, after guiding the Fed’s resolve to normalise policy by reducing the balance sheet and raising interest rates in late 2018, current Fed Chair Jerome Powell quickly reversed the policy normalisation, as the extent of the market reaction was particularly negative. Volcker served for two terms (1979 – 1987) and was replaced by Greenspan. Fed Chair Greenspan (1987 – 2006) took a slightly more Machiavellian approach in acknowledging that communications about intended courses of policy actions can drive market movements and was very much involved in communicating with markets. However, the central bank mystique continued with what he referred to as ‘purposeful obfuscation’, which he defined as the usage of words to give the impression of answering a question but not really answering it (El-Erian, 2016, p. 18; Jasper, 2007). Greenspan’s communication-style was coined ‘Fedspeak’ or ‘Greenspeak’ (Dauble, 2007). Rather than relying on the Fed's staff model to assess the effect of tightening, Greenspan had taken to gaming through the market's probable reactions in his head. If monetary policy operated through the psychology of traders, then he, too, would think like a trader, summoning up the lessons he had learned in the commodity-futures pit. (Mallaby, 2016, p. 461). From 1994, policy decisions at the Fed were publicly announced. The presumably close encounters with these changes led Fed Vice-Chair Alan Blinder, then confidant of Bill Clinton, to coin the term ‘quiet revolution’ (Blinder, 1998, 2004). Central banks have become increasingly transparent about their views and analysis of key elements of monetary policy over the past few decades. The 2008 GFC furthered the need for policy certainty from the Fed amid market uncertainty around the sub-prime mortgage crisis and failure of the large investment banks. The fundamental tenet behind this has been that 68 In other words, it may be difficult for the Fed to react to new incoming data and reverse its policy course without causing market disruption as participants adjust their positions. 146 markets work better and are more likely to be in synch with, and reinforce the effects of, central bank policies. Market participants needed to understand how policymakers were expecting the economy to evolve and what likely policy was to be implemented. Fed Chair Ben Bernanke (2006 – 2014) instituted many changes to increase transparency and communication by publishing the dot plots in 2011 (governing council estimates of the path of future interest rates), releasing meeting notes and crucially giving detailed purchasing amounts in the QE programs during his tenure. Thus, the research on the quiet revolution in central banking remains relevant for the ECB until today. 6.3 European Central Bank communication As elaborated above, the focus of information exchange in this research is on active market participants in the asset management industry. The ECB holds regular hearings with the European parliament, particularly with the European Committee for Economic and Monetary Affairs. We hereby focus on speeches, face-to-face meetings and transcripts as part of the monthly briefings on policy decisions and the BMCG. Investment research reports by investment banks are also included in this analysis. In contrast to the US Fed, the ECB started with monthly press briefings from its inception. In a sense, the quiet revolution never happened for the ECB as it started amidst the new paradigm of information sharing. The aim for the first ECB president Wim Duisenberg (1998 – 2003) was to institute a modern and transparent communication strategy from the very outset. Duisenberg’s approach could be summarised as central bank communication needing to reflect internal deliberations of the ECB. In a letter to the Chairperson of the Committee for Economic and Monetary Affairs, Duisenberg stated that “transparency requires that our communication closely reflects our internal decision-making process. Adopting ‘too simple’ a form of presentation would not honestly convey the complexity of the analysis we have to conduct” (Duisenberg, 2001). The monthly press briefings format usually starts with the President reading a prepared statement, lasting around 10 – 15 minutes, on the current policy decision and economic conditions during the prior period and gives guidance on likely policy behaviour. This is usually followed by 45 - 50 minutes of questions from journalists in the audience. The meetings are broadcasted live so that any market participant or the public can follow them. However, Duisenberg still faced an uphill struggle with communicating policy 147 and the expected paths of interest rates to the market. The main issue seemed to have been that it wasn’t “always clear when ECB officials are sending signals, and when they are simply inarticulate” (Sims & Wessel, 2000). The transparent approach to communication by no means reversed under ECB President Jean-Claude Trichet (2003-2011), and the press briefings were used to also get some feedback for policy makers in the Question and Answers section. As Trichet put it “…the value, of course, of these press conferences is that they permit us to learn, as far as the main issues are concerned, what is in the minds of the European and global press” (Quoted in Stern, 2004). The US Fed seemed to have more trouble communicating effectively with the market. As Stern pointed out “we've been struggling with [the communication topic] in the Federal Reserve off and on…we've had a lot more concern about whether we've been able to communicate our intentions effectively to the markets” (Stern, 2004). Trichet (2008) was still noting the changing trend of increased transparency in central bank communication, while under Mario Draghi’s presidency, central bank communication, in the form of forward guidance, evolved as a policy tool in itself. Unconventional monetary policy in the form of QE needed more efficient and precise information (Draghi, 2014e). In 2014, the ECB also started to publish meeting summaries. The ECB has gone to great length to define communications with the market for higher-level officials, such as GC members. They should only accept speaking engagements where market-sensitive information is divulged, if the same information is published online at the same time via webcast or on the website (European Central Bank, 2019b). This guidance presumes that it can be known a priori what information is and can be market-sensitive. More interestingly the code of conduct further elaborates on guidelines for one-on-one meetings with market participants “e.g. with bankers, industry representatives, or with special interest and advocacy groups, the members of high-level ECB bodies and alternates will ensure that no market-sensitive information is divulged” (European Central Bank, 2019b). This guideline is sensible, but given how market participants gauge information for investment decision-making, the tacit understanding of a situation derives from information gathered from diverse networks and one-on-one meetings. Thus, the level of information extracted from one-on-one meetings can differ significantly from the prepared statements of a press conference followed on live webcasts, despite equal factual content. This is the case, because, in addition to verbal communication, other cues, such as facial expressions, can add 148 to the understanding of a given situation. As an illustration, Japanese researchers are examining the facial expressions of BOJ governor Kuroda and ECB President Mario Draghi to ascertain conclusions about likely policy utilising facial recognition software (Gray, 2018; Uetake, 2017). When implementing European-style QE through its purchase programs in 2015, the ECB went ahead and defined, specifically, what sort of information could be exchanged with counterparties at the implementation stage of the policy. The guiding principle to communication is transparency, equal and timely distribution of data and hence attempting to avoid divulging material non-public information to market participants. This is in line with the US Fed’s history of central bank communication as discussed in section 6.2. Looking back at the period of 2012 to 2014, the ECB introduced possible unconventional monetary policy measures as optionality into the expert networks without explicitly launching and implementing a new policy. This will be discussed below. 6.3.1 Communication during the Mario Draghi Era (2011 – 2019) In 2012, ECB communications with markets experienced a sea-change with Mario Draghi’s ‘bumblebee speech’. While the analogy of likening the Euro to a bumblebee was at a minimum ambivalent, the quote “within our mandate, the ECB is ready to do whatever it takes to preserve the euro. And believe me, it will be enough” (Draghi, 2012) was unequivocal. It meant that there was a new, proactive ECB president that would use the modern central banking toolbox to its fullest. Since that speech, the nick name ‘Super Mario’ evolved and found enthusiastic acceptance in the investment communities. The belief in Draghi’s leadership to steer Europe out of the crisis grew, particularly coming out of a muddled multi-lateral decision-making history through the 2010 Greek crisis period and cliff hanger decisions out of Brussels involving the Troika (European Commission, ECB and the International Monetary Fund), German Chancellor Angela Merkel and French President Nicholas Sarkozy. Draghi’s 2012 speech de facto launched the Open Monetary Transactions (OMT) programme specifically, but more generally Fed-style QE in the eurozone, thereby sending stock markets higher and eurozone bond yields lower. Former Governor of the Central Bank of Ireland, Patrick Honohan, described the context to this speech that people at the time “painted scenarios in which Greece might exit the euro by creating its own currency and defaulting on euro-denominated debt. By making exit conceivable, such a step might lead 149 to irresistible speculative pressure on Portugal” (Honohan, 2019, p. 93). ‘Grexit’, Greece leaving the Euro, could have caused significant risk of a domino effect of other countries leaving the euro and a collapse of the system. During the ensuing period, previous ECB Deputy Director of General Research REDACTION: Name removed for confidentiality reasons joined Goldman Sachs as Chief European Economist.69 REDACTION: Name removed for confidentiality reasons reception by market participants was particularly welcoming and depicted a newfound excitement for European rates and fixed income space. In his first client calls he would lay out the current thinking at the ECB and explain the paradigmatic shift under Mario Draghi. REDACTION: Name removed for confidentiality reasons, himself a former academic in the US, was very familiar with QE and large scale asset purchases. The ability to translate new concepts to market practitioners was significant as the view of the ECB prior to Draghi was one of inaction and having to work powerlessly on the basis of consensus across the eurozone leaders, the Commission and Parliament for previous bail out programs. The demand for ECB experts in the asset management industry rose given the need to understand and contextualise the bank’s policies under the new presidency. Changing demand for expert knowledge also transferred to other asset classes outside of fixed income such as equities, that had traditionally not paid too much attention to central bank communication. In 2011, the previous Morgan Stanley Fixed Income Strategist REDACTION: Name removed for confidentiality reasons became head of trading strategies at Citi Group in Hong Kong. This was a newly created role for the purposes of servicing clients with a more cross asset class approach and shorter term trading recommendations. Needless to say, these trade ideas had to be integrated into the context of the firm’s trading recommendations of the specialist Equity and Fixed Income/FX teams. However, it was a reflection of central bank monetary policy, given that during European and Japanese QE, there had been many disagreements between Citi’s traditional equity strategists and REDACTION: Name removed for confidentiality reasons in this new role. Citi’s equity strategy team around REDACTION: Name removed for confidentiality reasons was an ironclad fundamental and valuation-focused investment team. As the equity strategists 69 REDACTION: Name removed. Personal data removed for confidentiality reasons 150 continued to focus on valuations, profitability and fundamentals for their asset allocation recommendations, REDACTION: Name removed for confidentiality reasons introduced new models to predict market levels based on correlations with central bank balance sheets, notably for the Japanese Equity Index Nikkei at that time. Given his mathematics background, his supply-demand model of financial markets gained significant interest in the investment community and propelled fundamental equity investors to start paying attention to global central bank communications. Robert Buckland, long-time head of equity strategy at Citi, put this aptly in a note at the height of ECB QE: However we have always found it odd that…economists and central bankers have shown little interest in the part that equity markets play in shaping the corporate response to low rates. Our fixed income and credit colleagues are always telling us about their invitations to meet with central bankers. As far as we know, none of Citi’s equity strategists have ever been asked to join them (Buckland et al., 2016, p. 3). Buckland’s colleague in New York, Matt King, Global Head of Credit strategy at Citi, echoed this: And what’s more, when we do, actually almost embarrassingly simple things, like just plotting what the global central banks are buying each month and plotting that against the change in equity prices or the change in credit spreads, we come out with these really, really good relationships without looking at any fundamentals whatsoever (Alloway & King, 2017). This rising industry focus on fixed income coincided with the establishment of the ECB’s BMCG in 2013, which will be discussed in the next section. 6.3.2 The ECB’s Bond Market Contact Group and Expert Networks The interaction and communication with market participants at the ECB was intensified with the establishment of the different contact groups. These groups are expert networks of systemically important market participants and industry experts. Usually, annual membership consists of two to three ECB members and around 20 members from AMCs, investment banks, industry bodies and regulators. The most relevant group for the study of the bond 151 purchasing programs is the BMCG, established in 2013. For this group, select senior individuals at large AMCs and some investment banks are annually selected to meet on a quarterly basis with ECB officials, to exchange views, collaborate on research and contextualise market events (European Central Bank, 2018). Membership criteria revolve around the importance of the institution the individual represents, as well as the knowledge and expertise of the individual. The individuals are expected to represent not only the firm’s views but act as representatives of the market as a whole and make active contributions to the discussion and body of knowledge generated in the group. Continued membership is additionally contingent upon attendance, active participation, contribution to the discussion and preparation of presentation material. Presentations are often jointly prepared by members with different company affiliations and cover topics chosen mostly by the ECB. These collaborations create strong ties as e.g. individuals would meet in Frankfurt or London in their spare time, prior to the ECB meeting in Frankfurt, to discuss market developments, agree on specific investment views or technical analysis of monetary policy and are thereby creating a significant input for the ECB itself. The mandate of the BMCG was established during the first meeting: The two main aims of the BMCG are: (1) to improve the ECB’s market intelligence on structural and on-going market issues in the euro bond markets; and (2) to provide market participants with an opportunity to bring up issues that, in their view, deserve attention at euro area level (European Central Bank, 2013). In that sense, the meetings serve both for establishing and shaping the consensus of the group. Meetings serve to make clear what the ECB intends to do and ascertain participants’ investment strategies given a certain set of scenarios. According to contacts familiar with the group, the benefits for members participating are reputational, the ability to network, being close to the main information source and being aware of what the ECB thinks. Implicitly, by presenting their views, members of the asset management industry also give guidance on their respective evaluation of market situations. Hence it is an intense two way communication exercise in which the ECB can gauge market interest, perceptions and trends. The then Executive Board member Coeure, known for his good standing with the asset management industry, concluded the first session in 2013 by saying that “the BMCG would complement the other sources of market intelligence on bond market area and contribute to 152 the dialogue between policy makers and market participants” (European Central Bank, 2013, p. 3). The BMCG had 213 members and participants from 2013 to 2019 and I recorded the names, institutional affiliations, job titles and domiciles for these individual nodes in a social network from the ECB website. The ties between individuals represent the shared membership during a particular year from 2013 to 2019. The resultant interlocking network yields a network structure focused on the member domiciles. Figure 6.1 shows that most members and participants in the BMCG meetings have been from either Frankfurt (25%) or London (39%) over the past 6 years. Paris (9%) and Munich (8%) are the next significant member domiciles. The size of the city nodes is a measure of the number of years an individual from this city has been a member in the BMCG measured as degrees. The sociological literature highlighting the importance of regular physical, face-to-face meetings in establishing a consensus is of relevance here. The BMCG members in London and Frankfurt will have more chance encounters and connections become more intense as conferences and meetings are shared in respective cities. Social networks become less coherent in an age of global travel and rise of far flung connections (Larsen, Axhausen, & Urry, 2006), so that bonds between those that have more frequent chance-encounters based on location may forge stronger ties and relations become thicker and intermeshed. As Urry puts it for another context, face-to-face conversations are “produced, topics can come and go, misunderstandings can be quickly corrected; commitment and sincerity can be directly assessed. Trust between people is thus something that gets worked at involving a joint performance by those in such conversations” (Urry, 2003, p. 164). 153 Figure 6.1 BMCG membership affiliation network (2013 – 2019) by city The most central individuals can be found in figure 6.2. Zoeb Sachee, Head of Euro Area Government bond trading for Citi Group based out of London, and Michael Krautzberger, CIO of Fixed Income for Blackrock in London, have both been members since the establishment of the BMCG in 2013 and have outlasted and participated more than all internal ECB staff over the past 6 years. While Sachee has been with Citi for 29 years, mostly in the same role, Krautzberger has been with Blackrock since 2004, regularly winning Morningstar’s Fixed Income awards and was serving at Union Investment and DWS previously (individuals from both institutions have been continuous BMCG members). These two individuals alongside Christoph Rieger of Commerzbank and Ingo Mainert of Allianz make up the four most central members of the BMCG. Given the nature of increasing job mobility and staff turnover in the finance industry, having strong ties with members that have a lot of weak ties, enables information and views to travel quickly through different networks (Granovetter, 1977). Ingo Mainert, for instance, is also affiliated with the European Fund & Asset Management Association (EFAMA) representing 23 member countries and its asset managers in the BMCG. As CIO, one naturally has lots of weak ties, particularly given the oversight responsibilities that come with the position and the maintenance of external stakeholder relations. Figure 6.2 shows the different groupings of BMCG members to date. As head of European sovereign bond trading, Zoeb Sachee is in contact with both the ECB in these meetings, as well as his buy-side clients. These individuals have plenty of weak ties as a result of their function and seniority. Carlos Egea, 154 another central figure in the network, led the initial discussion on the scenario analysis of a potential sovereign bond purchase program in the April 2014 BMCG meeting (detailed in the next section). While originally working for the ECB itself as an economist, he moved on to Morgan Stanley as Macro Trading Strategist for 10 years and then transitioned to a prominent hedge fund as strategist. European strategists contextualise the ECB policies and their likely impact on markets. Other peripheral group members were invited just for a single year and should have less influence on the shaping of the agenda and thus group consensus as a result. It is also interesting to note that only 10% of memberships over the 6 years were occupied by female individuals. Figure 6.2 Social Membership Network of the BMCG (2013 – 2019) Similar to the corporate interlocking networks of chapters four and five, the institutions affiliated with the BMCG membership are also the most central in this social network in Figure 6.3. Blackrock and Allianz are the most central asset managers, while Morgan Stanley is the most prominent investment bank represented in the network. Hedge funds (Moore 155 Capital, Spinnaker Capital), industry bodies and political institutions (European Stability Mechanism, Eurex Clearing, European Commission Economic and Financial Committee) are at the periphery of the network. For instance, the European Commission’s Economic and Financial Committee representatives Anne Leclerq and Maya Atig were only invited around the time of the implementation of the PSPP in April 2015, as the government debt purchase program faced increasing scrutiny and political opposition, as seen in the European Parliament Hearing, 23 March 2015. Members of this committee were not re-invited thereafter. In his research on interlocking directorates, Useem (1980, p. 66) found that corporate interlocking ties on the boards of the late 1970s were barely replaced when board seats were vacated by death. After Carl Norrey retired in 2016, JP Morgan practically lost the seat at the BMCG and only had one representative member in 2017 and none in 2016, 2018 and 2019. After his departure, a client remarked on Norrey that Carl has a remarkable ability to read markets and so I often turned to him in order to better understand what is going on. His execution skills were perfect, and he was just fun to work with. As a leading figure he contributed a great deal to the development of the SSA [Sovereigns, Supranational and Agencies] market (McGlashan & Fildes, 2016). A rival banker with a head of capital markets position also mentioned that: For what it’s worth, I’d rank him as the top fixed income DCM [Debt Capital Markets] and trading professional in the market over the last 25 years. He’s been a huge asset to JP Morgan, and frankly a huge asset to the wider industry, and we — banks, issuers, investors — have all benefited from him being around these past decades (McGlashan & Fildes, 2016). Although a firm’s size and centrality does seem to matter for AMCs, JP Morgan looks to have lost its influence in the BMCG with the retirement of Norrey and this example indicates that the individual expert may have more influence in this network than the individual’s employer, particularly if the institution is an investment bank. 156 Besides Norrey’s JP Morgan, two US institutions, Morgan Stanley and Citi Group are the most central investment banks in the BMCG interlocking network which communicate directly with different AMC clients on a daily basis. This is also a reflection of US investment banks taking over European bond trading after the eurozone crisis of the 2010s.70 For instance, Allianz, Blackrock and Commerzbank will deal with these two brokers on a daily basis in trading activities. Also, more peripheral asset managers in this network such as APG Asset Management, Norges Bank, Union Investment or Deutsche Asset Management will deal with these counterparties on a daily basis. Hence there are other social ties between the elites within the network of the BMCG and this helps strengthen the ties within the BMCG itself. Simply put, the social ties between members extend the BMCG network. ECB experts transcend the institutional boundaries and share a common history with other peers at this elite level. Hence, central actors such as Krautzberger or Sachee, become important actors over the history of the BMCG, detached from their institutional affiliation. In his earlier work on interlocking directorates, Burt contended that central directors with multiple board seats make up the “inner group within an American capital class” (Burt, 1983, p. 259). With the ongoing decline of interlocking board seats (Chu & Davis, 2016), other affiliation networks in elite circles look to take precedence over board seats, such as in the example of the BMCG. For an investment bank, a key criterion of its market standing would be the number of ties established with other market participants, while for a pure investor, an AMC, a key measurement of importance would be the amount of assets managed. Krautzberger represents Blackrock as the single largest investor in European bond markets and has strong policy views pertaining to the eurozone and on how to trade these. In the run up to the PSPP, he voiced his views in the following way: That is basically the whole purpose of the exercise, that the ECB is really trying to make the safe haven unattractive...we are not going out and just saying buy all the periphery, even within the periphery we prefer more the Western end of the periphery and maybe be a bit more careful on Italy (Krautzberger, 2014). 70 See a Bloomberg analysis on changing market share trends in European bond trading https://www.bloomberg.com/news/articles/2019-02-25/jpmorgan-s-traders-nab-market-share-while-deutsche- bank-s-slip [accessed 03.04.2020]. 157 His established strategy later in 2015 after the PSPP announcement developed into a country allocation guided by ECB liquidity. In a conversation with Jonathan Ferro, he pointed towards this. Liquidity is obviously part of the consideration and that is a big advantage of Italy, but if you trade those markets on a relative value basis, it’s even good, if sometimes, [for] your short leg, for example long Spain, short Italy, we like at the moment… the liquidity in Italy is simply good to include them in trading strategies (Krautzberger, 2015) The influential Rieger of Commerzbank,71 also expresses strong policy views with respect to the eurozone. In 2013, he already spoke about ways to inject liquidity into the corporate sector: “The ECB could consider a liquidity facility geared towards corporate loans… modelled on the UK’s FLS [Funding for Lending Scheme]…[it] could offer three year liquidity at a fixed rate, with the condition it is used for these ABS” (Quoted in Hayes, 2013). By appointing such individuals, the ECB is both able to bring together discrepant views of highly connected and influential market actors but also establish a consensus by delegating co-presentations and research efforts in regular face-to-face meetings. As mentioned in chapter one, the discussion above shows how the ECB co-opts BMCG members in highly controlled and animated member-only meetings, with the ability to diffuse ideas and create consensus. This will be analysed in detail in the next section. 