Essays on Information in Financial Markets

Hennig, Tristan 

Thumbnail Image
Change log

Chapter 1 demonstrates how to jointly quantify the size of search and information frictions in OTC markets. I use transaction data for the U.S. corporate bond market to structurally estimate a model featuring both frictions via the simulated method of moments. The data support the notion that trades are informative and that uncertainty about the fundamental value diminishes as time passes. The model allows me to evaluate the rise of electronic trading as well as recent regulation such as MiFID II, both of which have reduced the severity of the search friction in OTC markets. Increasing the probability of finding a trading partner by 20% decreases spreads by 18 percent on average, increases welfare by 21.8 percent on average, and decreases price volatility by 40 percent on average. The effects are biggest in bonds with long time to maturity and smallest in bonds close to maturity. However, these improvements come at the cost of a substantial slow-down in price discovery. The speed of convergence of the price to the true value decreases by up to 26 percent.

Chapter 2 is joint work with my supervisor. We present a sequential trade model in the tradition of Glosten and Milgrom (1985). In our model, agents receive informative but imperfect information about the fundamental value of a security and learn by observing others’ trading decisions. Herding occurs when traders ignore their private information to switch in the direction of the crowd. Similarly, contrarianism occurs when traders ignore their private information to switch against the crowd. Using NYSE TAQ tick data, we estimate our model by maximum likelihood and show that it fits the data significantly better than existing models by Easley et al. (1997) and Cipriani and Guarino (2014). The event uncertainty setup commonly used in this literature is rejected in favour of a general three-state setup. We find that contrarianism occurs very rarely and has negligible price impact. In contrast, about 35% of days in our sample feature herd behaviour at some point during the day. Compared to existing results, we show that herding is more frequent, has a more variable duration, occurs later in the day, and causes higher deviations of the price from an informationally efficient benchmark. The data also support the notion that herd behaviour is self-reinforcing rather than self-destructive.

Chapter 3 is divided into two self-contained but related parts. The first part addresses a key assumption made in sequential trade models, namely that changes in the security's fundamental value only occur in between trading days. I show that this assumption can easily be relaxed and present two results. First, the data do not support the assumption that fundamental value changes only occur over night but instead are much more frequent. Second, when the model is adjusted to allow for within-day value changes, the extent of herd behaviour that is predicted by the model is substantially reduced. The second part of this chapter develops a model that departs from another assumption typically used in sequential trade models. Instead of restricting agents to either buy or sell a single unit of the security, I allow the action space to be the real line. Despite the unrestricted continuous action space herding and contrarianism can still arise in equilibrium. Using transactions data I show how to estimate the model via maximum likelihood. The results demonstrate that using only the trade direction instead of all information contained in the volume data can bias the results.

Chapter 4 studies the nature of trading when market participants are not perfectly rational Bayesians. The standard way of modelling decision making under uncertainty assumes that agents update their beliefs about uncertain variables in a statistically correct way. In this chapter, I introduce a notion of bounded rationality that constrains agents to entertain only a finite number, rather than a continuum, of posterior beliefs. Given this constraint, I analyse how coarse beliefs can be structured optimally. I then introduce a mass of traders with coarse beliefs into a simple financial market and explore the effects on the price of an asset with an uncertain payout and on trading between the agents. Introducing agents with coarse beliefs into an asset market generates trading that would otherwise not occur.

Sabourian, Hamid
Financial Markets, Information, Market Microstructure, Financial Economics, Microeconomics
Doctor of Philosophy (PhD)
Awarding Institution
University of Cambridge
Cambridge Trust Vice Chancellor's Scholarship Faculty of Economics Trust Funds