71 Rieger is regularly voted as ‘Zinsexperte des Jahres’ (interest rate expert of the year) by €uro magazine. 158 Figure 6.3 Institutional affiliation in the BMCG membership network (2013 – 2019) 6.3.3 Diffusion of the Purchase Programs in the BMCG network As outlined in chapter one, a central question of the corporate interlocking research in organisation studies and sociology attempts to address the success of co-optation practices of board members. Cohen, Frazzini and Malloy (2012) examined whether co-optation of board members yields a positive outcome from the corporate perspective. This section portrays the ECB’s co-optation of BMCG members and aims to evaluate the outcomes of these practices in the example of the diffusion of the PSPP. To establish the context of information exchange on the PSPP requires analysing official transcripts back in time, roughly a year before its implementation in March 2015. In the April 2014 ECB press conference, ECB president Draghi gave the following answer to a question on details of the unconventional monetary policy option included in the GC’s statement of that month: 159 I think you have rightly pointed to the key sentence in the statement: ‘The Governing Council is unanimous in its commitment to using also unconventional instruments …’ – meaning that we haven’t finished with our conventional measures – ‘... also unconventional instruments within its mandate in order to cope effectively with risks of a too prolonged period of low inflation.’ This statement says that all instruments that fall within the mandate, including QE, are intended to be part of this statement. During the discussion we had today, there was indeed a discussion of QE. It was not neglected in the course of what was actually a very rich and ample discussion (Draghi, 2014a). This in effect introduced the possibility of government bond purchases into the information exchanges. Deliberations by ECB experts give interesting clues. In a May 2014 report, Goldman Sachs’ team evaluated the muted market reaction to Draghi’s April statement and the lack of an ‘announcement effect’ (such as in the ‘whatever it takes’ episode) in that investors were uncertain how to interpret Draghi’s ‘unanimous commitment’ statement (Pill, Daly, Schuhmacher, Benito, Holboell Nielsen, et al., 2014). The interjection of this statement in the ECB briefing also led to the diffusion of a new policy idea, that of sovereign bond purchases. Subsequently, in the August 2014 press conference, a journalist asked an indirect question on what QE constitutes. Draghi gave the following answer on whether he would call purchases of ABS a European QE: Well, no. It depends very much on what we define by QE. If we define QE as broad- based asset purchases, then QE would include ABS, but would certainly not be reduced to ABS only. The QE broad asset purchases programmes include government bonds, in general public assets, and private assets. So ABS would be an example of private assets, but then you have QE into government bonds that are still on the table (Draghi, 2014c). The idea of government bond buying continued to emerge in the ECB’s November 2014 briefing: 160 Looking ahead, and taking into account new information and analysis, the Governing Council will closely monitor and continuously assess the appropriateness of its monetary policy stance. Should it become necessary to further address risks of too prolonged a period of low inflation, the Governing Council is unanimous in its commitment to using additional unconventional instruments within its mandate. The Governing Council has tasked ECB staff and the relevant Eurosystem committees with ensuring the timely preparation of further measures to be implemented, if needed (Draghi, 2014d). Hence this statement alone was probably the clearest hint that a large scale government bond purchase program would be launched. While Goldman Sachs’ team was still sceptical about a sovereign QE throughout 2014, the tune changed in a note from November 27. It was concluded, while still unlikely, that if conducted in accordance with the capital key, a potential sovereign bond purchase program could amount to around EUR 500bn in the first instance (Pill, Daly, Schuhmacher, Benito, Durre, et al., 2014). In the ECB’s December 2014 press briefing, Mario Draghi was asked about the program again to which he replied: On the first question, you see, you are in a very intelligent way trying to extract from me the date of the next decisions, and you won’t get it. Early means early. It doesn’t mean at the next meeting. It depends very much on how our assessment will go. We’ll certainly have a lot of facts to examine, namely and especially the big movement in the price of oil and what the impact of this is going to be, not only on economic activity, but also on inflation (Draghi, 2014b). ‘The date of the next decisions’ hereby implied the launch of a sovereign bond purchase program. In a 11 January 2015 Goldman research report, the ECB was expected to launch Sovereign QE on January 22, citing deteriorating economic factors and comments from Draghi’s December conference. It would be a response to reduce the risk of deflation (Gazarelli, Ardagna, & Cena, 2015). A base case of EUR 500bn in government bond purchases was formulated and what investors would not have thought possible only a few months ago would, through tighter spreads in the eurozone, ease financial conditions significantly (Gazarelli et al., 2015). The PSPP was announced on January 22, 2015. 161 On the 23rd of March 2015 in a European Parliamentary hearing by the Committee for Economic and Monetary affairs, Draghi mentioned that On 9 March, we started purchasing public-sector securities as part of our expanded asset purchase programme...The pace of purchases so far puts the overall program on track to reach a total of €60 billion in March. At this point in time we see no signs that there will not be enough bonds for us to purchase. Feedback from market participants [emphasis added] so far suggests that implementation has been very smooth and that market liquidity remains ample (Draghi, 2015a). Hence these prepared remarks depended on feedback gathered from the ECB’s contacts in expert networks, be it trading counterparties or BMCG members. While the question of government bond buying started to emerge in the monthly speeches and during the course of 2014, the topic also transpired in ECB meetings with senior experts in the BMCG. In practice the phrase “unanimous commitment to using also unconventional instruments” removed the policy from the private realm of the ECB into the discourse of the investment community. The time lag from the diffusion of the idea to implementation was roughly one year. During that time, the ECB would meet with key decision makers in the asset management industry to gauge their expectations and understanding and take this into consideration when shaping the implementation of the policy. During the BMCG meeting on the 27th of January 2015, both the CIO of AXA and the Head of Research for Commerzbank jointly analysed the market impact of the announced policy of the 22nd of January. Despite the fact that the meeting was held only five days after the announcement, the discussions show that participants were familiar with the program and analysed different scenarios in terms of monetary size and potential impact on market structures and liquidity. Rieger for instance made the point that “if the ECB were to buy €600bn (€50bn per month for one year) in sovereigns it would end up holding some 10% of the outstanding euro area government bond market” and goes on to state that “a negative impact on bond market liquidity (measured by turnover) could not be found (see BMCG presentations by Carlos Egea and Campbell Gilbert at April 2014 meeting)” (Rieger, 2015, p. 6). Earlier in 2014, BMCG members Egea and Gilbert already weighed different scenarios and cost/benefits of a potential government bond purchase program, thereby contributing to the sensemaking processes of the ECB (Egea & Gilbert, 2014, p. 3), as well as providing specific views on the monetary size of the policy. 162 A week after the PSPP announcement, the BMCG held an ad-hoc conference call with all BMCG members to gauge market reception of the policy. It was concluded that despite the ECB’s attempts to diffuse the policy throughout 2014, members were surprised at the extent of the market reaction to its announcement. Indeed, the summary of the March Teleconference states that members were particularly surprised about the decline in Euro area long-term government bond yields beyond a sole impact of the PSPP purchases, indicating that other market participants pre-empted the proposed central bank purchases. During the call, members “attributed the move in part to speculation about reports on the potential implications that a further decline in short-term yields could have for the PSPP-eligible universe [emphasis added] in some jurisdictions” (European Central Bank, 2015a, p. 1). The policy may have not been unexpected to BMCG members but looks to have been so to the market as a whole given the outsized market reaction. While there is no specific material non-public information, the BMCG expert network established certain scenarios of future policies, for instance, the previous ABS program was analysed and potential long term asset purchase programs were discussed before the new policy announcement (Bond Market Contact Group, 2015). Using prior examples of the BOE’s QE and the higher levels of the ECB’s prior Securities Market Program (SMP), Goldman Sachs’ Schuhmacher contended that the likely effect of Sovereign QE (the PSPP) would be muted as rates for sovereigns were already low (Pill, Daly, Schuhmacher, Benito, Holboell Nielsen, et al., 2014). A Morgan Stanley report from November 2015, recalled the market reaction to the PSPP announcement in detail. O’Brien and his team concluded: When QE purchases first started in March 2015, investors reacted by preferring sovereigns that had the largest disparity between QE demand and government supply as well as where ECB holdings would be the highest as a percentage of issuance. This favoured the smaller nations of the Netherlands, Belgium and Portugal over the heavier issuers such as France and Italy (Hornbach, O’Brien, Heese, Kawano, & Liang, 2015, p. 29). This shows that equipped with detailed information of how much purchases amount to and where the most likely shortages of supply will happen, investors will aim to front run the ECB’s action and extract profit. They thereby confirm what Volcker was worried about, namely that communication would hinder operational flexibility of the central bank (see section 6.2). It also illustrates the value that individuals can extract from information 163 exchanges in expert networks, filling Burt’s (1992) structural holes between the ECB and other market participants. BMCG meetings are a form of sensemaking processes to both establish consensus as well as to gather information on the part of the ECB and BMCG members. Market situations and contexts are analysed and monetary targets of policies are discussed in scenario analyses, resulting in detailed accounts. The BMCG member selection and research efforts are coordinated by the ECB as well as the regularity and dates of the meetings. Members need to travel to Frankfurt for face-to-face gatherings at ECB facilities. The gatherings are also intense in terms of informational content and engagement in that members need to prepare presentations and collaborate on research efforts and scenario analyses prior to the meeting. One of our contacts mentioned that membership requires significant input and work. Maitlis (2005, p. 35) contends that such guided sensemaking results in rich contextual accounts bringing together diverse stakeholders and establishing focus on distinct courses of action. In other words, consensus in a group of elite decision makers can be established. The BMCG meets on a quarterly basis and also hosts ad-hoc conference calls should the need arise. The UK’s Brexit referendum and the announcement of the PSPP were incidents that resulted in an ad-hoc conference call between the BMCG members. In those calls, the main discussion was led by the ECB representatives, outlining what they are planning to do, how they see things developing and what the ECB’s view on Brexit and its impact on the economy and financial markets would be. So while the regular meetings seem to be structured by the ECB to ascertain and shape a consensus among members on pre-selected topics, delegate research presentations and collaborations among members, the ad-hoc calls seem to be more devoted to a communication of the ECB’s intended actions, contextualising global macro events and to get instant feedback on the ECB’s policy plans. Thus, the Brexit teleconference seemed to pre-empt volatility to market events and to calm BMCG members. Likewise, the ad-hoc PSPP related call seemed to explore what members thought of the policy, where two members were tasked to analyse scenarios thus giving a valuable input from the view of investors into the technical aspect of the program’s implementation. From the discussion above, it also transpired that a consensus of €500 – 600bn size of the PSPP had been established, thus for the ECB to triangulate its own policy decision, which later translated into €720bn (annually) including pre-existing asset purchases. 164 Given that the ad hoc conference calls happened when there was either market turmoil or significant policy announcements and implementation, the periods were marked by uncertainty and hence information ambiguity. White (1981) contends that complex or ambiguous information can lead to market participants imitating each other or adjusting behaviour in relation to peers. The ECB is an actor in this expert network that seeks to ascertain which purchase amounts may be satisfactory to significant market participants and adjusts its policy according to consensus of said network. To make a bigger impact, it exceeded the expected monetary target of the program. This leads the discussion on to regulation of information sharing in ECB networks. The official communication guidelines after the announcement of the PSPP are far reaching in detail and specify in what ways counterparties are allowed to tell other clients about the ECB’s action: For the PSPP, counterparties could indeed communicate to third parties that the Eurosystem had been buying in a certain market and maturity bucket. However, our counterparties shall not disclose the amounts transacted, the individual securities involved, or which Eurosystem member was buying. Particular care has to be taken for issuers with relatively few outstanding issues, where substantially broader maturity ranges have to be chosen (European Central Bank, 2019c). For the later CSPP, the wording is slightly different. Counterparties can “communicate that the Eurosystem has been buying in the corporate space, the maturity bucket and the sector (e.g. utilities), but not the exact amounts, the issuers of the bonds purchased, the securities involved or the Eurosystem member involved” (European Central Bank, 2019a). These guidelines explicitly acknowledge the importance of trading counterparties in policy implementation and the knowledge such interactions with the ECB bring. When asked about transparency of the later CSPP transactions by Member of the European Parliament Ramon Tremosa, Draghi responded that “..we are not going to disclose the guidelines or the volumes because that would simply foster activity by market participants, which could actually hamper the achievement of our objectives” (Draghi, 2017, p. 9). Thus while not explicitly formulated, information exchange that could give certain market participants advantages to 165 front-run the ECB’s policy is a key concern and as discussed above difficult for the ECB to control. Another prominent example of this dilemma occurred only a few months after the PSPP was announced. Member of the ECB’s Executive Board Coeure’s participation in the invitation- only dinner with Hedge Fund PMs caused significant stir in both financial markets and the press, given accusations of divulging material non-public information with significant market participants (M. Jones & O’Donnell, 2015; Spence, 2015). The event was sponsored by Brevan Howard, one of the largest Macro Hedge Funds at that time. In a response to a letter from the European Ombudsman Emily O’Reilly, the ECB referred to this incident as an ‘internal procedural error’. During the dinner, Coeure mentioned that: We are ... aware of seasonal patterns in fixed-income market activity with the traditional holiday period from mid-July to August characterised by notably lower market liquidity. The Eurosystem is taking this into account in the implementation of its expanded asset purchase programme by moderately frontloading its purchase activity [emphasis added] in May and June ... The slightly higher purchase volume that market analysts may observe in the coming weeks is therefore unrelated to the recent episode of market volatility (European Central Bank, 2015c, p. 2). It was contended that market participants during that dinner leaked this information or traded on said information (Spence, 2015). The discussion has shown that the ECB is not acting in a policy vacuum and both engages with and extracts information from key market actors. By co-opting members in the BMCG and thereby filling the structural holes between the ECB and financial market participants, the ECB attempts to establish consensus. Based on my field interviews, the BMCG meetings are a two-way conversation and require significant input and preparatory work by members. Thus, rather than dictating what the market is supposed to do, the ECB takes significant market actors’ expectations, intentions and potential courses of actions into consideration, particularly, when considering both the technical implementation and the size of the purchases. As a result, it becomes striking how, through market-based monetary policy, the ECB becomes entangled in financial networks of active market participants. The discussion around the diffusion of the policy idea of the PSPP also exemplifies that if market participants analyse and follow the ECB closely, new policy implementation can be fathomed 166 early and taken into consideration when portfolio management decisions are made. The interview data in chapter seven will look at the respondents' recollections and interpretations of the ECB’s communication around new policy implementation periods. Coming to the end of Draghi’s term and also the end of net asset purchases for the PSPP and CSPP, former Goldman Sachs Economist Dirk Schuhmacher put it this way when asked whether a new leadership at the ECB would mean significant change to its policies: I don’t think so. In the end it is a consensus driven body. So, whoever will be at the steering wheel at the ECB will have to find a broad consensus for whatever course she or he will want to do. So, I don’t think it matters that much. Of course, Draghi at the right point in time stepped forward and was very proactive in safeguarding the Euro with his famous speech in London, and you can question whether others would have done the same, so it does matter, but in the end when it comes to normalisation, these kind of more technical questions, I don’t think it matters that much (Schuhmacher, 2019). This brings the discussion back to Axilrod’s contention, that a great central bank leader has to primarily gauge the social and political limits to new policies and successfully manage the implementation of new policies. 6.4 Information exchange in asset manager networks In 2012, PMs were still able to share Bloomberg group chatrooms with other investors or brokers. That was one of the ways, PMs could quickly gauge expectations of different bid levels for new issuance in the eurozone which inadvertently harmonised decision-making processes between different asset managers. After the GFC and massive bank bail outs, investment banks were facing increasing scrutiny by regulators. In 2012 UBS faced fines for illicit trading on its London swap desk (Miediema & Wilkes, 2011). Collusion in the pricing of the London Interbank Offered Rate (LIBOR) market making caused fines for large investment banks such as Barclays (Treanor, 2012). The implementation of the Volcker rule in 2014 (through the Dodd-Frank reform) meant an end to proprietary trading for investment 167 banks.72 With the progression of Markets in Financial Instruments Directive II (MiFID II) and tighter internal compliance, Bloomberg chatrooms were coming under increasing scrutiny and required both approval by internal compliance teams, and were also monitored. Hedge funds were under scrutiny as well with the FBI and Securities and Exchange Commission investigating the largest hedge funds in the US for potential insider trading. In this new regime, brokers were not able to text any personal messages to clients on their work phone. How can expert communities survive when regulatory restrictions aim to break these communication networks, both between expert PMs and investment bank market makers, but also between PMs and industry experts (such as technology or medical experts)? The ECB created expert networks of PMs, traders and strategists in their contact groups. As shown, the previous accounts of central bank communication emphasised the concept of a sender and receiver. Burt’s (2010) stresses the non-neutrality of language and the tacit knowledge and jargon that develops in expert groups. PMs operate in such networks in which they share knowledge, information, views and contexts to specific market situations under their coverage. While regulation had the implicit aim of breaking up these relationships to create a level-playing field for market participants, there are still shared formulaic interpretations spread through investor conferences, professional bodies such as the CFA institute and sensemaking narratives through the use of Bloomberg terminals, the Financial Times and more unstructured data exchanges in social networking platforms (e.g. Twitter). Expert networks incorporate such shared knowledge and historical conceptualisation of fluid financial market situations. Communication thus becomes a reflexive information exchange, rather than factual information being sent from Actor A to B. The clearer the ECB attempts to communicate and specify its likely behaviour, the more market participants are required to second-guess other market participants’ likely views as the ECB’s view becomes better understood. Hence, the market reaction or interpretation of the ECB’s view becomes ever more important as a known unknown and determinant of market prices. 72 Details in the Federal Register “Prohibitions and Restrictions on Proprietary Trading and Certain Interests in, and Relationships With, Hedge Funds and Private Equity Funds” https://www.govinfo.gov/content/pkg/FR- 2014-01-31/pdf/2013-31511.pdf [accessed 03.04.2020]. 168 While central bankers from Volcker to Draghi concerned themselves with an equal informational footing for market participants, PMs strove to obtain an information edge in order to outperform peers. Bloomberg and Reuters were there to supply PMs with the regular financial data but data gathering happened at the fringes of what one could refer to as gathering ‘unstructured data’. Unstructured data requires different evaluative and formulaic processes from the financial data captured by financial data providers. To give context to this information exchange, illustrations have to be drawn from the global asset management industry, particularly in the US. In the early 2000s, both US and European hedge funds utilised expert networks such as provided by the Gerson Lehrman Group (GLG) to obtain specialist information.73 GLG arranged hourly meetings or conference calls for PMs or analysts to speak with industry experts in a variety of industries ranging from semiconductors to pharmaceuticals. GLG and the expert were paid a handsome fee for an hour’s conversation with a PM. Given the high cost of the calls, it was mostly hedge funds, who charge higher management fees, that were able to pay for such services. Information or data points obtained in these arranged meetings may or may not have been material or confidential information, but add pieces to a ‘mosaic’ of financial and unstructured information supporting or disproving the PM’s investment thesis. GLG made each expert sign an agreement not to disclose material non-public information to the client (Kolhatkar, 2017), yet often it is difficult for an expert to discern which information may or may not be material and market sensitive. Other hedge funds specialised in obtaining an information edge by speaking to firm insiders to obtain potentially valuable information or to directly hire industry experts as investment professionals.74 Regulators quickly aimed to level the playing field again, as seen in the example of Steve Cohen’s SAC Capital, who had to pay $1.8bn in fines for insider trading (Gapper, 2017) or Raj Rajaratnam of the Galleon Group, who had to serve an 11 year prison sentence for insider trading. During 2011 – 2014 there were significant efforts from regulators, particularly in the US, to counter excessive profits made from insider trading. The motivation for more regulation, however, was not based on principles of sharing non-public information, but was assessed on whether market participants made profits on these actions. Kolhatkar’s (2017) account of the ‘black edge’ – an illegally obtained information edge for purposes of insider trading – details this period in 73 GLG was founded in 1998: https://glg.it/gerson-lehrman-group/ [accessed 03.04.2020]. 74 REDACTION: Removed for confidentiality. 169 information exchange. At the time there was political will to prosecute the hedge fund industry and the Securities and Exchange Commission and the US Attorney for the Southern District of New York Preet Bharara were pulling on the same string with the aim of convicting Steve Cohen. Kolhatkar’s (2017) analysis showed that prosecutors failed to prove that Cohen read an email containing material non-public information and that this was the data point causing him to short-sell Dell Computer shares. As a result, Cohen was not indicted and the case was settled. Even when material non-public information is acted upon, it is not always clear what investment action the data points warrants. In the example of Cohen, he was aware of the quarterly gross margin for Dell, constituting material non-public information (Securities and Exchange Commission, 2013, pp. 12–16). However, this data point needed to be contextualised in market participants’ likely interpretations and consequent investment decision, which is often inconclusive. For instance, was the ECB announcement of the €720bn in the example of the PSPP higher than expected and if so which asset classes would be affected in what ways? Hence, even if PM A receives material non-public information from information source B, it does not imply an unequivocal action. These vignettes illustrate the same problem that regulators face globally. It was not clear whether participants in the Brevan Howard dinner benefitted from ECB’s Coeure’s comments elaborated in the previous section. Starbuck and Miliken stated that “people may operate very effectively even though they characterize a shared situation quite differently, and people’s unique backgrounds may reveal to them distinct, but nevertheless accurate and valid, aspects of a complex reality” (Starbuck & Milliken, 1988, p. 29). With the spread of social networking sites, experts share more unstructured data in information networks. Even in the political sphere, leaks of confidential information have become more common place. Hence from a regulator standpoint, it becomes increasingly difficult to link a single data point to a single investment decision. The information edge of ECB experts could be termed what Burt (2007, p. 122) outlined to be ‘information arbitrage’. Those who are able to enter separate expert networks can obtain valuable information that others in the network may not be able to get. They also obtain a tacit understanding (Barker et al., 2012) of fluid market developments as discussed in chapter two. Particularly in episodes of higher uncertainty, unstructured information, tacit 170 understanding of a given context becomes more valuable. This also enables ECB experts to partake in specialist meetings and increases their ability to remain of value to a diversity of market actors at both AMCs and at the ECB itself, and facilitate more informed decision- making. Another recent illustration from the asset management industry for the changing footing of information exchange in expert networks is that of Twitter and social media use. An ardent user of Twitter, Cathie Wood is also one of the few managers who runs actively managed ETFs at ARK Investments, with a notable $6.6bn assets under management. Wood pursues, what was described above in the hedge fund examples, investing in disruptive technology companies. Given that the company manages these positions in an ETF, all the positions are communicated daily, thereby greatly increasing transparency. When addressing the topic of information sharing in expert networks using social media she stated that: We’re living in the communities we are researching, because we have people who are connecting with us saying ‘hey, that’s our world, and yes you’re getting this right or you’re not getting this right and maybe you should think about this…’ so we’re getting insights from, eh, the innovators we’re researching that others cannot get because they’re not living in those communities and they’re not sharing what they’re doing (Wood, Balchunas, & Weber, 2019). While regulators have worked to even the level-playing field of information dissemination during the SAC and Galleon period described earlier, Wood goes on to state that “I think it is a huge competitive advantage and, em, I think compliance departments will have to let up, because I do believe Twitter and other social networks are going to be very important knowledge networks” (Wood et al., 2019). The growth of unstructured data and information exchange significantly changes the concept of data sources that shape investment decision- making. On the one hand, the ‘black-edge’ investment process and business model, described by Kolhaktar (2017), has been de facto terminated with the imposition of the $1.8bn fine on SAC. Yet on the other hand, the concept of non-public information has significantly changed. Single data points become less significant as richer context is openly shared in expert networks. As described in chapter two, PMs utilise a myriad of data sources and contextualise single data points within the overall investment strategy. 171 In summary information exchange in the asset management industry has evolved from the period of the prosecution of US hedge funds and the incident of the ECB’s hedge fund dinner. Currently, information is exchanged in expert networks where it is unclear for regulators, what information is material and non-public. Even when Mario Draghi recently mentioned a potential revamp of the existing QE program in April 2019, there was literally no market reaction to his narrative. This does not necessarily mean that the ECB president has lost the ability to surprise markets, but more that policies and their rationale have been widely understood by market participants. The BMCG is a good example in which the ECB retains some control in shaping consensus of said expert network, but is also influenced by views of these experts and becomes increasingly entangled in these interactions. 6.5 Conclusions This chapter has discussed in what ways the ECB spreads its messages and attitudes towards possible scenarios in the economy and financial markets. It has also shown that it is the intent of the ECB to spread its messages through different communication channels to prepare certain courses of action for market participants. Active market participants contribute to the ECB’s own interpretation of events as it looks for guidance from experts in the asset management industry and takes views into account, particularly when contemplating implementation of new policies as seen in the example of the PSPP above. However, during market episodes where there are greater unknowns, the ad-hoc conference calls from the BMCG on the Brexit vote of 2016, the ECB takes the lead in providing clear and ‘assertive’ communication. Not only do the ECB members of the BMCG organise and coordinate the group calls and the agenda, but also provide guidance on likely behaviour. With the size of the balance sheet as high as the ECB has maintained throughout Mario Draghi’s presidency, this forms an important input into decision-making processes of market participants. Thus, central bank communication and information exchange is a two way relationship. A key issue in the discussion of central bank communication and financial networks in this chapter remains the notion of ‘group think’. Communication and meetings play an important role to harmonise behaviour. With the establishment of the BMCG, the ECB is able to communicate with key market participants in great depth, its intended actions and interpretations of data, thereby creating an expert network. 172 As in the corporate interlock literature, co-optation and diffusion of the PSPP through the BMCG may not have yielded success given the outsized market reaction to the implementation in 2015. The BMCG serves as another empirical example of the influence of nomination practices (such as Cohen et al., 2012) on the overall benefit to stakeholders. Likely, given their idiosyncratic strategies, hedge funds have not been part of the BMCG membership for some time now and are peripheral in the membership network, despite the European industry amounting to around US$486 bn in assets and US$ 2.3tln globally (Eurekahedge, 2019). The BMCG membership constitution is relatively homogenous in terms of the investment products offered by the members. This may reflect the falling influence of active asset managers. However, engaging diverse market participants can act against a bandwagon effect and group think. Particularly, active asset managers, including hedge funds, are able to counteract forces of social cohesion. The BMCG network seems to be focused primarily on members in Frankfurt and London. The under 10% female participation in the 6 years of BMCG existence is also something to be aware of. If policy makers are ‘in bed’ with significant market participants and create a harmonisation of behaviour by building consensus in BMCG meetings, this only encourages group think. As Uzzi (1997, pp. 53–54) pointed out in the context of shared equity investments, the higher the level of multiplexity, i.e. the number of shared ties between two agents, the higher the investment activity within the embedded network. More diversity in the membership composition would help mitigate this group think. If positions at significant asset managers, investment banks and at the ECB are occupied by the same network of individuals, this inevitably leads to policies being shaped by the same individuals. Methods, scenario analysis and monetary size of purchasing programs are all material and contentious issues. Sensemaking in organisations was applied to the practices of the investing elite discussed above. Sensemaking is also a process in which preconceptions guide explorations of an unknown situation (Weick, 1988, p. 306). The BMCG experts consulted previous experiments to establish frameworks to the potential shape of European QE before it was launched in January 2015. Experts looked at previous experiments of the BOE and the ECB itself in its ABS purchase program to establish a detailed account and decision framework. Both BMCG experts and experts in the asset management industry then went on to deduce a potential size of the sovereign QE as elaborated above. In doing so the “act of exploring itself has an impact on what is being explored, which means that parts of what the explorer discovers retrospectively are consequences of his own making” (Weick, 1988, p. 306). These 173 modalities resemble what Abolafia and Kilduff (1988) referred to as enacted environments in which social actors construct and order information and scenarios in the process of sensemaking. The BMCG offers an empirical example of how PM’s individual portfolio actions are integrated in social processes of sensemaking in ECB expert networks. The meetings of the BMCG both serve as gatherings to make sense of uncertain environments, collaboratively analysing previous market situations and to reduce volatility of decisions by establishing a consensus. Sensemaking processes in the BMCG produce constraints, detailed assessments and courses of action. The fact that the market reaction to the implementation of the policy was a surprise is evidence that beyond the BMCG consensus, market participants adhere to their own investor frame and put into context the information in their own sensemaking processes so that the ECB remains just one input to frame construction and adaptation. This research on the ECB’s communication channels and the BMCG also echoes findings from earlier literature in economic sociology (Burt, 2010; Granovetter, 1985; Lazarsfeld & Merton, 1954) in that information exchange in social networks shapes beliefs, behaviour and systematic ways of interpreting information. Information exchanges within these groups become highly specialised and it becomes difficult to communicate this knowledge to other networks. This also means that market participants with membership in the BMCG network can utilise information arbitrage for career benefits, such as described in Seabrooke’s (2014) notion of epistemic arbitrage. The examples of Norrey and other influential members were used to show that ECB experts can transcend the importance of their institution and in what the corporate interlock literature found as the discontinuation of ties with the death of board members (Burt, 1983). ECB experts continue to be important in that they help translate jargon and concepts from the ECB networks to others. However, these experts are now not necessarily working only at brokers, in this case investment banks, but move freely from the ECB to the buy-side, hedge funds or other AMCs. From a historical perspective, Mario Draghi’s communication with market participants can be divided into the period of 2012 up until 2016 after the implementation of the PSPP and CSPP, and from 2016 until the end of the PSPP and CSPP in December 2018. The former period was marked by significant adjustments in the asset management industry to cater for and contribute resources to the analysis of ECB communication policies in which the need 174 for understanding new innovative policies grew commensurate to the full usage of the monetary policy tool kit. The role of the ECB experts in this period is akin to the concept of diffusion in economic sociology, such as that of the medical professionals in Coleman, Katz and Menzel’s (1957) research. There, the social networks of experts helped the initial diffusion of a medical innovation at an early stage but lost influence in later stages of the adoption, as the product became more widely known. The latter period (2016 until present) had shown lower volatility in capital markets as behaviours have been learnt. The next chapter explores this and the idea of arbitraging information exchange in expert networks in more detail by analysing interview data conducted with systemically important market participants. 175 7 Adaptive Frame Construction: Strategies and Tactics “Don’t look for the needle in the haystack. Just buy the haystack.” Jack Bogle “It is difficult to predict how the CSPP will impact market behaviour. On the one hand, it could help provide support for the market, reducing the likelihood of any major sell-off in the near future, and smoothing volatility. The ECB ‘back-stop’ bid could therefore make holding European investment grade non-bank corporate bonds relatively attractive, limiting any potential downside, at least in the near term. However, this could go completely the other way, particularly if the size or nature of purchases are perceived to jeopardize market functioning and liquidity. Investors may see this as an opportunity to ‘cash in’ on their eligible holdings and instead move into other parts of the credit curve or different asset classes, with the potential for higher returns. To a large extent this is possibly the intention of the CSPP, driving portfolio rebalancing into less safe corporate sectors, such as high yield or SMEs, where the need for credit and market investment is more pertinent…” ICMA IG Corporate Bond Committee, 21 April 2016 7.1 Introduction The International Capital Market Association Group (ICMA) is a not-for-profit membership association serving AMCs, investment banks, central banks and public and private sector issuers. The elaborate quote from its committee above was made shortly after the surprise announcement of the CSPP and underlines some of the implications on how the ECB’s buying behaviour could impact PMs decision-making. As Zuckerman (2012) argues in the context of the EMH, the better a policy or theory is understood and known, the less it will have the intended effect. The preceding chapter has detailed the diffusion and exchange of information in networks affected by the ECB’s monetary policy. It was highlighted that the missing link in current research is knowledge of the behavioural patterns of key actors in the asset management industry, an industry crucial to the implementation and transmission of monetary policy. It was also contended that while expectation formation is important, a sociological analysis of central bank communication is inherently linked to action and needs to examine behavioural patterns by nodes in the network. 176 In the following chapter, I explore the topics raised in this thesis with senior PMs, CIOs and traders in Frankfurt, London, Munich, Singapore and Hong Kong. Going into these meetings with a semi-structured interview approach (Creswell & Creswell, 2014), I decided to focus on a handful of issues to leave enough space for conversations to develop. The first group of topics were changes in methods and adaptative behaviour to QE. The second line of questions focused on central bank communication, expert knowledge/networks and reflecting on what information is used to make investment decisions. Relevant literature, industry reports and the holdings-based network analysis were consulted after conducting interviews and topics were built on and developed as the research progressed. Another area touched on in the interviews were the hypotheses from the network analysis on imitation, co-optation and whether monetary theory is actually interpreted as proposed by central bankers as in the case of the portfolio rebalancing effect and the EMH. As discussed in section 1.4.2, the meetings went very smoothly and my professional background fostered a more reflective dialogue than would have been possible if I had been a total outsider. Hence, on the one hand, I was an insider, somebody fluent in the assessment and usage of trading strategies and market behaviour with a shared history, but on the other hand, I was a specialist in a different region (emerging markets) and asset class (equities), allowing me to ask open and seemingly naive questions. The information collected was thus rich and grounded in both time and geography. This chapter is structured as follows. Section 7.2 explores how PMs in different asset classes reacted, perceived and adapted to the ECB’s PSPP and CSPP and how behaviour changed as a result. Section 7.3 explores how PMs perceive communication by the ECB and how this is integrated into investment decision-making and practices. Section 7.4 analyses sensemaking in the aftermath of the programs and looks at whether PMs thought the policy was successful. 7.2 Strategic and tactical adaptation Investment action can either be strategic or tactical, so either action with a longer or shorter term time horizon respectively. The following data from my fieldwork show how the ECB changed the context to PMs’ investment decisions. While participants still adhered to the modalities of their investor frame and process, the ECB continues to distort the conditions under which decisions are made. 177 7.2.1 European Corporate Credit and the CSPP Bottom-up fundamental investors across bonds and equities share a focus on idiosyncratic risk and company fundamentals. Absolute and relative valuations and security selection constitute the main concern for both corporate bond PMs and Equity PMs with a stock picking focus. Fundamental analysis of bonds entails analysing cash flow statements, assessing the ability of a company to repay its debt and use comparative ratios. These inputs act as a filter to bracket a pool of securities for possible selection. Then PMs would have to decide on the duration, i.e. how many years until the expiration of a given bond. The longer the time until maturity the higher the duration and the higher the duration risk. Duration risk is the risk that the market price of a bond moves disproportionately, should there be a change in interest rates. The Net Present Value of a bond with higher duration is more susceptible to changes in interest rates (given the additional return investors need to be compensated for holding a bond longer), as it still has more future cash flows that need to be discounted into the present. Figure 7.1 shows a typical list of bonds with valuation metrics to ascertain valuation mismatches in individual securities. A key metric is the YTM which is the annualised rate of return an investor would get over the life of a bond until maturity, taking into account the coupon payments and the price of the bond. Figure 7.1 Screenshot of corporate bonds from interview participant A fundamental corporate bond PM would look at credit fundamentals such as listed in figure 7.2 to estimate which bonds could experience a rating change or have a higher likelihood to repay their debt than priced into the bonds. Figure 7.2 lists the ticker, the bonds’ credit rating, 178 sector, its net leverage and interest coverage.75 If a company sees credit improvements, for instance a decrease in leverage or increase in interest coverage, the chances of a rating upgrade are higher and thereby the price of the bond would increase with the rating. A lot of trading in corporate credit is based on relative value strategies for instance comparing the YTM of a certain security relative to other issues in the same sector, or relative to benchmark rates or risk free rate, such as German Bunds. Figure 7.2 Screenshot of credit fundamentals from interview participant The difficulty for fundamental investors has always been how to adapt to changing contexts in capital markets, that is, how to assess market risk rather than idiosyncratic securities risk. Drastic changes in risk appetite bring about outsized market movements which are challenging to explain on the basis of an individual company’s fundamental business or even the industry environment. In the interviews, fundamental fixed income PMs stressed that the CSPP was a game changer for European corporates in initiating a bull market in their asset class. As analysed in chapter four, when the price driver for a large part of the European corporate bond market is the money flow from the ECB, it is difficult for fundamental PMs to make investment decisions based on credit fundamentals. If the credit fundamentals have not improved commensurate to the contraction in spreads, how can a PM justify buying bonds at much lower spreads? Immediately after the CSPP was announced in March 2016, there was significant uncertainty amongst PMs and analysts what to do. Citi Group summed it up very well in a note from April 2016 in that the issuer eligibility of the CSPP was not too clear and neither whether the ECB will buy in the primary or secondary market. However, CSPP- 75 Interest coverage can be defined as the ratio of operating income over the interest expense of a company, see Glossary. 179 eligible bonds performed very well after the announcement and PMs were able to adjust their security selection fairly quickly by reaching out to brokers and sell-side analysts. JP Morgan’s take was to ‘Focus on Scarcity’ and concluded to focus on €-IG bonds as these would likely become more expensive in light of the ECB purchases (J.P. Morgan, 2016). In our discussion around sensemaking, JP Morgan’s European Credit Team (2016) looked at the PSPP and other programs to gauge the likely upside to further tightening in the corporate sector bonds. In a presentation from April of the same year, Citi (2016) advised to stay in CSPP-eligible bonds as these had the highest potential of appreciation in their view and were the more conservative choice, piling into these bonds ahead of future ECB purchases. Given the relatively smaller size of the programme compared to the PSPP, the announcement of the CSPP on 16th of March 2016 came as a bit of a surprise and did not receive as much media attention. Despite the smaller size, the CSPP was arguably more unconventional as it was focused on private sector bonds. For the networks of PMs affected by this in the European corporate space this came as a surprise. AM7a: That [CSPP announcement] was a surprise, I would say, that they extended the program to corporates, I think that was not expected, and we didn’t expect this as well. So to that… and then we also saw it in the market reaction the announcement, we saw the spreads really tighten quickly, so this wasn't expected, and em, yeah you had a quick tightening, especially in the eligible bonds, but then the tightening path that happened afterwards that was, because everyone had to think about what does it mean for the market, and we needed some time to adjust the allocation. (AM7a interview, October 5, 2018) ‘Tightening’ in the fixed income space means yields contracting and bond prices rising. PMs deliberately targeted those PSPP-eligible issues that would face supply shortages in 2015. Pre-empting the behaviour of the ECB, active PMs decided to crowd into likely CSPP- eligible bonds first after the announcement. After the actual implementation of purchases by the ECB, flows continued into eligible securities and tightening continued. Risk appetite rose but the technical investment methodology and analysis of corporate bond PMs did not fundamentally change. Months into the implementation of the CSPP, only some active PMs were willing to move into riskier non-eligible issues and high yield. As two fundamental credit PMs from a large European asset manager deliberate here: 180 AM7a: I think the first, em, the announcement was at a time when the spreads were on a widening path, in ‘15 and the beginning of ‘16, they announced it in March 2016…There were several topics like China, commodity prices, things like that so the market was coming from the bear mood, and then with the announcement the whole thing changed, more or less from one day to the other…That's where's the most benefit. No matter if it's eligible or not, we wanted to be long risk. AM7b: Just to add, still at the beginning we asked, is it eligible or not eligible, just to see, ok is the ECB going to buy it or not, but ja, we didn't necessarily buy only the eligible bonds but as AM7a said, into higher beta bonds, and as the CSPP continued, we, I think, investors didn’t even ask whether it is ECB eligible or not. I think it was relatively clear which ones are eligible but there are some issuers like Heineken, for example, where some issues were eligible some weren't, but in the end it was just a beta play. (AM7 Interview, October 5, 2018) A ‘beta play’ refers to the fact that in a strong upward move, riskier securities would move comparatively more than the overall market. After the initial CSPP launch, PMs in the European corporates carefully weighed the fundamental credit risks of targeted bonds within the context of the ECB’s enormous monthly flows into the bonds. On the one hand, flows were guaranteed to drive up prices of bonds, but on the other hand, the fundamental valuations would not justify purchases to investors. The same held true for the 2015 PSPP where PMs analysed the potential supply shortages of German Bunds. PMs were carefully deliberating about eligibility and the extent of risk that they would take. Here an asset allocator at another large European Asset Manager elaborates on the dilemma at that time. I: And then when they introduced, em, the CSPP, 2 years ago, did you consider, what are the eligible bonds, we trade those, or we go for the non-eligible bonds, where there is a higher spread compression, for instance, were these trading ideas? AM6: Yeah, these are trading ideas especially the second one. At least the rationale for the first idea is, what you propose, we jump quickly now into the market, before the ECB is coming in, they pump up the bond price and I make a nice gain. Could work and it has worked at that point, but if you look at the, ehm, bonds, especially the 181 govies the ECB has been purchasing a lot at that point, like Italian ones… ‘bonds that are a bit under pressure’. You think that 8 years ago, where you got the hair cut on Greece, at that point people were a bit weary, are they really going to buy this? The influx at that point, when it was announced, the execution into govies was not that big, because people were quite uncertain, of what is going to happen at that point. So, they went for the second leg. Let's look for this, we will go into the more illiquid part, see what is going to be bought there and then do like a pair trade and make a spread play on it. This is something that happened a lot. (AM6 interview, October 5, 2018) Strategies to exploit the predicted behaviour of the ECB were manifold. First, there was the question of eligible securities and whether to crowd into these issues to front-run ECB purchases, which was mostly the initial reaction for both the PSPP and CSPP. Second, PMs could re-allocate to non-eligible securities, in this case high yield bonds, as these should disproportionately gain given the higher beta and higher spread compression. Third, given that the ECB was initially only participating on the secondary market for the PSPP, PMs could participate in primary issues and sell this on to the ECB making the spread. Fourth, PMs pursued relative value strategies, for instance with Italy as a short and long the core. Lastly, equity PMs could purchase shares of eligible issuers as the weighted average cost of capital falls with lower financing rates, thereby decreasing the discount rates for shares of the company and commensurately increasing the theoretical fair value.76 So there was a case to trade the whole capital structure, bonds and shares, of European corporates (see section 7.2.3). Liquidity soon dried up and it was more difficult, even for the largest AMCs in Europe to get bonds that were of interest to buy. AM7a: Ja, I mean, I think what happened is that, it was really easy to sell bonds to brokers because they were quite aware that they can just recycle it to the respective central bank. Even though they had lots of positions on their books they were just happy to buy more as it was easy to recycle. But then the uncertainty wasn't there that they cannot get rid of their positions. On the bid side you had quite good liquidity. (AM7a interview, October 5, 2018) 76 The discount rate in a discounted cash flow model is constituted proportionately of the cost of debt and the cost of equity, a weighted average cost of capital. Hence the lower the cost of debt for a company, the higher the theoretical fair value of the security is. 182 This vignette illustrates what HF1b mentioned about the decline of investment bank risk taking in chapter two. In the past, market makers could build bond inventories ahead of such policy announcements in anticipation of price increases, rather than scrambling for it after the announcement. In this episode of the CSPP, having good liquidity ‘on the bid side’ meant that PMs were always able to get a bid price for any bond they wanted to sell. However, with the competition from the ECB, brokers and other PMs wanting to buy eligible bonds, the offer side wasn’t that liquid. This lack of liquidity on the offer side meant that buying turned out to be less discriminate and careful. AM9: Eh, I think it drove the behaviour of less idiosyncratic analysis by the market because you just wanted to buy everything with a yield. So, it killed a little bit of that spirit of what we are supposed to do, which is, do your homework well. In terms of your analysis, my observation is that and it also drove herd behaviour, so more people would own it because the ECB would buy it, so you amplified the reaction in markets. (AM9 interview, January 18, 2019) Given the rise in prices, active PMs in the European corporates chose to sell holdings to brokers who were happy to stock up issues on their balance sheets with the new firepower of the ECB. As discussed in chapter four, passive investors would crowd into eligible issues alongside the ECB. Active managers were both competing for the offers in order to sell to the ECB at a higher price and liquidity providers by liquidating some holdings that appreciated in the rally. The problem for active PMs was where to recycle the freed up funds from the bond sales. With the ECB in the market, PMs were forced to adapt to a new organism in their financial network which required a tactical adaptation rather than a wholesale change in the analytical framework of their work. However, as the yields moved to negative territory, active PMs were not able to hold some bonds any longer as they couldn’t justify elevated valuations. Am7b: You could participate in new issues and then just recycle bonds from the portfolio to get the cash, but em, as AM7a said, there was positives, so liquidity went up and you could again trade…we saw these two spread tightenings in terms of gapping tighter at the announcement and then the actual start [of purchases], and then just continuously grinding tighter, and you saw like the flat spread curves and bonds 183 going into negative yield territory. I think that also squeezed us and probably other investors out of the bonds when we said ok for 20 years the spread curve is too flat, we want to sell or we don't necessarily want to hold negative yields on the bonds, so we sold those as well. So that was one effect, the bid was always there from the banks. (AM7 interview, October 5, 2018) Even though tactically participating in the rally was a strategy pursued by fundamental value investors, there was a threshold where the options ran out for active PMs to continue participating, at around 18-20 months into the program towards the end of 2017. Am7a: Ja, I mean you see it in the valuations, and then, we are fundamental asset managers so we want to really have spreads that are justified by, that are adequate for the risk, and if there is no spread at all then there's no steepness at all and as AM7b said, we started selling the bonds, as the valuations got more ridiculous. But then where do you invest? I think what happened is we bought some of the non-eligible bonds that maybe offered some spread, or sectors that don't trade too tight, or where the spread is slightly wider, we try to hide there, where there is some value at least, that was the strategy at the stage where it was the tightest, at the end of last year, where spreads were really tight and I think that was the strategy at that point. (AM7 Interview, October 5, 2018) As elaborated in chapter six, despite brokers not being able to disclose the quantities of trades they have transacted with the ECB and NCBs, PMs utilise the daily runs to ascertain who is the largest market maker in certain issues. PMs were taken by surprise by the CSPP, but were able to extract gains out of the projected behaviour of the ECB. Strategists at investment banks helped quantify the program to fundamental PMs who acted both as liquidity providers to the ECB but were also in competition with the ECB, rushing in and selling bonds back to the ECB at higher prices. As a result, the performance of €-corporate credit funds was good. The behaviour was co-ordinated between ECB experts in that the playbook from the PSPP was there as a reference and those able to move quickly did. The sensemaking within the organisations I visited occurred quickly once people figured out what the program meant for the structure and behaviour of the market. Thus, rushing into eligible assets seemed something very instinctive and something that tends to happen in many contexts. 184 7.2.2 Sovereigns and Rates PMs in the sovereign credit space are traditionally very much in tune with the ECB and actively trade any information edge. PMs trading global macro, an investment strategy allocating capital based on broader systemic trends, have historically followed central banks and governments closely in their trading activities. The best examples for this was George Soros and Stan Druckenmiller shorting the pound when the BOE broke with the European Exchange Rate Mechanism in 1992. Other examples which were brought up in earlier chapters are Moore Capital and Brevan Howard. Sovereign bonds, rates and FX are the focus of global macro and PMs in that space have been traditionally much more in tune with central bank communications, monetary policy and geopolitics. Usually, yields widen if debt is downgraded by a rating agency as the bond would require a higher return for the higher risk of default. As an example, Greece was continually downgraded at the height of the eurozone crisis in 2012 and the 10 year yield on its government bonds rose to ~30%. Conversely, US Treasuries are seen as a safe-haven asset and have different demand dynamics – investors would buy Treasuries in ‘risk-off events’ or crises. For example, when the US government hit its debt ceiling in September of 2011, US Treasuries lost their AAA credit rating from Standard and Poor’s. Although counter-intuitive, US Treasury yields tightened as the debt downgrade sparked a global risk-off event, leading investors to seek safe-haven assets and buy US Treasuries as a result. In the eurozone, the Austrian government issued a 100 year bond two years ago and it only yielded just over 1%.77 Suffice to say, a lot of things can happen in 100 years, so the fact that the current yield is just over 1% illustrates the low interest rate environment PMs faced. Hence PMs in global macro, rates and sovereigns have to take a myriad of factors into consideration when making investment decisions. Figure 7.3 shows a screenshot of European bond yields and the predicament PMs were facing in the low interest rate environment when the PSPP was launched in 2015. Almost half the bonds up to a 10 year duration had negative yields. Simply put, an investor buying a bond with a negative yield, in effect pays the issuer to hold the bond, rather than being compensated for the risk of holding the bond. That implies that the price of risk had pretty much disappeared. 77 https://www.bloomberg.com/news/articles/2019-06-25/austria-weighs-another-century-bond-for-yield- starved-investors [accessed 03.04.2020]. 185 Figure 7.3 Screenshot of 2015 European Bond Yields from interview participant Global macro PMs are often located in the Asset Allocation division of an AMC, contributing to the top-down strategy of a fund. Macro PMs could also be dedicated sovereign bonds and rates specialists. At hedge funds, global macro specialist would be involved in trading risk in all asset classes, including e.g. equity futures. In the context of this study into the ECB, corporate bond PMs would tend to be more bottom up, i.e. to analyse corporate fundamentals and try and extract alpha through idiosyncratic risk and mispricing. Macro PMs would focus more on risk-on and risk-off, thus implicitly focusing on the beta, the overall direction of the market. One of the European Rates PMs elaborated on changes in approach to analysing the ECB after the European Sovereign QE was launched. I: Let’s say, if you look at pre Draghi and Draghi's presidency, was there any change in your methods or investment strategies in your space? AM7c: Em, yes a lot of things have changed because the price of fundamental risk has… vanished, and you just buy what is being bought by the ECB, and it is roughly the same pattern we have seen in the US. I have spoken to some US investors and they say just buy what the Fed is buying and sell what the Fed is selling. That is the right thing to do, it is only the flow that is determining … assets and not the inherent risk. So in the rates space, there is not that much risk in there, but ehm, you can see that maybe risk in the periphery [European periphery] is not quite where it should be perhaps. But who knows where the fundamental prices of Italy would be. Or if you look at a long term fundamental model, you see now and then the 10 year yield is like inflation expectations plus the potential growth and a bit of a risk premium, and biggest part of this equation is the very negative term premium, and that's part of the 186 scarcity… Couere [ECB Executive board member] has shown a chart where he done a scatterplot of Bunds vs the percentage of Bunds available in the market. That shows the same pattern, the greater the scarcity in the market, the bigger the % of ECB holding Bunds, the more expensive it is for market participants to do a repo…And the second thing has changed in this market, there is no separation of doing a repo collateralised or non-collateralised. (AM7c interview, October 5, 2018) The scarcity in German Bunds was driven mainly by the ECB purchases and given the low interest rate environment, PMs were looking for capital gains rather than yields and follow the flow of the ECB as a result. That also meant that the portfolio rebalancing effect examined in chapter four did not unfold but investors would stay in German Bunds. This was also echoed by another PM in that the focus shifted to buying sovereigns on pure capital gains potential, thus buying what the ECB is buying, and ignoring the yield aspect of the return structure. HF1b: People think there's going to be search for yield, but this is also wrong…You can either get the yield or you can get the capital gain…US equities made 300% return and the US$ went up 50% against EM currencies and German bonds made 80% return. So investors started thinking…why should I go and buy an Ecuadorian government bond at a 9% yield, if I can make 80% in Germany, why should I go and buy three year government bonds in Brazil with a 10% yield if I'm going to lose 50% on the currency…By the time the ECB, really after the European debt crisis, the ECB really stepped up QE…at that point investors finally get it. (HF1b interview, September 25, 2018) Furthermore, AM9 and AM6 both underlined that the ECB has become ‘one of them’ in market participation, so that its behaviour becomes a more important input when they both perform asset allocation. AM9 sees this as a significant input, while AM6 echoes this but refers to it as a ‘residual’ input. In internal organisational sensemaking processes, such as the investment committee meetings at AM6, a ‘residual’, referred to in the literature as ‘residuum’, remains something external that cannot be left out of internal deliberation and that remains after the usual internal modalities are completed (Weick, 1988, p. 307). 187 AM9: Yeah, eh, for me yes, because I'm active in the European bond market space, so if I don’t understand how the dynamics of their purchasing programs shifts, then I could be making a mistake, understanding the net supply component of the market, for example so, let’s say if they are buying, just as a rough example more BTPs, so Italian government paper, or more French government paper for example, and that switches this year, and let’s say for example investments in those markets become smaller, it will drive significant underperformance purely because they’re taking down less of a supply coming to the market, so they’re basically like us now right? One component of the market right? So it’s like not understanding what the demands of pension funds are, or real moneys. So basically, it’s an important constituent. I: But.. is the tendency to follow, so to imitate or…? AM9: Hmmm, I can’t say for myself, because my decisions would be primarily driven by, for example, do I like duration and if I like it, where do I like it. It could be an input, but it’s not going to be the driver of the decision. Certainly not to imitate. Ehm, it would be difficult to make money that way. Certainly, in this day and age there might have been, in the past, for example a lot of times people would be, investors would be bearish Bunds, just purely from the level basis, because they are so low yielding. You know, why should one own them? And yet they consistently outperformed expectations. Year in, year out, since QE started. You know you obviously had a few wobbly sell-offs so generally they outperformed. So then why? The only reason is that they’ve [ECB] taken out so much of the supply of German paper, out of the market, so they basically own a third of German paper, over a 1/3 of German paper eh government paper. So de facto they do affect the performance yes. (AM9 interview, January 18, 2019) AM9 elaborated on what had changed since Mario Draghi and more generally since the GFC. While AM9 stressed that in global macro, you always care about central banks, the shift and context has changed significantly in terms of the ability to extract returns. AM9: Well… I would say that since the financial crisis central banks have played more of an essential role, because they've had to come in and act as a stabiliser when markets have been broken, ehm, and they certainly have done that, ehm, in terms of alleviating fear in markets. Of course, on the other hand, I am sure we talk about this later, they created other issues, but that’s a different topic altogether. Ehm, so I would 188 say, I would say, 'how has it changed?', We have to be much more in tune with how they are thinking… I: Hm. AM9: That would be the main change I would say, so as an investor I feel that if you don’t understand their reaction function you do so at your own peril, because as much as you think, and there are a lot of equity investors for example that are big believers in the bottom up approach and that is fine, it’s part of one’s process, but if you don’t understand what’s actually driving the Beta component and a lot of times since the financial crisis it’s been about central banks, that could affect your returns humongously. Now, I’m slightly biased because I am on the macro side, so in macro we always care about central banks. I: Yes, hm. AM9: But if I had to pan out my career before and after the financial crisis, then of course there’s a significant or drastic rise in importance of, eh, central banks in regards to the reaction function in markets. (AM9 interview, January 18, 2019) The ECB has shifted and changed the investment context significantly for PMs through its policies. While each PM continued to follow their investor frame regarding security selection, PMs faced an unknown territory and through sensemaking processes and creative adaptation managed to extract economic profits from this new context. However, not all PMs were able to adjust to the changing context brought about by the PSPP and CSPP as well as the low interest rate environment such as the large insurer I spoke to. Thus, in a sense the intent of the central bank to encourage PMs to rebalance portfolios to riskier assets and move out of their preferred habitat does not seem to work as in the first instance, PMs stuck to investment methods and tactically adapted to the new context. PMs were very much aware of the limits of the policy and used this insight to pre-empt ECB purchases. 7.2.3 Equities On the equity side, the adjustment made to QE was mainly through the discount rate, in that lower rates will cause higher Price to Earnings ratios and PMs considered this as a simple adjustment that despite downward earnings trends, the asset class received a re-rating through a lower discount rate and higher Net Present Values. For instance, instead of a market PE of 189 15x, PMs would have to adjust this to 20x, or start looking at forward earnings to justify purchases. Either way, central bank liquidity entered capital markets on a massive scale. In Europe specifically, the PSPP caused a significant rally in assets shortly after the PSPP announcement, but this quickly stalled after a few months as investor flows focused more directly on credit. HF1a: So the way we try to look at it is, that ultimately the only thing is distorted is you know the valuation I put on something because the cost of equity has changed, right? money is for free, multiples go up, but the business day to day doesn’t change, so to me that's the challenge is what used to be 15x PE now it's worth 20x PE, because the cost of equity is lower. Structurally what central banks have done, and conversely what it means, it spreads everywhere else right? Whether you need or you don't, it's just your proxy and that the distorting factor that became so extreme now. (HF1a interview, June 8, 2018) The strategy to hold bonds with equity like returns also shifted perceptions of equity. AM6 highlights that the whole capital structure, including shares, of companies included in ECB bond buying benefit from money flows as a sort of second leg to the initial contraction in spreads of the company’s bonds. A strategy in equities was to consider dividend yielding stocks in the eurozone, that have bond-like characteristics but also expose investors to growth. AM6: I mean it is something that is, actually you need to go a step further than that, what is the actual investor looking for, … for the larger investors it is a substitution for bond returns, so that is why you have larger inflows into growth, as they act in a kind of way like a bond proxy, so you have large inflows into strong dividend players, this is like your second leg into a bond proxy. Yeah so you need to purchase like, investors buy these stocks, in that sense that they em, that their return, their assumptions, … I make a stupid example here, but we don’t choose Deutsche Telekom because they're being purchased [by the ECB], this is not a criteria, it's a residual, nothing else. (AM6 Interview, October 5, 2018) 190 As discussed in chapter six, new ways of looking at stock markets evolved on the basis of global central banks’ balance sheet expansion. In figure 7.4 forecasted returns for overall stock markets would be based on the extent of projected balance sheet expansion, in other words the amount of asset purchases of the Fed, ECB, BoE and BoJ. Figure 7.4 Screenshot of US Investment Bank Forecast of the S&P 500 US equity index based on Central Bank purchases Source: Contact data. Hence with a challenging yield environment and lack of improvements in corporate fundamentals, PMs in different asset classes start to derive more of their expected returns from central bank balance sheet expansion. The behavioural patterns of PMs after the announcements of the PSPP and CSPP discussed in this section could be summarised as actors in financial networks following the flows and detailed guidance of the ECB in the absence of yield and growth. PMs aim to exploit this in their tactical adaptive processes, such as buying stocks of companies that are included in the CSPP. 7.3 Communication and Information exchange 7.3.1 ‘Whatever it takes’ PMs echoed the situation leading into Draghi’s 2012 speech in what was described in chapter six by former Irish Central Bank Governor Honohan. PMs saw words uttered by Mario Draghi as a reflection of likely courses of policies. Hence, PMs viewed words contextualised Graph removed due to confidentiality. 191 in actions of the ECB. In that sense the interpretation of his words became grounded in a very particular historical context, the Euro breakup scenario (AM7c), the absence of a trend (AM3) and a panic to act (AM4), discussed in this section. As AM7c elaborated when discussing the ‘whatever it takes’ quote. AM7c: Em, well before the 'whatever it takes', markets were very scared about default risk in the market, breakup of the ECB, … before that quote, the market was scared of defaults and Euro breakup, and afterwards the market was more convinced about the fact that the ECB is going to calm down markets, the freeze in some markets like the periphery to break it up and make the market tradeable again, there are a couple of different phases that were followed by the quote, to bring back liquidity to the market was the first step, to bring down the short end of the yield curve, that eh, like it is possible to borrow short term liquidity and then to anchor the short end of the yield curve and then to flatten the entire part of the long term maturities. So that worked all very well, em, it brought back quite a good amount of growth, even though inflation is not yet where it used to be. (AM7c interview, October 5, 2018) AM3 and I also discussed this quote and the circumstances surrounding it. Looking at AM3’s past observations and dealings with the BOJ, AM3 asserted that the ECB language became more powerful in certain contexts, for instance when inflation and rates were low, PMs look for guidance from the central bank. Words can definitely not be removed from their context and are inherently linked to potential policy action of the bank. AM3: Language and meaning becomes separate - words become more than words. Words become more important when inflation is low. A continued low spell of inflation puts more emphasis on words of the central bank... When markets don’t see a significant trend, language becomes more powerful and investors rather 'ride the nose of the shark’. (AM3 interview, August 17, 2018) In the recent past, ECB communications have been used to moderate, finetune and contextualise policy and interpretations have become more nuanced. Hence the ECB prepared PMs to adjust positions ‘slowly’, such as in the ECB tapering and the unwind of the 192 Fed’s balance sheet. AM7c describes the development of ECB communication under Mario Draghi further and AM4 highlights the significance of Draghi’s communication style grounded in the current economic context. AM7c: What the Fed started to begin with Yellen is to do some dovish hikes, and the same pattern communication is done by Draghi in a much more sophisticated way to do hawkish measures, like cutting purchases, but sounding dovish and keeping the yields low. And this way of doing hawkish measures with a dovish language it still works in the market, the wording of central bank quite dovish as the market believes it, so that’s the recurrent procedure you see in almost every ECB communication. (AM7c interview, October 5, 2018) AM4: It was a sign of panic [of the ECB], you could only take his word to be good for 2 months… 'Do whatever it takes' is unmeasurable and undefined. If you say it at the wrong time, [the] market will crash, so it's all about timing and the audience… Positioning has been a bit more diverse than in the past, before it was starting from the lowest [exposure], central banks, then real money funds, then banks and hedge funds. Central banks were never the front runners, but this has become very lopsided now. (AM4 interview, August 21, 2018) ECB communication is thus perceived by PMs as grounded in the specific social context of the market and a tool to reflect potential courses of action. Given that the ECB is an oversized market participant, words are used to finetune policy and engage PMs. 7.3.2 Information and expert networks All interview participants in this study had a high seniority, but the different levels of understanding of the ECB’s monetary policy and its impact on different asset classes varied by the location of the participants. While AM4 was a previous central banker and a current PM, AM4’s level of detail and awareness of the historicity of events of Mario Draghi’s presidency did not eclipse the knowledge of interview partners in Frankfurt and London. IB1 in Interest Rate Sales and IB2 in Macro Sales regularly interact with central banks on the trading side and have not acquired the level of detailed analysis respondents from London 193 and Frankfurt had established, despite being active in the same asset classes. Interview respondents in Munich were also not as tuned into the debate as their Frankfurt counterparts. Chapter five showed the inherent home biases in the sovereign bond markets of the eurozone. Coval and Moskowitz (2001) also showed that geographical proximity can result in outsized returns as investment decisions are highly informed. This was also echoed in the PM comments on this topic. While global macro and rates PMs agree that it is important to have a detailed understanding of internal events at the ECB, corporate fundamental PMs don’t necessarily network to the same extent. AM6: I think there are two different levels of communication. So if you open the TV, your Bloomberg and look at Mario Draghi speaking , everybody gets it, but the point is… and that is something our fixed income guys are very good at, they have good connections into the ECB…You can go there, meet these people and you can get some insights in the third layer. But then you go into this, what I call, probability phase. On the top it's what Mario is speaking, something that is 100% certain, and then you can go like a layer 90, 80, 70, 60 so it's not like they are facts of course, but they are attached to probabilities. So their outcome, let’s say, is a little bit odd and it's up to the PMs to decide, what is the probability I can work with or not. But they are closer to the ECB than let's say someone in New York or Hong Kong... (AM6 interview, October 5, 2018) Contextualising ECB communication becomes even more important when considering the different social positions of the individual within the ECB uttering the words. Coeure seemed to be popular with interview participants for instance, with most of the Frankfurt and London participants regularly referencing him. AM6: That's the reason why you have the [Fed’s] Dot Plots, you can see what each person [is forecasting] ….you don’t have that in Europe. This is why they try to read what is Rehn saying, what is Weidmann saying. This is what I said before, people talk about probabilities. For instance, if they come out and say something, you are trying to weigh how big is their role in the council, you're trying to be ahead of the curve. (AM6 interview, October 5, 2018) 194 This echoes that the central nodes in the BMCG network (see chapter six) are based out of Frankfurt and London and the importance of having reference points other than the official ECB transcripts. 7.4 Reflective sensemaking 7.4.1 Communicating the exit The ECB can help PMs understand the context and applied rationale when facing new incoming data and in a way clarify the difference between what the central bank does and what it thinks. The example of dovish hikes raised by AM7c in Section 7.3.1 is a case in point. The ECB is announcing the net withdrawal of liquidity, but saying that the greater context of its decision has not changed and that the bank will remain accommodative, trying to calm PMs. However, there is a difference between what the ECB wants PMs to know and how PMs ‘play’ it. If PMs know that dovish hikes mean tightening in the end, the PM knows that unless there is a significant market turmoil, the ECB will continue on this path. In June 2018, only three months after the reduction in net purchases of its QE programs, the ECB hinted at an end to the programs in December 2018. The end of net purchases just meant that the ECB will maintain a static balance sheet and didn’t mean that the ECB would unwind its balance sheet altogether. Hence the size of the balance sheet was maintained and maturing bonds would be replenished after the end of net purchases. AM9 described the end of net purchases and the likely courses of future action for the ECB. AM9: The same was true, I believe, when QE was tapered last year in Europe, and you saw pure underperformance of European corporates, both in High Yield and IG. Now, of course, it’s not the whole story, not the whole truth ehm, you know there’s also the fear of a slowdown in European growth, were also a key driver of that, eh but, even the severe underperformance which takes you to the top quartiles of underperformance historically, especially on a relative basis, it probably tells you that it was overplayed by this reduction in QE and this expectation that QE ends. So, it had a meaningful impact on the way in, and a meaningful impact on the way out. I: I suppose the ECB, can’t really unwind the balance sheet without causing significant rise in yields especially in the corporate bond market? 195 AM9: Eh, I am not sure how they are going to be able to achieve that. I think they will keep driving the market with a static balance sheet, I mean certainly the Fed did that, if you think about it, kept the balance sheet static for quite a few years. And as soon as it signalled to the market that we might start to, ehm, actually even taper, just the taper announcement itself, caused a huge shock in the markets, emerging markets especially. (AM9 interview, January 18, 2019) AM7b described the impact on liquidity in the face of the tapering announcement. AM7b: And now I would say it is more fading but you could actually see that, those players who were bidding strongly last year [brokers] are kind of, maybe scaling down a bit as they see, CSPP is also fading. It's still good but the question is whether liquidity is going to remain the same when the CSPP is coming down. Now, coming to the end of CSPP, it is more like about not being in the eligible bonds and being in the non-eligible bonds to avoid the spread widening of the artificially tight bonds. (AM7b interview, October 5, 2018) Hence PMs were actively deliberating about the end of QE and those with an active mandate were trying to avoid the potential drawdown. The changing context from expansionary monetary policy to tighter financial conditions required PMs to adjust behaviour throughout 2018 and 2019. The balance sheet unwind was top of mind even in Hong Kong in mid 2018. AM1: It's nothing new really, but they do know what's going on, I attended a meeting with the top 10 people [of respondent’s organisation]. People are unaware of what the balance sheet unwind will mean for markets and I'm concerned that this has not been priced in and people are too complacent… Market participants still assume that the ECB knows more than the market, although I doubt that's the case. People assume they have a superior analysis… It's not healthy to have the ECB participate as a market actor to the extent they do, they haven't fixed anything. (AM1 interview, August 16, 2018) 196 The change in the competitive landscape by an actor such as the ECB entering markets certainly distorted the context for PMs to make decisions. AM9 goes into more detail on the impact the end of QE will have. I: From an investment management perspective, do you feel that this has changed the competitiveness and competitive field of the investment for portfolio managers? AM9: I think more in the sense of going forward we probably have to go back to doing our jobs..[laughs] which we get paid to do. I would actually say it pays to have more active management going forward. If we are going to be optimistic, cause this is being hugely optimistic, then say, we are going back to operate in a pre-2008 environment, then security selection, sector selection, because last year you could have made a lot of alpha just by deciding inter-sector, inter-country, making your correct choices, your asset allocation choices, there were huge opportunities. A lot of PMs will tell you everything underperformed last year, absolutely true everything did. But for example, if you owned health care vs European banks, you made a killing right, 50% …35% at one point, outperformance. Those are the type of returns, basically what that is saying, you have to be more discriminate where you put your money… (AM9 interview, January 18, 2019) The modalities described by PMs in different market segments reflect Weick’s (1988) enacted environment. In the case of the CSPP, PMs first react to the announcement and then figure out what this actually means for the market structure and after the initial tightening of spreads in eligible bonds, the rally spreads further. Action often preceded sensemaking and strategies were later adjusted and re-developed as a reaction function to explain what had happened. Thus, it is difficult to anticipate how markets react to certain ECB actions, as actions often construct the rationale of the occurrence and thus consequences of the announcement of the PSPP and CSPP have been difficult to forecast even for BMCG experts. BMCG members in chapter six were surprised at the market reaction during the initial announcement of QE. As a result, it became very difficult for the ECB to withdraw from QE as the organisation became committed to the path it laid out, for the PSPP was initially only planned until September 2016, which then turned out to be December 2018. “Once a person becomes committed to an action, and then builds an explanation that justifies that action, the 197 explanation tends to persist and become transformed into an assumption that is taken for granted” (Weick, 1988, p. 310). 7.4.2 End of QE? PM evaluations As noted in section 1.2.1, increasing responsibilities were progressively put on central banks. The dangers of centralising and consolidating too much responsibility and authority with the ECB are manifold. As noted elsewhere, governments had given the sole responsibility of the economic recovery to central banks after the GFC, de facto replacing fiscal with monetary policy (El-Erian, 2016; Tucker, 2018). As former ECB central banker Papadia put it: “Central banks have become too important players on the market for government securities in the USA and the euro-area, but also in the UK and Japan, to maintain that there is strict separation between fiscal and monetary policy” (Papadia & Välimäki, 2018, p. 253). The process of centralising responsibility and authority can lead to a reduction in the level of competence dealing with a particular problem (Weick, 1988, p. 312) and this had become clear when speaking to PMs. As the interviews were conducted in 2018 and 2019 with the preliminary end of QE on the horizon, the interviews also focused on whether PMs thought that QE achieved what it was set out to do and how PMs evaluated it. Overall, responses on whether the ECB had been successful were conflicted and there was a certain amount of criticism from PMs about the ongoing QE. While PMs’ portfolios performed well and benefited from the bull market in European fixed income in 2017, the tapering and announcement of the end of QE in 2018 caused this performance to reverse. Furthermore, the policies have distorted other areas of the market which seemed to be an issue particularly for fundamental PMs. AM9 elaborates on the negative impact on bank lending in the eurozone. AM9: I have a lot of respect for the ECB as an institution, it has dealt with very difficult situations, it’s a very professional institution from my interactions…They have a very difficult situation, because they come last to the party, you know, they waited q long [with QE], it’s much more complicated for the ECB because it’s got negative rates now, I personally think they should exit from negative rates, and negative rates doesn’t mean that you are starting a super hiking cycle. But I think negative rates are terrible in Europe, because they create all these other problems with banks, as I said. And you have seen banks have been totally beaten down. And beaten 198 down banks are not going to be induced to lend, and because Europe relies heavily on borrowing from banks, then you defeat the cycle from that, so you win with the Euro but you lose with your banks medium term. And eh, I know that they might say we are not in the business to make banks profitable, but your policy is directly making banks unprofitable, I would counter that. (AM9 interview, January 18, 2019) Towards the end of our conversation AM9 concluded that we wouldn’t know what would have happened otherwise without QE and with the foreseeable end of the asset purchase programs at the end of 2018, AM9 expected PMs having to do more fundamental analysis again as the stimulus to bond markets gets withdrawn. In AM4’s concluding remarks, the fixed income PM from a large Asian Asset Manager, made it clear that AM4 disagreed with the overall idea of QE. AM4: They shouldn't dictate the market! Let good old recession take place? They just don't believe in the business cycle. The Fed has an easier job, while the BoJ had too many different roles in the past… ECB has too many countries which makes it more difficult for them. (AM4 interview, August 21, 2018) AM5 from a large European Asset Manager and Insurer concluded that it was increasingly difficult for the group portfolio to adjust to the purchase programs and low yield environment of the ECB. AM5: Some market actors can but we don’t move that quickly, for us to make such decisions at the periphery takes a long time. The more levered mutual companies such as Aegon, NL insurers and Talanx, who owns Hannover RE, are now moving into unlisted assets such as utilities, guaranteed toll, residential mortgages etc, and there has been a crowding effect as a result of the ECB policy. So this has caused some bubbles in different parts of the market. Solvency 2 regime was introduced at the end of 2015 and market players are getting used to new systems and the mark to market. So the new solvency regime causes slower adjustments to monetary policy. I: Overall, how do you feel about the ECB's market-based monetary policy? 199 AM5: Well, overall they shouldn’t be doing this. We can learn from the Japan example. There's no income on savings and all the M&A deals I look at make sense because of the low discount rate. And you have bubbles building as a result. All the inputs for valuation models are distorted. So, all the research I do is affected by it. (AM5 interview, October 1, 2018) While AM7 expressed overall support for the expansionary monetary policy of the ECB and the temporary revitalisation of bond markets, concerns emerged over the longer term cost of the programs. I: …has QE been a success in the eurozone? AM7C: My personal opinion is yes, of course, but we haven't seen the cost yet. It's just we are having a party and it's unclear whether we will have a hangover in a couple of years. AM7a: Yes, I agree, cause we are still in this exceptional monetary policy, but we have to get out of it, and then we will see what happens then. The economy has to be in a good growth path to do that, and it would be the next, I don't know what will happen to the economy next year, but it's maybe quite difficult to get out of the policy we are in now, so but if that works what the Fed did, then I think it was very good, but hard to say now. (AM7 interview, October 5, 2018) As pointed out in chapter five, the issue of the PSPP legality was raised to the German Constitutional Court which was initially backed by QE-critic and ECB GC member Jens Weidmann. To recap, the issue that was brought forward to the ECJ was that the PSPP could constitute the financing of government budgets, which is not permitted under the Treaty. The previous comments from AM5 put doubt on this and it was then stated explicitly by AM7c. I: So you don’t have any problem with the ECB becoming an active market participant, e.g. through the purchase programs? AM7c: It's roughly on the edge, because the ECB is not allowed to finance countries for example, to say it in general, em, financing of public debt, but this program leads to better market conditions, that enable countries to issue debt at a more attractive level, and without this program it would be perhaps harder for them to meet their 200 financing needs. And so you cannot really say they're financing this country, like Italy or Greece or whatever, at least not now but Italy, or Portugal before, but without the ECB, there might be some circumstances or situations in the market where it would be harder for them to come to the market and issues that they buy. (AM7 interview, October 5, 2018) HF1a pointed out that the ECB and central banks globally have run out of options and hinted at what others have highlighted as the last option in helicopter money. HF1a used an analogy to underline the failure of QE. HF1a: It's basically like going back to the old age where there was a court jester, there was a king, the king was the guy who dictated what they ate, so he had the table of, they were dancing and what, the court jester was there to entertain people. Central banks are like the king, everyone else's feasting and enjoying it. As long as you got entertainment you got everything, you got food and the problem there is when that ends, it's horrible right? People don’t know what to do with themselves… The Fed had 10 action points, when they did QE. They've done 9 already. The last one they have to do is helicopter money. You can’t actually give people money but what you do is you get them to pay no tax, it's a subsidy. If you have zero tax you actually feel richer and you spend. But you go down that way, then it’s over and then how do you raise tax from zero to 30%? So that's the last thing left. (HF1 interview, June 8, 2018) 7.4.3 Analysis In summary, almost all PMs see the QE of the ECB as a ‘party with unknown costs’. Despite PMs being aware that fundamental asset prices were significantly distorted in pretty much all asset classes, from equities, fixed income to property and private equity, this exuberance resulted in ‘lazy decision-making’ and some PMs were in riskier assets than they would otherwise be invested in and as a result, experienced underperformance during the period of tapering by the ECB in early to mid 2018. The reflections of PMs in our interviews were a sober assessment of the ECB’s monetary policies, both in acknowledging the difficult environment the central bank faces, but also the scepticism about the cost of QE. German 201 Bunds were clearly the top choice for PMs to crowd into and returns were concentrated there at announcement as well as during the implementation of the policy. The CSPP had an outsized impact on the way in, and then again, on the way out. Corporate PMs bought more €-denominated IG corporate bonds and those who started to diversify into high yield a year into the program paid the price in 2018 when both IG and high yield underperformed on the back of the ECB tapering. Not only passive funds herded into IG corporates but the interviews showed that active managers did the same. Analysing investment behaviour around the PSPP helped PMs with sensemaking and contextualising the unexpected CSPP announcement and implementation. An analysis of the analyst notes around a month after the CSPP announcement also showed that analyst consensus was not very clear on which trading strategy to pursue. Behaviour was very much anchored in the past experience of the PSPP. In these interviews, the Macro PMs were very much following the ECB flow into PSPP eligible securities and PMs as well as analysts were acutely aware of the supply and demand in the various eligible-sovereigns. The multiplexity of relationships between the ECB and these senior PMs enabled a sort of ‘endorsement effect’ on investment conducted by the ECB and embeddedness increased co-investments. This was also noted to be the case not only during the PSPP and CSPP (2015-2018), but also in previous crisis episodes when the ECB bought Greek government bonds outright. As Uzzi (1997, p. 52) found in a different context, social networks reduce the complexity of investment decisions. When the ECB slowly signalled monetary tightening during 2018 and therefore a potential withdrawal of the network, PMs faced increasing uncertainty and were not able to react clearly. As the whole complex underperformed, there was no place to hide. With the still static balance sheet of early 2019, the damage done was not existential, but based on the interviews, the ECB won’t be able to unwind its balance sheet without causing major financial contagion. In other words, a scenario where the ECB would sell bonds and reduce the balance sheet would have had unknown consequences. 7.5 Conclusions Structures are certain patterns of exchanges that have some continuity over time. With the entrance of the ECB into sovereign and corporate bonds, the structure of the bond markets 202 was altered significantly. Brokers benefitted from the influx of liquidity into the market and higher transaction volumes, but are now sitting on inventory. Fundamental PMs faced difficulties in adjusting their usual investment behaviour, based on fundamental analysis, and had to inevitably pay more attention to the ECB’s purchasing behaviour and adjust strategies tactically. Sovereign PMs on the other hand have traditionally dealt with central banks closely and have established certain patterns of behaviour. Like fundamental credit PMs, equity specialists had to undergo a significant adjustment process and rely on strategies from peers in the credit space. From a theoretical perspective, PMs were acutely aware of the ECB’s intentions and what monetary theory suggests about portfolio rebalancing in the quote at the beginning of this chapter. However, this model of behaviour was undermined by PMs thinking on their feet and deliberating alternative strategies. With the help of communications from key strategists and by examining previous episodes, PMs followed a more behavioural process of investing and moved the focus away from pure yield investing and valuations to tactical adaptation. As in Weick’s (1995) sensemaking, plausibility outweighed accuracy. The ECB had created an unknown and uncertain context by sustaining near-zero to negative interest rates and launching unprecedented monetary policy. In this context of uncertainty, PMs chose to ‘follow the leader’ as the investor frames and decision-making processes were in the process of adjusting. This finding echoes the notion in organisational theory, that imitation is a response to uncertainty in decision-making (Rao, Greve, & Davis, 2001) and resulted in regret for some PMs that are now sitting on bonds that are riskier than what they would usually choose to incorporate in the portfolio. This feedback mechanism caused PMs, who usually operate in different habitats, to adapt behavioural patterns. Overall, PMs made sense of the European QE as they went along, learning how others react to announcements and ECB purchases. Sensemaking took place in the months after the announcements and market participants formulated strategies as events unfolded. The initial reaction was to follow the strong impetus that was felt through networks in the bond market. The findings underline that PMs follow their own investor frame, and at the same time, take into consideration the context they face, the assessment of how other market participants are acting and how to achieve their own interest. Going back to Burt’s (1982) action theory, PMs did not act in isolation, but through their own action and reflective monitoring of each other’s 203 action created social structures, which in turn, constrained meaning creation. Simply put, PMs were influenced by everyone else’s reaction to the purchase programs and crowded into the most popular trades, thereby limiting the diversity of trading strategies. The example of the unexpected CSPP policy launch has shown that sensemaking and decision-making in financial markets is deeply embedded in social relations of experts and based on historical experiences of these networks. Thus, as shown in chapters four and five, there is a common behavioural pattern that develops over time in such networks of the ECB, PMs and brokers. 204 8 Findings and conclusions My initial view of central banks when I started this project was that of powerful and omnipresent institutions in financial markets, able to implement affairs as they see fit and that instrumental independence had led them to be relatively shielded from outside influence. However, the thesis showed that the ECB is entangled in social and financial networks of senior decision makers. By examining the content of ties of the ECB’s financial networks and examining the multiplexity of relationships within the social networks of ECB experts, the thesis uncovers the ways in which the ECB is entangled in the social networks of a European investing elite and is influenced by the interactions within these networks. The ECB aims to enact sensemaking in its communications in expert networks, it communicates its purchasing behaviour through forward guidance, it diffuses new policy ideas before actual implementation to gauge expectations of systemically important market participants. As an unintended consequence, the ECB’s policy objectives of portfolio rebalancing and international diversification didn’t really work out as the ECB’s actions alters behavioural patterns in unpredicted ways. The thesis has applied methods from the economic sociology of networks and re-configured tools to analyse the decision-making in social and financial networks around the ECB. Structural holes, homophily, strength of ties and centrality were used in the network analysis. ECB experts actively engage in the networks and fill structural holes. They exert influence on the interpretation of information in financial markets. Corporate interlocking networks are shaped by decisions being made in social networks of this financial elite. These social networks exist as an overlay to the corporate interlocking financial networks. The findings underline that the main interlocutors of the ECB, senior PMs and their institutions in reverse, influence the actions of the ECB itself in occasions of sensemaking. 8.1.1 Findings This study found that PMs organise their everyday working lives around occasions of sensemaking, they prepare for regular presentations in the expert network of the BMCG, they contextualise this tacit understanding and translate this in internal investment committee meetings, into their own investment decisions and asset allocation and in meetings with investors or the financial media. Occasions of sensemaking give a rhythm to the PM’s 205 working life, how she organises her time, prepares and positions herself on issues in anticipation of such meetings. As seen in the case study of the BMCG, structural holes in the European financial markets enable individuals to advance their careers either by moving from the ECB, to a hedge fund, to traditional asset management houses and investment banks, or by gaining significant analytical input in the asset allocation process, for instance by avoiding Italy where the ECB would like more investors to move to and allocating to places where supply is tight such as Belgium, Austria or Germany. The consequence is either outperformance by such funds which brings financial rewards and new clients, or bolstering of reputations as seen in members garnering investor awards by Morningstar or Euro Magazine, or invitations to elite conferences such as the Annual Meetings of the International Monetary Fund and World Bank. The implications of these findings are two-fold. On the one hand, the ECB is able to gauge some kind of consensus and likely reactions to its policies aiding their own sensemaking processes. The ECB then uses these cues to surprise market expectations with an even larger monetary amount for its PSPP policy in 2015 or launch of the unexpected CSPP in 2016. On the other hand, the ECB is inadvertently influenced by other market actors in the actual implementation and content of its policies. While these networks can also be seen from an institutional perspective, i.e. the main actors are organisations such as the ECB or Blackrock, the findings underline that key decision makers transcend the institutions they represent and are able to move freely between these institutions for beneficial employment and career advancements. Thus, members are able to implement different strategies, discussed in chapter six and seven, such as relative value strategies (short Italy and long the core, Austria, Germany, Belgium), that are antithetical to the assumptions of the portfolio rebalancing effect or of portfolio diversification. In the example of chapter four, ECB experts may allocate towards eligible assets and recycle primary deals to the ECB. Other ECB experts yet again may move from an employment at the ECB to the sell side or to hedge funds to exploit the tacit knowledge gained. The process of how decisions are made in circles of the European financial elite is complex. The findings of chapter six have both shown that the ECB itself can utilise the consensus of the BMCG to increase the impact of a policy by ‘beating consensus’ with a higher than expected monetary stimulus, but also AMC members of the BMCG can utilise the analytical input for better asset allocation decisions. Ties in social networks hence undermine the 206 rational assumptions of the policy and at the same time challenge the democratisation of the project. The implication of this is that the corporate interlocking networks are an empirical expression of the social networks of leaders within these organisations. As opposed to the literature in economic sociology, where co-optation has been seen as more deterministic and from the regulator’s viewpoint, co-optation happens both by the ECB trying to reduce the uncertainty of reactions to its policies, but also by members from important AMCs shaping decision-making of the ECB, such as in co-presentations and face-to-face meetings. BMCG members co-opt the ECB itself. The importance of studying these dynamics further become evident when considering the impact of quantitative easing on the housing market, on pension plans, on retirement income, interest income and so on. Regular inclusion of other interest groups in BMCG meetings could serve the democratisation process well. This struggle for material interest is an important finding. It shows the complexity of social networks and that the assumptions on which quantitative easing is based are challenged by this very dynamic. The findings underline that the multiplicity of frames and multiplexity of ties in bond-investor networks result in unanticipated reactions by bond investors. Networks are dynamic and re-shaped through interactions in multiplex relationships. Indeed, Burt stipulates that purposeful agents in networks are only able to extract and interpret contrasting ideas and find innovations by active engagement in divergent expert networks. Here the organisation literature becomes relevant in which the enacted environment is shaped by occasions of sensemaking in the social networks of experts. The application of investor frames in the analysis revealed that each situation in financial markets embeds a multiplicity of frames. Further, multiplex relations in social networks of investors yield more multifaceted and intense interactions, attending the same conferences and meetings, sharing information on investments and exchanging philosophical views on the investment universe. The multiplexity of ties in these expert networks enable the individuals to transcend company affiliation illustrated in the career moves of BMCG members between the ECB, the sell-side, hedge funds and traditional fund houses. The question about home bias, the tendency to have economic connections with market participants that are geographically close is a central tenet of homophily in investor networks. Networks can form between issuer (governments) and investors (AMCs), but also between multiple AMCs, particularly when investing abroad. Such network characteristics supersede the assumptions of the rational investor and the findings of homophily and home bias in 207 geographical blocs in the eurozone have shown this. In chapter five it is demonstrated that the concept of homophily in sociology and home bias in behavioural finance is closely related in the tendency for market participants to be more likely to establish links between counterparties that are geographically proximate. The findings underline the relationship between the investor, the purchaser of bonds, and the local government, the issuer of bonds as home bias. For the ECB to break up these relationships has proven difficult as these socio- cultural ties override any price differentials and rational assumptions an economist would make. The connection between the two concepts of homophily and home bias has not yet been fully explored and has so far only been loosely sketched (Lin & Viswanathan, 2016). Given its importance, more research is required into home bias in the eurozone. The network findings suggest that the financial networks of the CSPP are becoming comparatively dense and nodes are increasingly resembling the ECB’s portfolio. This indicates a certain imitation effect, which is an unintended policy outcome. At the same time, the networks of the PSPP underline the home bias among nodes in the network, particularly amongst the nodes based in the Southern European bloc. The PSPP has over time led to more consolidated and dense networks that amplify the geographical segmentation of the government bond markets in the eurozone. In following Krackhardt’s (1988) assumption that crisis episodes require more external (outgroup) connections and assessing the ECB’s policy objective to stabilise peripheral markets such as Spain and Italy, I concluded that the PSPP has not made the government bond markets more resilient to financial contagion. In order to overcome the uncertainties of unprecedented monetary policy, PMs are led to imitate the ECB’s investment action and peripheral markets remain reliant on local nodes. The fact that investors do not rebalance and not move outside of their geographical habitat, also challenges the assumption that the economic theory applied by central bankers is performative. This is an interesting case to show the nuances of central bank interactions with financial market participants and highlights the need to allow for a multiplicity of frames when studying financial markets. Market participants follow their own investor frames and also have input, in turn, to the decision-making processes at the ECB through participation in the BMCG. Further cultural factors influence decision-making in the networks of investors in the eurozone. This adds granular analysis to the more complex relationship between the ECB and significant market actors and underlines the multi-directional nature of this relationship. It also leads to the conclusion that investment decision-making relies on divergent kinds of 208 calculation by heterogenous market actors. Furthermore, as chapter two has shown, the power dynamics in the investment chain that the social studies of finance laid out are changing and to account for these changes, network analysis can be used to isolate and analyse the most influential market actors. The thesis explored the ways in which central banks act inter-organisationally by interacting with experts from industry, buy- and sell-side, and in which such occasions of sensemaking are used as input for both the shaping of monetary policy but also the implementation aspect. Interestingly, this phenomenon can also be observed outside of Europe, with the recent application of ETF purchases by the US Fed. The complexity of the relationship between quasi-public institutions, such as central banks, and providers of the instruments that are purchased in quantitative easing programs by institutions such as Blackrock, is seen in the establishment of a separate entity, the ‘Financial Markets Advisory Group’ within Blackrock and is an example of yet another expert group of individuals that fills structural holes. By focusing on co-investor networks and more longer lasting ties, rather than on the shorter term networks of market exchanges or the investment chain as in the previous social studies of finance literature, the thesis calls attention to the notion of financial markets as social networks in which market participants act with knowledge of each other, as opposed to the notion of the anonymous market or the ‘market exchange between strangers’. The introduction of a novel monetary policy such as the PSPP and CSPP could be portrayed as in other sociological literature such as the network studies of how innovations spread. Here, sensemaking in expert networks is a useful analytical tool through which social scientists may analyse decision-making processes in the asset management industry. Sensemaking offers ways in which uncertainty and ambiguities can be reduced for participants, while at the same time, leaving open different interpretations and investment actions emerging out of these situations. Furthermore, the increasing use of ‘unstructured data’ in investment research prevents the PM from describing the ways in which her decisions are made in a reductionist way. PMs seek extra analytical input from different contacts, and put these in the context of the different social positions of the person providing such information, the past voting histories or philosophical viewpoints. In the example of European bond markets, these constitute ECB experts or individuals with access to the elite networks. As an illustration, in an ideal world, 209 interview respondent AM7b will just purchase bonds that are trading at a >3% spread over bunds with low credit risk or AM6 would implement a relative value trade in short Italy and long Belgium and Austria as response to the launch of QE. Likewise, interview respondent HF1a would just like to buy securities where the company has a net cash position on the balance sheet, is trading at significant discount on all valuation metrics and is experiencing a significant earnings turnaround. However, the PM is facing a dynamic and uncertain environment and portfolios built with such ideal-type positions are difficult to implement. That is the reason why investors find themselves in networks with other investors with differing frames. The ECB made clear that they would like to see investors move into Italy and through their policies make the risk-free core unattractive with negative yields. However, PMs adapt to uncertain environments and adjust their investor frames through sensemaking processes. As a result, multifaceted strategies are pursued and frames are adapted, some PMs may go long Austria, Germany and Belgium and short Italy as a relative value trade, given the heightened funding needs of the Italian government and supply shortage of government bonds from the core eurozone. Others may follow the ECB’s buying behaviour and allow style-drift, as one BMCG member mentioned in response to the unusual interest rate environment, “people are getting more comfortable with buying negative coupon bonds”. A multiplicity of frames exists in these bond investor and co-holder networks. Amid unprecedented market contexts and through occasions of sensemaking, PMs justify and rationalise previously ambiguous positions and investing rationales and adjust frames. 8.1.2 Summary and conclusions Chapter one has outlined the theoretical framework to the thesis. It thereby introduced an original angle to the literature, as elaborated above. Chapter two has shown the connection between sensemaking processes and investor frames in light of the changing regulatory environment of the asset management industry. It showed that the previous position of investment banks trying to influence investment decisions of PMs as so-called frame-makers has shifted to a more neutral position of information gatherers and data providers. PMs select and put data points into the investor frame to confirm, shape or alter the strategy. Market participants are not constantly self-reflective and self-referential. PMs will pause and reflect when meeting investors, company management, analysts, when writing monthly newsletters 210 or at internal meetings with the investment committee. This rhythm to frame adjustment is thus not a constant process but involves deliberate occasions of sensemaking. In order to examine the viability and success of the ECB’s monetary policy and its underlying assumptions, I analysed the social structure and behavioural patterns of the financial networks the ECB is operating in. To do this, I formulated a holdings-based model of financial networks in focusing on the ties between co-holders rather than ties between buyers and sellers of a given security. Two explicit policy objectives were examined: the portfolio rebalancing effect and international diversification through cross-border investments. What I introduced as holdings-based network analysis is a useful tool to examine the material interest in financial markets from an institutional angle, but the thesis research incorporated the analysis of social networks that exist within the corporate interlocking networks of financial institutions. The social networks of senior experts in the industry often supersede the financial networks in which firms are tied to other firms in various ways and underlines the embeddedness argument of the economic sociological literature. This was shown by analysing the adaptation processes of PMs and their investor frames and the influence of ECB experts on investment decision-making. Likewise, sensemaking is a useful concept to study the investment decision-making in financial markets. The thesis has shown that given the highly autonomous nature of the profession, PMs require past socialisations, occasions of sensemaking, idiosyncratic research tools and analytical inputs from divergent expert networks for their investor frames to be successful. Like Starbuck and Miliken mention, Stock prices plummeted today. Is this an exceptional opportunity to buy underpriced stocks that will rebound tomorrow, or does today’s drop portend further declines tomorrow? To understand today’s prices, one needs a forecast of tomorrow’s, and this forecast derives from ones past experiences” (Starbuck & Milliken, 1988, p. 40). Therefore, the thesis challenged the notion that it can be known what information is material for asset price changes and that ECB expert networks rely on tacit understanding. While the ECB continues to make attempts to disseminate information ‘equally’, inadvertently, the non- public communications become more important in sensemaking processes, both for ECB 211 experts in the BMCG, but also to PMs following the ECB closely. Central actors in these expert networks are able to react and adapt to new policies quicker. Explicating a decision or trading strategy through the application of the investor frame binds the investor to it in future contexts. The example of the BMCG underlined that collaborating on views on a regular basis, communicating views from the BMCG to their own colleagues and to peers in investor meetings binds the individual to certain courses of action and trading strategies. The regularity of meetings and the time invested into preparation, collaboration and physical presence creates strong ties. The fact that these meetings occur at least every three months and that continued membership is dependent upon the level of engagement facilitates the multiplexity of ties. Given the seniority of BMCG members and the confidentiality of the information exchange, the BMCG member relationships supersede internal relationships with analysts and employees supporting the PMs in their companies. The reputational and informational value of these relationships thus resemble what Burt (1992, 2007, 2010) referred to as information arbitrage. To an extent, ECB experts become dis-embedded from their local company connections and spend more time travelling and meeting other ‘elites’. As a result, the BMCG membership becomes a matter not of company affiliation but based on personal circumstances and the connections formed outside the company. This was also shown in the BMCG affiliation network analysis in chapter six where institutions were not guaranteed a membership when an important individual left the firm (see section 6.3.2). In chapter two, I discussed the changing organisation of the asset management industry and the implications this has for the decision-making processes of PMs. Historically, coverage of the rising influence of central banks in financial markets in the thesis serves as another case study of the continuing decline of the power of investment banks and is emblematic of other economic sociology studies, which highlighted the changes in business models and competition in the industry (Baker, 1990; Mintz & Schwartz, 1985; Podolny & Phillips, 1996). Democratisation of financial markets is high on the agenda of both the European Project and recent regulatory efforts. The regulatory measures of MiFID II are to an extent breaking up the co-operative action in financial markets and with it the social networks surrounding this. While occasions of sensemaking are becoming increasingly controlled by the PMs in dealing with financial intermediaries, at the heart of the ECB, new expert networks in its contact groups are formed and maintained which challenge this move towards 212 democratisation. These expert networks are a vivid illustration of how the ECB forms social networks to both test and diffuse policy ideas and the findings also underline that these experts’ predictions have shown limitations and a diversification of these groups in terms of its membership may be fruitful. It is not surprising that central banks are coming under increasing criticism of ‘being in bed’ with the market and that trust in the institution of the ECB is ambivalent. According to Eurobarometer in June 2019, surveyed citizens across Europe distrust the ECB to a tune of 51% in Spain, 71% in Greece 48% in Italy and 40% in Germany. To open these meetings to a more diverse set of market actors may counteract this concern. This leads to similar conclusions to those Annelise Riles (2018) has spoken about in the US, namely that central bankers should seek a more rigorous engagement with different stakeholders in society to overcome the culture clash between central bankers and their critics. Chapter four has shown that central nodes in financial network are often imitated. This reverts back to the fact that when individuals face uncertainty, discerning decision makers are adjusting their own processes to emulate the ‘leader’. Thus, the entrance of the ECB into corporate bond markets needs to be conducted in a reflexive manner, insofar as the ECB’s own interactions in networks of corporate bond holders have unintended consequences for issuers and holders and are difficult to predict. The paucity of active AMCs and the lack of diversity of market actors in the CSPP network causes some concerns. The fact that the most central nodes in the European corporate bond market are passive investment vehicles, mainly ETFs, which are not sensitive to the fundamental valuations of securities that are purchased, could potentially lead to fire sales and spread contagion throughout the network in an event of operational breakdown of central nodes. Should a central node experience redemptions in their investment vehicles, indiscriminate selling could cause such contagion. As AM1 put it, “it’s not humans making such sell decisions but when funds are withdrawn, the computer sells” (AM1 Interview). The risk of ETFs in European capital markets has received increased attention and the findings of chapter four add depth to the conclusions of the Advisory Scientific Council in their study of systemic risk and ETFs: “Given the high levels of concentration in the ETF market, a large event leading to the materialisation of operational risks in one of the providers may generate massive fire sales of ETFs, resulting in large price movements of their constituent securities … or of the ETFs themselves” (Pagano, Serrano, & Zechner, 2019, p. 33). 213 Chapter five outlined the connection between homophily and home bias across the disciplines of sociology and behavioural finance. It was shown that there is still a significant home bias in the holders of government debt in the eurozone, particularly those in the Southern bloc. This underlines the lack of risk sharing across different countries and blocs of the eurozone. However, with the establishment of the eurozone and its monetary policy, problems in Greece or Italy directly affect individual nodes in Germany or the Netherlands and vice versa. As nodes in the network forge ties between market participants across national boundaries, risks emanating from a specific country quickly spread through the region. Thus, the European project requires a rethink of national boundaries and approaches to, at a minimum, regional networks. Chapter five also showed that from a network perspective, the PSPP has not helped to strengthen and diversify the holder base of Southern European sovereign bonds. In his reflections on the eurozone crisis, former Irish central bank governor Honohan (2019, pp. 348–349) highlighted the need for increased integration and risk-sharing as the European project comes under mounting threat and that the central bankers should work more on their cooperation with governments more. Including different stake holders in the processes of policy formation should mitigate the policy’s unintended consequences. The thesis findings also encourage a more vigorous engagement with the economic sociology of networks literature for financial and behavioural economists. It is implicit, that the work of the new economic sociology tradition of Granovetter, Burt, Podolny and Mizruchi offers theoretical and methodological approaches that are highly relevant to the research published in financial journals on herding and social learning (e.g. Clement & Tse, 2005; Sias, 2004), in behavioural finance and the study of home bias (Coval & Moskowitz, 1999, 2001; Grinblatt & Keloharju, 2001) and in management research of executive ties and board appointments (Cohen et al., 2008, 2012; Fracassi, 2016). Particularly, it links factors that determine informational advantages to positive portfolio returns in the area of behavioural finance. The potential of the approaches used in the thesis for further research is vast both in terms of methods and across academic fields. In addressing the theoretical contributions to the study of corporate interlocks, the thesis offered an approach that looks at the social attributes of actors in interlocking networks, which in turn, influence investment decision-making. Some of the literature in the social studies of finance, discussed in chapter one, does not analyse both the social structure and agency of nodes in financial networks but defines and uncovers the influence of central banks on financial markets from an institutional perspective. By 214 utilising corporate interlocks and focusing on sensemaking processes of the ‘decision elites’, the thesis has offered a novel approach to the study of central bank influence. When considering the way in which the findings underline the impact individual decision- making in elite financial networks has on the overall financial market and how the ECB as central actor influences the financial networks it enters, it also becomes clear how it could institute positive change through its investing activity. More recently, for instance, the ECB is eyeing green investing and green central banking, promising to incorporate environmental concerns into its decision-making. This development would indicate a clear shift away from the previously neutral investment mandate and could solidify a more-broad based approval of its policies across society. 215 Appendix A – Formal Interview Participant Demographics Note: Regions represent the location of place of work for respondents. Years of Industry Experience is put into date ranges to protect anonymity. Appendix B – Glossary of Terms Bear market – a market characterised by falling prices. Brexit – Britain exiting the European Union. Broker – the investment services arm of an investment bank. Responsibilities can include investment research services, sales service, trade execution, provision of leverage, Over-The Counter options, swaps, IPOs. BTPs – Italian government bonds. Bull market – a market characterised by rising prices. Bunds – German Government Bonds. Buy-side – Generally, the asset management industry/institutional investors. Buy-side analyst – investment professional tasked with the analysis of securities, mostly broken down by sector or country. Usually reports to the PM. Capital Asset Pricing Model (CAPM) – a common model to price an asset. Alternative methods are Discounted Cash Flow or Earnings Retention Model. Chief Investment Officer (CIO) – an investment professional typically responsible with oversight of more than one investment fund or product and leading the general asset allocation for the group. Corporate bonds – bonds issued by companies in the private sector, oftentimes companies listed on the stock exchange as well. Cost of equity – the required rate of return an investor should be compensated for to hold the shares of a company. There are mainly two methods to calculate this based on either the Price to Earnings of the shares or the Beta relative to the market or country index (CAPM). Credit Rating Agencies – institutions tasked with attributing credit ratings for bonds, mostly corporate or sovereign bonds. Examples are Fitch or S&P. Daily runs – listings/messages with bid and offer prices from brokers to clients and at times other brokers, giving an indication in which issues the broker has a market for the day. These are also used by PMs to calibrate bond inventories at brokers. 216 Deflation – the year on year decline in the price of a basket of goods. Discounted Cash Flow (DCF) model – model to ascertain the Net Present Value (NPV) of future cash flows. Equities – the asset class of stocks and its derivatives (e.g. options, futures, warrants). Exchange Traded Fund (ETF) – a stock-like instrument that reproduces a select number of single securities, such as an equity index, and enable intraday trading. ETFs are primarily known as passive replication strategies, thematic stock baskets, factor investing. Fixed Income – the asset class consisting of bonds and its derivatives. Forward Guidance – The practice of a central bank commenting and evaluating the current economic situation and giving guidance on likely courses of policy action. The central tenet of forward guidance is greater transparency, the assumption that markets work better and that policies are reinforced with forward guidance. FX Trading – Foreign Exchange/Currency trading. Hedge Fund – an investment fund with the ability to take leverage and with looser supervision and monitoring than traditional mutual funds. While the concept of the hedge fund is changing, common characteristics of hedge funds are short selling strategies, use of leverage, absolute return benchmarks, discretionary investing, differing time horizons. High Yield Bonds or High Yield – non-investment grade bonds usually with higher yields to compensate for the increased default risk. Inflation – the year on year growth in the price of a basket of goods. Institutional investors – an institution/company investing funds on behalf of others in capital or private markets. Interest Cover – Operating Income divided by the Interest expense of a company. A useful measure to assess the ability of a company to repay its debt. Investment Bank – An entity, either as boutique or part of a larger universal bank, with the general operations of market making in different asset classes, bond and equity issuance, investment advice and research services to institutional investors. Investment Grade bonds (IGs) – bonds from issuers that received a threshold rating from credit rating agencies such as Fitch or S&P. Material information – information that a ‘reasonable investor’ would consider important. Usually information that would impact the market price of a security. Material non-public information – material information that is not known to the public. Portfolio Manager (PM) – an investment professional tasked with managing or co- managing a portfolio of securities or funds. Responsibilities vary and can include, security selection, asset allocation, supervision of analysts, communication with investors, communication with the sell-side, trade execution, monthly reporting. Price to Earnings Ratio (PE) – A ratio of the price of a share divided by the earnings per share, or alternatively the market capitalisation divided by the net income of a company. It yields a value that can be used either as comparison to the sector average, other companies, relative to its own history or relative to its growth rate. Proprietary trading – trading and taking risk with the firm’s own internal capital, such as from a bank. Purchasing Managers’ Index (PMI) – An index measuring purchasing managers’ views in the manufacturing and servicing industry on whether they are expanding, stable or contracting. Quantitative Easing – Expansion of the central bank balance sheet via large scale purchases of assets, for example Government bonds, Asset-Backed Securities, Corporate Bonds, Equities. Rates Trading – trading instruments related to interest rates, mostly sovereigns. Risk appetite – the willingness to take risk. 217 Sales trader – execution only trader at a broker. Sell-side – investment banks, boutiques and independent investment research services catering to investors, mostly institutional investors. Sell-side analysts – Investment professional working in investment research services catering to investors and usually responsible for a group or sector of companies. Typical responsibilities include financial modelling, industry research, report writing, marketing ideas to clients via one-on-one meetings or conference calls for companies under coverage. These can include bonds or stocks. Spread tightening – the yield of a bond falling compared to a benchmark yield, e.g. when the BTP spread over German Bunds is falling, indicating rising appetite for riskier assets. Taper – to reduce the monthly amount of asset purchases, reversal of quantitative easing. Taper Tantrum – the fallout in financial markets on the announcement of the future reduction of monthly asset purchases by the Fed in 2013. Weighted average cost of capital – proportional percentage cost of total capital, consisting of equity and debt. 218 References Abolafia, M. Y. (1996). Making markets : opportunism and restraint on Wall Street. Cambridge, MA: Harvard University Press. Abolafia, M. Y. (2004). Framing Moves: Interpretive Politics at the Federal Reserve. Journal of Public Administration Research and Theory, 14, 349–370. https://doi.org/https://doi.org/10.1093/jopart/muh023 Abolafia, M. Y. (2010). Narrative Construction as Sensemaking: How a Central Bank Thinks. Organization Studies, 31(3), 349–367. https://doi.org/10.1177/0170840609357380 Abolafia, M. Y. (2020). Stewards of the Market : How the Federal Reserve Made Sense of the Financial Crisis. Cambridge, MA: Harvard University Press. Abolafia, M. Y., & Kilduff, M. (1988). Enacting Market Crisis: The Social Construction of a Speculative Bubble. Administrative Science Quarterly, 33(2), 177. https://doi.org/10.2307/2393054 Acton, M., & Brunsden, J. (2017, November 7). The Hanseatic League 2.0. Financial Times. Retrieved from https://www.ft.com/content/83e0427c-c399-11e7-a1d2-6786f39ef675 Allen, F., Morris, S., & Shin, H. S. (2006). Beauty Contests and Iterated Expectations in Asset Markets. Review of Financial Studies, 19(3), 719–752. https://doi.org/10.1093/rfs/hhj036 Alloway, T., & King, M. (2017, November 27). Why Historic Relationships in Markets Have Been Totally Upended. Odd Lots [Audio Podcast]. Retrieved from https://www.bloomberg.com/news/articles/2017-11-23/odd-lots-why-historic- relationships-in-markets-have-been-totally-upended Alloway, T., & Mackenzie, M. (2014, February 17). Goldman retreats from bond platform. Financial Times. Retrieved from https://www.ft.com/content/15c2e47c-9588-11e3- 8371-00144feab7de Arjaliès, D.-L., Grant, P., Hardie, I., Svetlova, E., & MacKenzie, D. (2017). Chains of finance how investment management is shaped. Oxford, UK: Oxford University Press. Retrieved from https://ebookcentral.proquest.com/lib/liberty/detail.action?docID=4877961 Asmussen, J. (2012). Building trust in a world of unknown unknowns: central bank communication between markets and politics in the crisis. In Speech by Jörg Asmussen, 219 Member of the Executive Board of the ECB, European Communication Summit 2012. Frankfurt am Main: European Central Bank. Retrieved from https://www.ecb.europa.eu/press/key/date/2012/html/sp120706.en.html Axilrod, S. H. (2009). Inside the Fed : Monetary Policy and its Management, Martin through Greenspan to Bernanke. Cambridge, MA: MIT Press. Baerveldt, C., Van Duijn, M. A. J., Vermeij, L., & Van Hemert, D. A. (2004). Ethnic boundaries and personal choice. Assessing the influence of individual inclinations to choose intra-ethnic relationships on pupils’ networks. Social Networks, 26, 55–74. https://doi.org/10.1016/j.socnet.2004.01.003 Bagehot, W. (1873). Lombard Street: A description of the money market (4th ed.). London: Henry S. King & Co. Baker, W. (1984). The Social Structure of a National Securities Market. American Journal of Sociology, 89(4), 775–811. https://doi.org/10.1086/227944 Baker, W. (1990). Market Networks and Corporate Behavior. American Journal of Sociology, 96(3), 589–625. https://doi.org/10.1086/229573 Baker, W., & Iyer, A. (1992). Information networks and market behavior. The Journal of Mathematical Sociology, 16(4), 305–332. https://doi.org/10.1080/0022250X.1992.9990093 Balchunas, E. (2016). The institutional ETF toolbox : how institutions can understand and utilize the fast-growing world of ETFs (1st.). Hoboken, New Jersey : John Wiley & Sons, Inc. Banerjee, A. (1992). A simple model of herd behavior. The Quarterly Journal of Economics, 108(3), 797. https://doi.org/10.2307/2118364 Banerjee, A., Chandrasekhar, A. G., Duflo, E., & Jackson, M. O. (2013). The Diffusion of Microfinance. Science, 341(6144), 363–370. Barberis, N., Shleifer, A., & Vishny, R. (1998). A model of investor sentiment. Journal of Financial Economics, 49(3), 307–343. https://doi.org/10.1016/S0304-405X(98)00027-0 Barker, R., Hendry, J., Roberts, J., & Sanderson, P. (2012). Can company-fund manager meetings convey informational benefits? Exploring the rationalisation of equity investment decision making by UK fund managers. Accounting, Organizations and Society, 37(4), 207–222. https://doi.org/10.1016/J.AOS.2012.02.004 Barro, R. J., & Mccleary, R. M. (2003). Religion and Economic Growth across Countries. American Sociological Review, 68(5), 760–781. https://doi.org/10.2307/1519761 Bastian, M., Heymann, S., & Jacomy, M. (2009). Gephi: An Open Source Software for 220 Exploring and Manipulating Networks. International AAAI Conference on Web and Social Media; Third International AAAI Conference on Weblogs and Social Media. Beckert, J. (2016). Imagined Futures: Fictional Expectations and Capitalist Dynamics. Cambridge, MA: Harvard University Press. Beckert, J., & Bronk, R. (2018). Uncertain Futures: Imaginaries, Narratives, and Calculation in the Economy. Oxford, U.K.: Oxford University Press. Beunza, D., & Garud, R. (2007). Calculators, Lemmings or Frame-Makers? The Intermediary Role of Securities Analysts. The Sociological Review, 55, 13–39. https://doi.org/10.1111/j.1467-954X.2007.00728.x Beunza, D., & Stark, D. (2004). Tools of the trade: The socio-technology of arbitrage in a Wall Street trading room. Industrial and Corporate Change, 13(2), 369–400. https://doi.org/10.1093/icc/dth015 Blackrock. (2018). Blackrock Investor Day 2018. Retrieved from https://s24.q4cdn.com/856567660/files/doc_presentations/1500122082.pdf Blau, P. M. (1977). Inequality and heterogeneity : a primitive theory of social structure. New York: Free Press, Collier MacMillan. Blau, P. M. (1982). Structural Sociology and Network Analysis: An Overview. In P. V. Marsden & N. Lin (Eds.), Social Structure and Network Analysis (pp. 273–280). Beverly Hills, CA: Sage Publications. Blinder, A. S. (1998). Central banking in theory and practice. Cambridge, MA: MIT Press. Blinder, A. S. (2004). The quiet revolution : central banking goes modern. New Haven, Conn.: Yale Univ. Press. Blinder, A. S., Ehrmann, M., & Fratzscher, M. (2008). Central Bank communication and monetary policy : a survey of theory and evidence. Journal of Economic Literature, 46(04), 910–945. https://doi.org/10.3386/w13932 Bloomberg Professional Service. (2017). Bloomberg Barclays Index Methodology. New York. Retrieved from www.bloombergindices.com Bonacich, P. (1987). Power and Centrality: A Family of Measures. American Journal of Sociology, 92(5), 1170–1182. https://doi.org/10.1086/228631 Bond Market Contact Group. (2015). Summary of Discussion, Tuesday 27 January 2015. Frankfurt am Main. Retrieved from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/150127/2015-01-27-summary.pdf Borgatti, S. P., & Everett, M. G. (1997). Network analysis of 2-mode data. Social Networks, 19(3), 243–269. https://doi.org/https://doi.org/10.1016/S0378-8733(96)00301-2 221 Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). Ucinet for Windows: Software for Social Network Analysis. Analytic Technologies. Retrieved from citeulike-article- id:6031268 Borgatti, S. P., Everett, M. G., & Johnson, J. C. (2013). Analyzing social networks. Thousand Oaks, CA: Sage Publications. Borgatti, S. P., & Feld, S. L. (1994). How To Test the Strength of Weak Ties Theory. Connections, 17(1), 45–46. Retrieved from http://www.analytictech.com/mb874/Papers/test_swt_theory.htm Brandes, U., Borgatti, S. P., & Freeman, L. C. (2016). Maintaining the duality of closeness and betweenness centrality. Social Networks, 44, 153–159. https://doi.org/10.1016/J.SOCNET.2015.08.003 Brauer, M., & Wiersema, M. (2018). Analyzing Analyst Research: A Review of Past Coverage and Recommendations for Future Research. Journal of Management, 44(1), 218–248. https://doi.org/10.1177/0149206317734900 Braun, B. (2015). Governing the future: the European Central Bank’s expectation management during the Great Moderation. Economy and Society, 44(3), 367–391. https://doi.org/https://doi.org/10.1080/03085147.2015.1049447 Braun, B. (2018). Central banking and the infrastructural power of finance: the case of ECB support for repo and securitization markets. Socio-Economic Review, 1–24. https://doi.org/10.1093/ser/mwy008 Braun, B., Gabor, D., & Hübner, M. (2018). Governing through financial markets: Towards a critical political economy of Capital Markets Union. Competition & Change, 22(2), 101–116. https://doi.org/10.1177/1024529418759476 Brida, J. G., Matesanz, D., & Seijas, M. N. (2016). Network analysis of returns and volume trading in stock markets: The Euro Stoxx case. Physica A: Statistical Mechanics and Its Applications, 444, 751–764. https://doi.org/10.1016/J.PHYSA.2015.10.078 Brivot, M., Himick, D., & Martinez, D. (2017). Constructing, Contesting, and Overloading: A Study of Risk Management Framing. European Accounting Review, 26(4), 703–728. https://doi.org/10.1080/09638180.2016.1180254 Brown, A. D., Colville, I., & Pye, A. (2015). Making Sense of Sensemaking in Organization Studies. Organization Studies, 36(2), 265–277. https://doi.org/10.1177/0170840614559259 Buckland, R., Genc, M. C., Manthey, B., Recchia, C., Stubss, J., & Tambi, A. (2016). Global Equity Strategist. Monetary Madness: An Equity Perspective. London. Retrieved from 222 https://www.citivelocity.com/rendition/eppublic/uiservices/open_watermark_pdf?req_dt =cGRmTGluaz1odHRwcyUzQSUyRiUyRmlyLmNpdGkuY29tJTJGOFV0NWNubVRP aEpseTROTHJpM3ZZMzdIS21Ma1ZkdTNnUlVWSEdma0pKbGZIWTg3TmRHT3NR JTI1M0QlMjUzRCZ1c2VyX2lkPW1nMzc3NjYmdXNlcl90eXBlPU Burne, K., & Turner, M. (2015, May 8). Creator of ‘GSessions’ Bond Platform Leaves Goldman Sachs. Retrieved October 30, 2018, from https://www.wsj.com/articles/creator-of-gsessions-bond-platform-leaves-goldman- sachs-1431124065 Burt, R. S. (1982). Toward a Structural Theory of Action. New York: Academic Press. https://doi.org/10.1016/c2009-0-21953-2 Burt, R. S. (1983). Corporate Profits and Cooptation: Networks of Market Constraints and Directorate Ties in the American Economy. Cambridge, MA: Academic Press Inc. Burt, R. S. (1992). Structural holes : the social structure of competition. Cambridge, MA: Harvard University Press. Burt, R. S. (2000). Decay functions. Social Networks, 22(1), 1–28. https://doi.org/10.1016/S0378-8733(99)00015-5 Burt, R. S. (2007). Secondhand Brokerage: Evidence on the Importance of Local Structure for Managers. Source: The Academy of Management Journal, 50(1), 119–148. https://doi.org/10.2307/20159844 Burt, R. S. (2010). Neighbor Networks. Competitive Advantage Local and Personal. Oxford, U.K.: Oxford University Press. Callon, M. (1998a). An Essay on Framing and Overflowing: Economic Externalities Revisited by Sociology. The Sociological Review, 46, 244–269. https://doi.org/10.1111/j.1467-954X.1998.tb03477.x Callon, M. (1998b). The Laws of the markets. Oxford, U.K.: Blackwell. Callon, M. (2007). What does it mean to say that economics is performative? In D. Mackenzie, F. Muniesa, & L. Siu (Eds.), Do Economists Make Markets?: On the Performativity of Economics (pp. 311–357). Princeton, N.J.: Princeton University Press. Castillo, S., & Varela, C. (2017). Target2 imbalances are rising, should we worry? BBVA Research. Retrieved from https://www.bbvaresearch.com/wp- content/uploads/2017/01/Target2_20170130VF.pdf Chambers, A. (2010, May 7). European banks shaken by Greek debt exposure. Reuters Business News. Retrieved from https://uk.reuters.com/article/uk-greece-banks/european- banks-shaken-by-greek-debt-exposure-idUKTRE6463KE20100507 223 Chang, C. (2009). Herding and the role of foreign institutions in emerging equity markets. Pacific-Basin Finance Journal, 18, 175–185. https://doi.org/10.1016/j.pacfin.2009.11.001 Chong, K., & Tuckett, D. (2015). Constructing conviction through action and narrative: how money managers manage uncertainty and the consequence for financial market functioning. Socio-Economic Review, 13(2), 309–330. https://doi.org/10.1093/ser/mwu020 Chu, J. S. G., & Davis, G. F. (2016). Who killed the inner circle? The decline of the American corporate interlock network. American Journal of Sociology, 122(3), 714– 754. https://doi.org/10.1086/688650 Citi Research. (2016). Why € credit increasingly needs more than ECB buying. London. Clement, M. B., & Tse, S. Y. (2005). Financial Analyst Characteristics and Herding Behavior in Forecasting. The Journal of Finance, 60(1), 307–341. https://doi.org/10.1111/j.1540- 6261.2005.00731.x Coenen, G., Ehrmann, M., Gaballo, G., Hoffmann, P., Nakov, A., Nardelli, S., … Strasser, G. (2017). Communication of monetary policy in unconventional times. CFS working paper series ; No. 578. Frankfurt am Main: Center for Financial Studies. Retrieved from https://ssrn.com/abstract=3043098 Cohen, L., Frazzini, A., & Malloy, C. (2008). The Small World of Investing: Board Connections and Mutual Fund Returns. Source: Journal of Political Economy, 116(5), 951–979. https://doi.org/10.1086/592415 Cohen, L., Frazzini, A., & Malloy, C. J. (2012). Hiring Cheerleaders: Board Appointments of “Independent” Directors. Source: Management Science, 58(6), 1039–1058. https://doi.org/10.1287/nmsc.1110.1483 Coleman, J., Katz, E., & Menzel, H. (1957). The Diffusion of an Innovation Among Physicians. Sociometry, 20(4), 253. https://doi.org/10.2307/2785979 Constâncio, V. (2014). Reflections on financial integration and stability. In Financial Integration and Stability in a New Financial Architecture (p. 6). Frankfurt am Main: Bank of International Settlements. Retrieved from https://www.bis.org/review/r140428b.htm Constâncio, V. (2018). Why EMU requires more financial integration. In Joint conference of the European Commission and European Central Bank (p. 9). Frankfurt am Main: Bank of International Settlements. Retrieved from https://www.bis.org/review/r180507a.htm#footnote-2 224 Cornelissen, J. P., & Werner, M. D. (2014). Putting Framing in Perspective: A Review of Framing and Frame Analysis across the Management and Organizational Literature. The Academy of Management Annals, 8(1), 181–235. https://doi.org/10.1080/19416520.2014.875669 Coval, J. D., & Moskowitz, T. J. (1999). Home Bias at Home: Local Equity Preference in Domestic Portfolios. The Journal of Finance, 54(6), 2045–2073. Retrieved from https://www.jstor.org/stable/797987 Coval, J. D., & Moskowitz, T. J. (2001). The Geography of Investment: Informed Trading and Asset Prices. Journal of Political Economy, 109(4), 811–841. https://doi.org/10.1086/322088 Creswell, J. W., & Creswell, J. D. (2014). Research Design: Qualitative, Quantitative, and Mixed Methods Approaches (4th ed.). Thousand Oaks, CA: Sage Publications. Creswell, J. W., & Plano Clark, V. L. (2017). Designing and Conducting Mixed Methods Research. Thousand Oaks, CA: Sage Publications, lnc. Cristina McEachern Gibbs. (2010). A New FIX for Algo Trading - ProQuest. Retrieved October 29, 2018, from https://search.proquest.com/docview/356687092?rfr_id=info%3Axri%2Fsid%3Aprimo Daft, R. L., & Weick, K. E. (1984). Toward a Model of Organizations as Interpretation Systems. The Academy of Management Review, 9(2), 284. https://doi.org/10.2307/258441 Dauble, J. (2007, September 17). Former Fed Chairman Alan Greenspan speaks extensively to Maria Bartiromo. CNBC News. Retrieved from https://www.cnbc.com/id/20819918 de Zeeuw, H. (2016). CSPP: lessons learned from CBPP3 Which three main lessons can we learn from CBPP3? ABN Amro. Amsterdam. Diessner, S., & Lisi, G. (2019). Masters of the ‘masters of the universe’? Monetary, fiscal and financial dominance in the Eurozone. Socio-Economic Review. https://doi.org/10.1093/SER/MWZ017 Dimitrios, K., & Vasileios, O. (2015). A Network Analysis of the Greek Stock Market. Procedia Economics and Finance, 33, 340–349. https://doi.org/10.1016/S2212- 5671(15)01718-9 Dincer, N. N., & Eichengreen, B. (2007). Central Bank transparency : where, why and with what effects? Cambridge, MA: NBER. Domm, P. (2014). Gross exit ripples bond market. Retrieved November 19, 2018, from www.cnbc.com/2014/09/26/gross-exit-ripples-bond-market.html 225 Draghi, M. (2012). Speech by Mario Draghi, President of the European Central Bank at the Global Investment Conference in London 26 July 2012. London: European Central Bank. Retrieved from https://www.ecb.europa.eu/press/key/date/2012/html/sp120726.en.html Draghi, M. (2014a). Introductory statement to the press conference (with Q&A), 3 April. Frankfurt am Main: European Central Bank. Retrieved from https://www.ecb.europa.eu/press/pressconf/2014/html/is140403.en.html Draghi, M. (2014b). Introductory statement to the press conference (with Q&A), 4 December. Frankfurt am Main: European Central Bank. Retrieved from https://www.ecb.europa.eu/press/pressconf/2014/html/is141204.en.html Draghi, M. (2014c). Introductory statement to the press conference (with Q&A), 7 August. Frankfurt am Main: European Central Bank. Retrieved from https://www.ecb.europa.eu/press/pressconf/2014/html/is140807.en.html Draghi, M. (2014d). Introductory statement to the press conference (with Q&A). Mario Draghi, President of the ECB, Frankfurt am Main, 6 November 2014. Retrieved July 17, 2019, from https://www.ecb.europa.eu/press/pressconf/2014/html/is141106.en.html#qa Draghi, M. (2014e, August 4). Central bank communications. Opinion Piece. Handelsblatt. Retrieved from https://www.ecb.europa.eu/press/inter/date/2014/html/sp140804.en.html Draghi, M. (2015a). Hearing at the European Parliament’s Economic and Monetary Affairs Committee (introductory remarks and Q&A). Brussels, 23 March 2015. Brussels: European Parliament. Retrieved from https://www.ecb.europa.eu/press/key/date/2015/html/sp150323_1.en.html Draghi, M. (2015b). The ECB’s recent monetary policy measures: Effectiveness and challenges, IMF presentation. Washington: IMF. Retrieved from https://www.ecb.europa.eu/press/key/date/2015/html/sp150514.en.html Draghi, M. (2017). Monetary Dialogue With Mario Draghi. Transcript, 29th May 2017. Brussels: European Central Bank. Retrieved from https://www.ecb.europa.eu/pub/pdf/annex/ecb.sp170529_transcript.en.pdf?3e4a59ef28a 7f45d8bc14c1085aecdf7 Draghi, M., Yellen, J., Carney, M., Kuroda, H., & Wessel, D. (2017, November 14). At the heart of policy: challenges and opportunities of central bank communication. Panel Discussion. Retrieved from https://www.youtube.com/watch?v=DI7p-g51O8g Duisenberg, W. F. (2001). The ECB’s monetary policy strategy and the quantitative definition of price stability. Letter of Dr. W. F. Duisenberg, President of the ECB to the 226 Chairperson of the Committee on Economic and Monetary Affairs. 13 December 2001. Frankfurt am Main: European Central Bank. Retrieved from https://www.ecb.europa.eu/press/key/date/2001/html/sp011217.en.html Easley, D., & Kleinberg, J. (2010). Networks, crowds, and markets : reasoning about a highly connected world. New York: Cambridge University Press. Egea, C., & Gilbert, C. (2014). Central Bank Asset Purchase Programs. ECB Bond Market Contact Group. 8 April 2014. Retrieved July 17, 2019, from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/140408/Item_2.pdf?d85d81554233ff 256e6067969891a858 Ehrmann, M., & Fratzscher, M. (2007). Communication by Central Bank Committee Members: Different Strategies, Same Effectiveness? Journal of Money, Credit and Banking, 39(2–3), 509–541. https://doi.org/10.1111/j.0022-2879.2007.00034.x El-Erian, M. A. (2016). The Only Game in Town : Central Banks, Instability, and Avoiding the Next Collapse. New Haven: Yale University Press. Emirbayer, M., & Goodwin, J. (1994). Network Analysis, Culture, and the Problem of Agency. American Journal of Sociology, 99(6), 1411–1454. https://doi.org/10.1086/230450 Erikson, E. (2013). Formalist and Relationalist Theory in Social Network Analysis. Sociological Theory, 31(3), 219–242. https://doi.org/10.1177/0735275113501998 ESMA. (2019). Questions and Answers On MiFID II and MiFIR investor protection and intermediaries topics. Paris. https://doi.org/ESMA35-43-349 Eurekahedge. (2019). The Eurekahedge Report July 2019. London. Retrieved from http://www.eurekahedge.com/Research/News/1921/Hedge-Fund-Report-July-2019 European Central Bank. (2013). Summary of the Discussion. Bond Market Contact Group. Frankfurt, Tuesday 22 January 2013, 13:00 – 16:00. Frankfurt am Main. Retrieved from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/130122/summary.pdf?bb6f54e8b535 cf72430c749c49437567 European Central Bank. (2015a). Ad-Hoc Teleconference, Bond Market Contact Group, Frankfurt am Main, 16th March 2015. Retrieved August 1, 2019, from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/150318/summary_of_teleconference _on_PSPP_implementation.pdf European Central Bank. (2015b). Increase in PSPP issue share limit enlarges purchasable universe (Press Release). Frankfurt am Main: European Central Bank. Retrieved from 227 https://www.ecb.europa.eu/press/pr/date/2015/html/pr151109.en.html European Central Bank. (2015c, June 10). Letter to Emily O’Reilly, European Ombudsman, re: Speech delivered by a member of the ECB executive board on May 18 2015. https://doi.org/https://doi.org/10.1017/bap.2017.6 European Central Bank. (2018). ECB Bond Market Contact Group (BMCG): terms of reference. Retrieved July 18, 2019, from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/BMCG_Terms_of_reference.pdf European Central Bank. (2019a). Corporate Sector Purchase Programme (CSPP) - Q&A. Retrieved July 15, 2019, from https://www.ecb.europa.eu/mopo/implement/omt/html/cspp-qa.en.html European Central Bank. (2019b). Guiding principles for external communication for high- level officials of the European Central Bank. Retrieved July 1, 2019, from https://www.ecb.europa.eu/ecb/orga/transparency/html/eb-communications- guidelines.en.html European Central Bank. (2019c). Public sector purchase programme (PSPP) - Q&A. Retrieved June 10, 2019, from https://www.ecb.europa.eu/mopo/implement/omt/html/pspp-qa.en.html European Commission. (2017). Analysis of European Corporate Bond Markets - Report from the Commission Expert Group on Corporate Bonds, November 2017. Retrieved from https://ec.europa.eu/info/publications/171120-corporate-bonds-report_en European Council. (2014). DIRECTIVE 2014/65/EU OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 15 May 2014 on markets in financial instruments and amending Directive 2002/92/EC and Directive 2011/61/EU. Retrieved from https://eur- lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014L0065&from=EN European Union. (2017). Drivers of Corporate Bond Market Liquidity in the European Union. London. https://doi.org/10.2874/499825 Fama, E. F., & French, K. R. (2010). Luck versus Skill in the Cross-Section of Mutual Fund Returns. The Journal of Finance, 65(5), 1915–1947. https://doi.org/10.1111/j.1540- 6261.2010.01598.x Financial Conduct Authority. (2016). Financial Conduct Authority Asset Management Market Study Interim Report MS15/2.2 Annex 8- Profitability Analysis Interim Report. London. Retrieved from https://www.fca.org.uk/publication/market-studies/ms15-2-2- annex-8.pdf Flick, U. (2014). An Introduction to Qualitative Research (5th ed.). Thousand Oaks, CA: 228 Sage Publications. Fracassi, C. (2016). Corporate Finance Policies and Social Networks. Management Science, 63(8), 2420–2438. https://doi.org/10.1287/mnsc.2016.2433 Freeman, L. C. (1978). Centrality in social networks conceptual clarification. Social Networks, 1(3), 215–239. https://doi.org/10.1016/0378-8733(78)90021-7 French, K. R., & Poterba, J. M. (1991). Investor Diversification and International Equity Markets. The American Economic Review, 81(2), 222–226. https://doi.org/10.3386/w3609 Frieda, G. (2015). What went wrong on October 15th: Perfect storm or global warming? Bond Market Communication Group Presentation. Frankfurt. Retrieved from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/150127/2015-01-27-Item-2- 1.pdf?77408c9d141930a16f384bd20b4bd756 Fruchterman, T. M. J., & Reingold, E. M. (1991). Graph drawing by force-directed placement. Software: Practice and Experience, 21(11), 1129–1164. https://doi.org/10.1002/spe.4380211102 Fuchs, A., & Gehring, K. (2017). The Home Bias in Sovereign Ratings. Journal of the European Economic Association, 15(6), 1386–1423. https://doi.org/10.1093/jeea/jvx009 Gabrieli, S., & Salakhova, D. (2019). Cross-border interbank contagion in the European banking sector. International Economics, 157, 33–54. https://doi.org/10.1016/J.INTECO.2018.07.002 Gapper, J. (2017, February 7). How Steven Cohen survived an insider trading scandal. Financial Times. Retrieved from https://www.ft.com/content/efda2ca2-ec69-11e6-930f- 061b01e23655 Gazarelli, F., Ardagna, S., & Cena, M. (2015). Macro Rates Analyst, 11 January 2015. London. Giorgi, S., & Weber, K. (2015). Marks of Distinction: Framing and Audience Appreciation in the Context of Investment Advice. Administrative Science Quarterly, 60(2), 333–367. https://doi.org/10.1177/0001839215571125 Goffman, E. (1963). Behavior in Public Places: Notes on the Social Organization of Gatherings. New York: Free Press of Glencoe. Goffman, E. (1969). Strategic interaction. Philadelphia: University of Pennsylvania Press. Goffman, E. (1974). Frame Analysis: An Essay on the Organization of Experience. Cambridge, MA: Harvard University Press. Goodfriend, M. (1986). Monetary Mystique: Secrecy and Central Banking. Journal of 229 Monetary Economics, 17(1), 63–92. https://doi.org/10.1016/0304-3932(86)90006-1 Goyal, S. (2007). Connections : an introduction to the economics of networks. Princeton, N.J.: Princeton University Press. Graham, B. (1959). The Intelligent Investor: A Book of Practical Counsel. New York: Harper. Granovetter, M. S. (1973). The Strength of Weak Ties. American Journal of Sociology, 78(6), 1360–1380. https://doi.org/10.1086/225469 Granovetter, M. S. (1977). The Strength of Weak Ties. Social Networks, 347–367. https://doi.org/10.1016/B978-0-12-442450-0.50025-0 Granovetter, M. S. (1985). Economic Action and Social Structure: The Problem of Embeddedness. American Journal of Sociology, 91(3), 481–510. https://doi.org/10.1086/228311 Gray, J. (2018, March 30). Deciphering the poker face of Draghi, central banker of Europe. Retrieved July 22, 2019, from https://www.handelsblatt.com/today/finance/face-values- deciphering-the-poker-face-of-draghi-central-banker-of- europe/23581666.html?ticket=ST-8045612-BIMpz163s43d5xoAzApq-ap5 Greenspan, A. (1993). Statement before the Committee on Banking, Finance and Urban Affairs, U.S. House of Representatives, October 19, 1993. Washington, D.C.: St Louis Federal Reserve. Retrieved from https://fraser.stlouisfed.org/files/docs/historical/greenspan/Greenspan_19931019.pdf Grinblatt, M., & Keloharju, M. (2001). How Distance, Language, and Culture Influence Stockholdings and Trades. The Journal of Finance, 56(3), 1053–1073. https://doi.org/10.1111/0022-1082.00355 Hannah Murphy. (2018). UK mid-caps suffer drop in liquidity and analyst coverage: Signs that new Mifid II rules around research distribution are beginning to bite. FT Trading Room Financial Times. Retrieved from https://www.ft.com/content/21e8b5de-91ae- 11e8-bb8f-a6a2f7bca546 Hayes, G. (2013, October 1). Central banks — saviours or distorters? Global Capital. Retrieved from https://www.globalcapital.com/article/jbx51gv0pjfx/central-banks- saviours-or-distorters Hendry, J., Sanderson, P., Barker, R., & Roberts, J. (2006). Owners or traders? Conceptualizations of institutional investors and their relationship with corporate managers. Human Relations. https://doi.org/10.1177/0018726706068793 Himick, D., & Brivot, M. (2018). Carriers of ideas in accounting standard-setting and 230 financialization: The role of epistemic communities. Accounting, Organizations and Society, 66, 29–44. https://doi.org/10.1016/J.AOS.2017.12.003 Hirschman, A. O. (1964). The Paternity of an Index. The American Economic Review, 54(5), 761–762. https://doi.org/10.1371/journal.pone.0083448 Hirshleifer, D., & Hong Teoh, S. (2003). Herd Behaviour and Cascading in Capital Markets: a Review and Synthesis. European Financial Management, 9(1), 25–66. https://doi.org/10.1111/1468-036X.00207 Holmes, D. R. (2013). Economy of Words : Communicative Imperatives in Central Banks. Chicago: The University of Chicago Press. Honohan, P. (2019). Currency, Credit and Crisis: Central Banking in Ireland and Europe. Cambridge, U.K.: Cambridge University Press. https://doi.org/10.1017/9781108680325 Hornbach, M., O’Brien, A., Heese, A., Kawano, K., & Liang, J. (2015). 2016 Global Rates Outlook. Dealing with Divergence. Huberman, G. (2001). Familiarity Breeds Investment. Review of Financial Studies, 14(3), 659–680. https://doi.org/10.1093/rfs/14.3.659 Hubert, L., & Schultz, J. (1976). Quadratic Assignment as a General Data Analysis Strategy. British Journal of Mathematical and Statistical Psychology, 29(2), 190–241. https://doi.org/10.1111/j.2044-8317.1976.tb00714.x Iannaccone, L. R. (1991). The Consequences of Religious Market Structure: Adam Smith and the Economics of Religion. Rationality and Society, 3(2), 156–177. https://doi.org/10.1177/1043463191003002002 Ingham, G. (1996). The “New Economic Sociology.” Workemplsoci Work, Employment & Society, 10(3), 549–564. J.P. Morgan. (2016). JP Morgan Europe Credit Research. What to Expect for the ECB’s CSPP. 18 March 2016. Jasper, W. F. (2007, October 29). Taking Delight in Deception: Greenspan’s “Purposeful Obfuscation.” The New American. Retrieved from https://www.thenewamerican.com/economy/commentary/item/3944-taking-delight-in- deception-greenspans-purposeful-obfuscation Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305–360. https://doi.org/10.1016/0304-405X(76)90026-X Jezek, M. (2016). Special Report Credit Financials IG Strategy EUR IG CSPP-Eligible vs. Ineligible Bonds: Performance and Strategy. 231 Jones, M., & O’Donnell, J. (2015, May 20). ECB under fire over market-sensitive closed- door speech - Reuters. Reuters Business News. Jones, S. (2017, September 14). BlackRock to Absorb Cost of Analyst Research Under MiFID II. Bloomberg Markets. Retrieved from https://www.bloomberg.com/news/articles/2017-09-14/blackrock-will-absorb-cost-of- analyst-research-under-mifid-ii Kellard, N., Millo, Y., Simon, J., & Engel, O. (2017). Close Communications: Hedge Funds, Brokers and the Emergence of Herding. British Journal of Management, 28(1), 84–101. https://doi.org/10.1111/1467-8551.12158 Keynes, J. (1936). The General Theory of Employment, Interest and Money. London: Macmillan. Khan, M. (2018, November 27). New ‘Hanseatic’ states stick together in EU big league. Financial Times. Kilduff, M., & Lee, J. W. (2020). The Integration of People and Networks. Annual Review of Organizational Psychology and Organizational Behavior, 7(1), 155–179. https://doi.org/10.1146/annurev-orgpsych-012119-045357 Kilduff, M., & Tsai, W. (2003). Social Networks and Organizations. Social Networks and Organizations. Thousand Oaks, CA: SAGE Publications Ltd. https://doi.org/10.4135/9781849209915 Kilka, M., & Weber, M. (2000). Home Bias in International Stock Return Expectations. Journal of Psychology and Financial Markets, 1(3–4), 176–192. https://doi.org/10.1207/S15327760JPFM0134_3 Knorr Cetina, K., & Bruegger, U. (2002). Global Microstructures: The Virtual Societies of Financial Markets. American Journal of Sociology, 107(4), 905–950. https://doi.org/10.1086/341045 Kohn, D. (2018). Central Bank Talk about Future Monetary Policy. In D. Schoenmaker, H. Huang, & P. Hartmann (Eds.), The Changing Fortunes of Central Banking (pp. 65–79). Cambridge, U.K.: Cambridge University Press. https://doi.org/10.1017/9781108529549.005 Kolhatkar, S. (2017). Black Edge. Inside Information, dirty money, and the quest to bring down the most wanted man on Wall Street. New York: Random House. Krackhardt, D., & Stern, R. N. (1988). Informal Networks and Organizational Crises: An Experimental Simulation. Social Psychology Quarterly, 51(2), 123–140. https://doi.org/10.2307/2786835 232 Krautzberger, M. (2014, October 31). Where to invest in a “low yield world”? CNBC Live TV. London. Retrieved from https://www.cnbc.com/video/2014/10/31/where-to-invest- in-a-low-yield-world.html Krautzberger, M. (2015, July 20). Blackrock Purchased Greek Debt Last Week. Bloomberg On the Move. London, U.K. Retrieved from https://www.bloomberg.com/news/videos/2015-07-20/blackrock-purchased-greek-debt- last-week-krautzberger Krippner, G. R. (2004). Polanyi Symposium: a conversation on embeddedness. Socio- Economic Review, 2(1), 109–135. https://doi.org/10.1093/soceco/2.1.109 Krippner, G. R. (2011). Capitalizing on Crisis: The Political Origins of the Rise of Finance. Cambridge, MA: Harvard University Press. https://doi.org/10.1177/0094306111430635y Krippner, G. R., & Alvarez, A. S. (2007). Embeddedness and the Intellectual Projects of Economic Sociology. Annual Review of Sociology., 33, 219. Kumar, S., & Deo, N. (2012). Correlation and network analysis of global financial indices. Physical Review E, 86(2), 026101. https://doi.org/10.1103/PhysRevE.86.026101 Landis, B. (2016). Personality and social networks in organizations: A review and future directions The Post-structural Tradition in Network Research. Journal of Organizational Behavior, 37, 107–121. https://doi.org/10.1002/job.2004 Lange, A. C., Lenglet, M., & Seyfert, R. (2016). Cultures of high-frequency trading: mapping the landscape of algorithmic developments in contemporary financial markets. Economy and Society, 45(2), 149–165. https://doi.org/10.1080/03085147.2016.1213986 Larsen, J., Axhausen, K. W., & Urry, J. (2006). Geographies of Social Networks: Meetings, Travel and Communications. Mobilities, 1(2), 261–283. https://doi.org/10.1080/17450100600726654 Lauman, E. O., Marsden, P. V., & Prensky, D. (1989). The Boundary Specification Problem in Network Analysis. In L. C. Freeman, D. R. White, & A. Kimball Romney (Eds.), Research Methods in Social Network Analysis (pp. 61–88). London: Transaction Publishers. Laurens, B., Arnone, M., & Segalotto, J.-F. (2009). Central bank independence, accountability, and transparency : a global perspective. New York: Palgrave Macmillan. Lazarsfeld, P., & Merton, R. (1954). Friendship as a social process: a substantive and methodological analysis. In M. Berger (Ed.), Freedom and Control in Modern Society 233 (pp. 8–66). New York: Van Nostrand. Lebaron, F. (2010). European Central Bank leaders in the global space of central bankers: A Geometric Data Analysis approach. French Politics, 8(3), 294–320. https://doi.org/10.1057/fp.2010.15 Lebaron, F., & Dogan, A. (2016). Do central bankers’ biographies matter? Sociologica, (2/2016), 1–36. https://doi.org/10.2383/85290 Leins, S. (2018). Stories of capitalism : inside the role of financial analysts. Chicago: The University of Chicago Press. Levy, H., & Sarnat, M. (1970). International Diversification of Investment Portfolios. The American Economic Review, 60(4), 668–675. https://doi.org/10.1287/mnsc.40.1.140 Liesman, S. (2014, August 25). Contrast between US and Europe highlights Jackson Hole. Retrieved April 6, 2019, from https://www.cnbc.com/2014/08/25/contrast-between-us- and-europe-highlights-jackson-hole.html Lin, M., & Viswanathan, S. (2016). Home bias in online investments: An empirical study of an online crowdfunding market. Management Science, 62(5), 1393–1414. https://doi.org/10.1287/mnsc.2015.2206 MacKenzie, D. (2018). ‘Making’, ‘taking’ and the material political economy of algorithmic trading. Economy and Society, 47(4), 501–523. https://doi.org/10.1080/03085147.2018.1528076 MacKenzie, D. A. (2006). An engine, not a camera : how financial models shape markets. Cambridge, MA: MIT Press. Mackenzie, D., & Millo, Y. (2003). Constructing a Market, Performing Theory: The Historical Sociology of a Financial Derivatives Exchange. American Journal of Sociology, 109(1), 107–145. https://doi.org/10.1086/374404 Maitlis, S. (2005). The Social Processes of Organizational Sensemaking. The Academy of Management Journal, 48(1), 21–49. https://doi.org/10.2307/20159639 Malkiel, B. G., & Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. Journal of Finance, 25(2), 383–417. https://doi.org/10.1111/j.1540- 6261.1970.tb00518.x Mallaby, S. (2016). The Man Who Knew: The Life and Times of Alan Greenspan. London: Bloomsbury Publishing. Mariolis, P., & Jones, M. H. (1982). Centrality in Corporate Interlock Networks: Reliability and Stability. Administrative Science Quarterly, 27(4), 571. https://doi.org/10.2307/2392531 234 Markowitz, H. (1952). Portfolio Selection. The Journal of Finance, 7(1), 77–91. https://doi.org/10.2307/2975974 Marsden, P. V. (1988). Homogeneity in confiding relations. Social Networks, 10(1), 57–76. https://doi.org/10.1016/0378-8733(88)90010-X Marsden, P. V. (1990). Network Data and Measurement. Annual Review of Sociology, 16(1), 435–463. https://doi.org/10.1146/annurev.so.16.080190.002251 Marsden, P. V. (2002). Egocentric and sociocentric measures of network centrality. Social Networks, 24(4), 407–422. https://doi.org/10.1016/S0378-8733(02)00016-3 McGlashan, C., & Fildes, T. (2016, February 11). Norrey see his like again: Carl to retire. Global Capital. Retrieved from https://www.globalcapital.com/article/wgflht122087/norrey-see-his-like-again-carl-to- retire McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a Feather: Homophily in Social Networks. Annual Review of Sociology, 27(1), 415–444. https://doi.org/10.1146/annurev.soc.27.1.415 Merton, R. K. (1948). The Self-Fulfilling Prophecy. The Antioch Review, 8(2), 193–210. https://doi.org/10.2307/4609267 Miediema, D., & Wilkes, T. (2011, September 15). UBS $2 bln Delta One loss puzzles experts. Reuters Market News. Retrieved from https://uk.reuters.com/article/uk-delta- one/ubs-2-billion-delta-one-loss-puzzles-experts-idUKLNE78F04B20110916 Mintz, B., & Schwartz, M. (1981). Interlocking Directorates and Interest Group Formation. American Sociological Review, 46(6), 851–869. https://doi.org/10.2307/2095083 Mintz, B., & Schwartz, M. (1985). The Power Structure of American Business. Chicago: The University of Chicago Press. Mishkin, F. S. (2007). Monetary policy strategy. Cambridge, MA: MIT Press. Mizruchi, M. S. (1996). What do interlocks do? An Analysis, Critique, and Assessment of Research on Interlocking Directorates. Annual Review of Sociology, 22, 271–298. https://doi.org/10.1146/annurev.soc.22.1.271 Mooney, A. (2018a, January 3). Ashmore to absorb costs of research under MiFID II. Financial Times. Retrieved from https://www.ft.com/content/1675871d-5119-3e11- be77-93948406265d Mooney, A. (2018b, February 22). Fidelity backtracks on passing external research costs to clients. Financial Times. Retrieved from https://www.ft.com/content/fba7885a-17b9- 11e8-9376-4a6390addb44 235 Mooney, A. (2018c, February 22). The definitive list of asset managers that will pay for research: Fidelity reacts to client complaints and will now absorb research costs under Mifid II. Financial Times. Retrieved from https://www.ft.com/content/d7be86de-8f1d- 11e7-a352-e46f43c5825d Morris, S., & Shin, H. S. (2008). Coordinating expectations in monetary policy. In J.-P. Touffut (Ed.), Central Banks as Economic Institutions (pp. 88–104). Edward Elgar Publishing. Muniesa, F., & Callon, M. (2007). Economic experiments and the construction of markets. In D. Mackenzie, F. Muniesa, & L. Siu (Eds.), Do Economists Make Markets?: On the Performativity of Economics (pp. 163–189). Princeton, N.J.: Princeton University Press. Namaki, A., Shirazi, A. H., Raei, R., & Jafari, G. R. (2011). Network analysis of a financial market based on genuine correlation and threshold method. Physica A: Statistical Mechanics and Its Applications, 390(21–22), 3835–3841. https://doi.org/10.1016/J.PHYSA.2011.06.033 Nederlandsche Bank, D., Boermans, M., & Keshkov, V. (2018). The impact of the ECB asset purchases on the European bond market structure: Granular evidence on ownership concentration. DNB Working Paper 590. Amsterdam. Retrieved from https://www.dnb.nl/binaries/Working Paper No. 590_tcm46-375317.pdf Nofsinger, J. R., & Sias, R. W. (1999). Herding and Feedback Trading by Institutional and Individual Investors. The Journal of Finance, 54(6), 2263–2295. https://doi.org/10.1111/0022-1082.00188 Pagano, M., Serrano, A. S., & Zechner, J. (2019). Reports of the Advisory Scientific Committee: Can ETFs contribute to systemic risk? No 9, June 2019. Frankfurt am Main. https://doi.org/10.2849/45983 Papadia, F., & Välimäki, T. (2018). Central banking in turbulent times. Oxford, U.K.: Oxford University Press. Pill, H., Daly, K., Schuhmacher, D., Benito, A., Durre, A., Holboell Nielsen, L., & Vernet, P. (2014). European Economics Analyst, Issue No: 14/42. November 27, 2014. London. Pill, H., Daly, K., Schuhmacher, D., Benito, A., Holboell Nielsen, L., Demongeot, A., & Graves, S. (2014). European Economics Analyst Issue No: 14/18. May 15, 2014. London. Podolny, J. M. (1993). A Status-Based Model of Market Competition. American Journal of Sociology, 98(4), 829–872. https://doi.org/10.1086/230091 Podolny, J. M. (1994). Market Uncertainty and the Social Character of Economic Exchange. 236 Administrative Science Quarterly, 39(3), 458. https://doi.org/10.2307/2393299 Podolny, J. M. (2001). Networks as the Pipes and Prisms of the Market. American Journal of Sociology, 107(1), 33–60. https://doi.org/10.1086/323038 Podolny, J. M., & Phillips, D. J. (1996). The Dynamics of Organizational Status. Industrial and Corporate Change, 5(2), 453–471. https://doi.org/https://doi.org/10.1093/icc/5.2.453 Powell, W., & Smith-Doerr, L. (2005). Networks and Economic Life. In N. J. Smelser & R. Swedberg (Eds.), The Handbook of Economic Sociology (2nd ed., pp. 379–402). Princeton, N.J.: Princeton University Press. https://doi.org/10.17323/1726-3247-2003-3- 61-105 Preda, A. (2007). The sociological approach to financial markets. Journal of Economic Surveys. https://doi.org/10.1111/j.1467-6419.2007.00512.x Rao, H., Greve, H. R., & Davis, G. F. (2001). Fool’s Gold: Social Proof in the Initiation and Abandonment of Coverage by Wall Street Analysts. Administrative Science Quarterly, 46(3), 502. https://doi.org/10.2307/3094873 Reagans, R. E., & Zuckerman, E. W. (2008). Why knowledge does not equal power: the network redundancy trade-off. Industrial and Corporate Change, 17(5), 903–944. https://doi.org/10.1093/icc/dtn036 Rieger, C. (2015). Design and impact of LSAPs. Presentation ECB Bond Market Contact Group. 27 January 2015. Frankfurt am Main. Retrieved from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/150127/2015-01-27-Item-3- 1.pdf?634fd27c4c5dec3e1269fd6284ae88ed Riles, A. (2011). Collateral Knowledge : Legal Reasoning in the Global Financial Markets. Chicago: The University of Chicago Press. Riles, A. (2018). Financial Citizenship: Experts, Publics, and the Politics of Central Banking. Ithaca: Cornell University Press. Roberts, J., Sanderson, P., Barker, R., & Hendry, J. (2006). In the mirror of the market: The disciplinary effects of company/fund manager meetings. Accounting, Organizations and Society, 31(3), 277–294. https://doi.org/10.1016/J.AOS.2005.02.001 Roy, R. B., & Sarkar, U. K. (2011). Identifying influential stock indices from global stock markets: A social network analysis approach. Procedia Computer Science, 5, 442–449. https://doi.org/10.1016/J.PROCS.2011.07.057 Roy, R. B., & Sarkar, U. K. (2013). A social network approach to change detection in the interdependence structure of global stock markets. Social Network Analysis and Mining, 237 3(3), 269–283. https://doi.org/10.1007/s13278-012-0063-y Sachee, Z. (2016). Electronic trading & the growth of automated trading strategies. ECB Bond Market Contact Group. London. Retrieved from https://www.ecb.europa.eu/paym/groups/pdf/bmcg/160119/2016-01- 19_Item_1.a.AutomatedTradingStrategies- dealerperspective.pdf?945d165db9fffc19b0ad4d92b60deb13 Scheinkman, J. A., & Xiong, W. (2003). Overconfidence and Speculative Bubbles. Journal of Political Economy, 111(6), 1183–1220. https://doi.org/10.1086/378531 Scheller, H. P. (2004). The European central bank : history, role and functions. Frankfurt am Main: European Central Bank. Schelling, T. C. (1958). The strategy of conflict . Prospectus for a reorientation of game theory. Journal of Conflict Resolution, 2(3), 203–264. https://doi.org/10.1177/002200275800200301 Schuhmacher, D. (2019, March 7). Makes sense for ECB to adjust forward guidance, analyst says. CNBC Interview. CNBC Live TV. Retrieved from https://www.cnbc.com/video/2019/03/07/makes-sense-for-ecb-to-adjust-forward- guidance-analyst-says.html Schweitzer, F., Fagiolo, G., Sornette, D., Vega-Redondo, F., Vespignani, A., & White, D. R. (2009). Economic networks: the new challenges. Science, 325(5939), 422. https://doi.org/10.1126/science.1173644 Scott, J. (2017). Social Network Analysis. Thousand Oaks: Sage Publications. Seabrooke, L. (2014). Epistemic arbitrage: Transnational professional knowledge in action. Journal of Professions and Organization, 1(1), 49–64. https://doi.org/10.1093/jpo/jot005 Securities and Exchange Commission. (2013). Administrative Proceedings. In the Matter of Steve Cohen, Respondent. United States of America: Securities and Exchange Commission. Retrieved from https://www.sec.gov/litigation/admin/2013/ia-3634.pdf Sharpe, W. F. (1991). The Arithmetic of Active Management. Financial Analysts Journal, 47(1), 7–9. https://doi.org/10.2469/faj.v47.n1.7 Sias, R. W. (2004). Institutional Herding. Review of Financial Studies, 17(1), 165–206. https://doi.org/10.1093/rfs/hhg035 Siklos, P. L. (2011). Central bank transparency: another look. Applied Economics Letters Applied Economics Letters, 18(10), 929–933. Siklos, P. L., & Sturm, J.-E. (2013). Central Bank Communication, Decision Making, and 238 Governance : Issues, Challenges, and Case Studies. Cambridge, MA: MIT Press. Sims, T. G., & Wessel, D. (2000, April 27). The European Central Bank Can’t Master Communication. Wall Street Journal. Retrieved from https://www.wsj.com/articles/SB956784524344568471 Sinn, H.-W. (2011). Why Germany is threatened by a debt tsunami - Lecture and discussion on the Euro Crisis, Humboldt University. Berlin, Germany: ifo Institute Center for Economic Studies. Retrieved from http://www.cesifo- group.de/ifoHome/events/individual-events/Archive/2011/event-20110509-vortrag- berlin.html Sinn, H.-W., & Wollmershäuser, T. (2012). Target loans, current account balances and capital flows: the ECB’s rescue facility. International Tax and Public Finance, 19(4), 468–508. https://doi.org/10.1007/s10797-012-9236-x Slater, D. (2002). From calculation to alienation : disentangling economic abstractions. Economy and Society, 31(2), 234–349. https://doi.org/10.1080/03085140220123144 Sorenson, O., & Stuart, T. E. (2001). Syndication Networks and the Spatial Distribution of Venture Capital Investments. American Journal of Sociology, 106(6), 1546–1588. https://doi.org/10.1086/321301 Spence, P. (2015, May 19). Hedge funds got early access to information that prompted euro slide. Retrieved July 16, 2019, from https://www.telegraph.co.uk/finance/economics/11616112/Hedge-funds-got-early- access-to-information-that-prompted-euro-slide.html Spyrou, S. (2013). Herding in financial markets: a review of the literature. Review of Behavioural Finance, 5(2), 175–194. https://doi.org/10.1108/RBF-02-2013-0009 Starbuck, W. H., & Milliken, F. J. (1988). Executives’ Perceptual Filters: What They Notice and How They Make Sense. In D. Hambrick (Ed.), The Executive Effect: Concepts and Methods for Studying Top Managers (pp. 35–65). Greenwich, CT: JAI Press. Stearns, L. B., & Mizruchi, M. S. (2005). Banking and Financial Markets. In N. J. Smelser & R. Swedberg (Eds.), The Handbook of Economic Sociology (2nd ed., pp. 284–306). New York: Princeton University Press. Stern, G. H. (2004). Interview with Jean-Claude Trichet. Federal Reserve Bank of Minneapolis. The Region, (December Issue). Retrieved from https://www.minneapolisfed.org/publications/the-region/interview-with-jeanclaude- trichet Stolper, O., & Walter, A. (2018). Birds of a Feather: The Impact of Homophily on the 239 Propensity to Follow Financial Advice. The Review of Financial Studies, 32(2), 524– 563. Retrieved from http://dx.doi.org/10.1093/rfs/hhy082 Strasser, G. (2018). The monetary policy transmission mechanism in the euro area. ECB Central Banking Seminar 11 July 2018. Frankfurt am Main. Retrieved from https://www.ecb.europa.eu/pub/conferences/shared/pdf/20180709_ecb_central_banking _seminar/2018-07- 11_The_monetary_policy_transmission_mechanism_in_the_euro_area_-_Strasser.pdf Suttard, J., Kini, D., & Edwards, C. (2016). Credit Telegram: ECB CSPP and credit: we know more. HSBC. London. Svetlova, E. (2018). Financial models and society : villains or scapegoats? Cheltenham, U.K.: Edward Elgar Publishing. Swedberg, R. (2000). Afterword: The role of the market in Max Weber’s work. Theory and Society, 29(3), 373–384. https://doi.org/10.1023/A:1007056722175 Taffler, R. J., Spence, C., & Eshraghi, A. (2017). Emotional economic man: Calculation and anxiety in fund management. Accounting, Organizations and Society. https://doi.org/10.1016/j.aos.2017.07.003 The Economist. (2018, December). Northern member states unite on euro-zone reform. The Economist. Retrieved from https://www.economist.com/europe/2018/12/08/northern- member-states-unite-on-euro-zone-reform Tortoriello, M., & Krackhardt, D. (2010). Activating Cross-Boundary Knowledge: The Role of Simmelian Ties in the Generation of Innovations. The Academy of Management Journal, 53(1), 167–181. https://doi.org/10.5465/amj.2010.48037420 Treanor, J. (2012, June 28). Barclays fined £290m as bid to manipulate interest rates is exposed. The Guardian. Retrieved from https://www.theguardian.com/business/2012/jun/27/barclays-chief-bob-diamond-bonus- fine Tribe, K. (2002). Reviewed Work: Börsenwesen: Schriften und Reden 1893–1898 (WG I/5), Part 1, Part 2 by M. Weber, Knut Borchardt, Cornelia Meyer-Stoll. Max Weber Studies, 2(2), 242–246. Retrieved from http://www.jstor.org/stable/24579611 Trichet, J.-C. (2008). A few remarks on communication by central banks. Keynote address. Retrieved July 2, 2019, from https://www.ecb.europa.eu/press/key/date/2008/html/sp080116_1.en.html#ftn.fnid1 Tucker, P. (2018). Unelected Power : The Quest for Legitimacy in Central Banking and the Regulatory State. Princeton, NJ: Princeton University Press. 240 Tuckett, D., Holmes, D., Pearson, A., & Chaplin, G. (2020). Staff Working Paper No. 870 Monetary policy and the management of uncertainty: a narrative approach. London, U.K. Retrieved from www.bankofengland.co.uk/working-paper/staff-working-papers Tuckett, D., & Nikolic, M. (2017). The role of conviction and narrative in decision-making under radical uncertainty. Theory & Psychology, 27(4), 501–523. https://doi.org/10.1177/0959354317713158 Uetake, T. (2017, October 20). In Kuroda’s face - researchers find ways to predict central bank changes - Reuters. Thomson Reuters. Retrieved from https://uk.reuters.com/article/uk-japan-boj-face/in-kurodas-face-researchers-find-ways- to-predict-central-bank-changes-idUKKBN1CP0GY Urry, J. (2003). Social networks, travel and talk. British Journal of Sociology, 54(2), 155– 175. https://doi.org/10.1080/0007131032000080186 Urry, J. (2004). Small Worlds and the New “Social Physics.” Global Networks, 4(2), 109– 130. https://doi.org/10.1111/j.1471-0374.2004.00083.x Useem, M. (1980). Corporations and the Corporate Elite. Ann. Rev. Social, 6, 41–77. https://doi.org/10.1146/annurev.so.06.080180.000353 Useem, M. (1996). Investor Capitalism: How Money Managers are Changing the Face of Corporate America. (1st ed.). New York, NY: Basic Books. Uzzi, B. (1996). The Sources and Consequences of Embeddedness for the Economic Performance of Organizations: The Network Effect. American Sociological Review, 61(4), 674–698. https://doi.org/10.2307/2096399 Uzzi, B. (1997). Social structure and competition in interfirm networks: The paradox of embeddedness. Administrative Science Quarterly, 42(1), 35–67. https://doi.org/10.2307/2393808 Velthuis, O. (2015). Making monetary markets transparent: the European Central Bank’s communication policy and its interactions with the media. Economy and Society, 44(2), 316–340. https://doi.org/10.1080/03085147.2015.1013355 Volcker, P. A., & Harper, C. (2018). Keeping at it : the quest for sound money and good government. New York, NY: PublicAffairs. Vollmer, H. (2007). How to do more with numbers: Elementary stakes, framing, keying, and the three-dimensional character of numerical signs. Accounting, Organizations and Society, 32(6), 577–600. https://doi.org/10.1016/J.AOS.2006.10.001 Walter, T., & Wansleben, L. (2019). How central bankers learned to love financialization: The Fed, the Bank, and the enlisting of unfettered markets in the conduct of monetary 241 policy. Socio-Economic Review. https://doi.org/10.1093/SER/MWZ011 Wansleben, L. (2018). How expectations became governable: institutional change and the performative power of central banks. Theory and Society, 47(6), 773–803. https://doi.org/10.1007/s11186-018-09334-0 Watts, D. J. (1999). Small worlds : the dynamics of networks between order and randomness. Princeton, N.J.: Princeton University Press. Watts, D. J. (2004). Six degrees: The science of a connected age. London: Vintage. Weber, M. (2000). Stock and Commodity Exchanges ["Die Boerse" (1894)]. Theory and Society, 29(3), 305–338. https://doi.org/10.1023/A:1007042728962 Weber, M., Borchardt, K., & Meyer-Stoll, C. (2000). Börsenwesen: Schriften und Reden 1893 – 1898 (WG I/5) 1. und 2. Halbband. Tübingen: J.C.B. Mohr. Weick, K. E. (1979). The social psychology of organizing. New York: McGraw-Hill. Weick, K. E. (1988). Enacted Sensemaking in Crisis Situations. Journal of Management Studies, 25(4), 305–317. https://doi.org/10.1111/j.1467-6486.1988.tb00039.x Weick, K. E. (1995). Sensemaking in organizations. Thousand Oaks: Sage Publications. Wellman, B. (1988). Structural analysis: from method and metaphor to theory and substance. In B. Wellman & S. D. Berkowitz (Eds.), Social Structures: A Network Approach (pp. 19–61). Princeton, N.J.: Princeton University Press. White, H. C. (1981). Where Do Markets Come From? American Journal of Sociology, 87(3), 517–547. https://doi.org/10.1086/227495 White, H. C. (2008). Identity and Control: how social networks are formed. (2nd ed.). Princeton, N.J.: Princeton University Press. White, H. C., Boorman, S. A., & Breiger, R. L. (1976). Social Structure from Multiple Networks. I. Blockmodels of Roles and Positions. American Journal of Sociology, 81(4), 730–780. https://doi.org/10.1086/226141 Whitman, M. J., & Shubik, M. (1979). The Aggressive Conservative Investor. New York: Random House. Wiersema, M. F., & Zhang, Y. (2011). CEO Dismissal: The role of investment analysts. Strategic Management Journal, 32(11), 1161–1182. https://doi.org/10.1002/smj.932 Wolf, M. (2011, May 31). Intolerable choices for the eurozone. Financial Times. Retrieved from https://www.ft.com/content/1a61825a-8bb7-11e0-a725-00144feab49a Womack, K. L. (1996). Do Brokerage Analysts’ Recommendations Have Investment Value? The Journal of Finance, 51(1), 137–167. https://doi.org/10.1111/j.1540- 6261.1996.tb05205.x 242 Wood, C. (Guest), Balchunas, E. (Host), & Weber, J. (Host). (2019, July 11). Disruption as an Investment Thesis. Trillions. [Audio Podcast]. Retrieved from https://podcasts.apple.com/https://www.bloomberg.com/news/articles/2019-07- 11/disruption-as-an-investment-thesis Zaghini, A. (2019). The CSPP at work: Yield heterogeneity and the portfolio rebalancing channel. Journal of Corporate Finance, 56, 282–297. https://doi.org/10.1016/j.jcorpfin.2018.12.004 Zaloom, C. (2003). Ambiguous numbers: Trading technologies and interpretation in financial markets. American Ethnologist, 30(2), 258–272. https://doi.org/10.1525/ae.2003.30.2.258 Zaloom, C. (2006). Out of the Pits: Traders and Technology from Chicago to London. Chicago: University of Chicago Press. Zuckerman, E. W. (1999). The Categorical Imperative: Securities Analysts and the Illegitimacy Discount. American Journal of Sociology, 104(5), 1398–1438. https://doi.org/10.1086/210178 Zuckerman, E. W. (2000). Focusing the Corporate Product: Securities Analysts and De- Diversification. Administrative Science Quarterly, 45(3), 591. https://doi.org/10.2307/2667110 Zuckerman, E. W. (2012). Market Efficiency: A Sociological Perspective. In K. Knorr-Cetina & A. Preda (Eds.), The Oxford Handbook of the Sociology of Finance (pp. 223–249). Oxford, U.K.: Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199590162.013.0013