Repository logo
 

Behavioural Macro-Financial Cycles


Type

Thesis

Change log

Authors

Lake, Alfred Fitzgerald 

Abstract

In the first chapter I study the optimal inflation expectations that it is possible for agents to estimate and the difference between these and rational expectations. It is typically not feasible for an agent with a macroeconomic sample of realistic length to estimate a conditionally unbiased predictor of future variables, such as rational expectations. It is also often not optimal, in terms of forecast error, to minimise the conditional biases imposed, as using statistically simple expectations will often reduce forecast variance sufficiently to outweigh forecast bias. I therefore introduce optimal feasible expectations, the expectations that are predicted to minimise the relevant measure of forecast error out of the set of expectations that agents can estimate, as a realistic alternative to infeasible rational expectations. I then empirically estimate the optimal conditional biases when forecasting US inflation using a factor weighted ridge approach. I find it is optimal to impose large conditional biases: one should essentially only use information on past changes in price indices, despite several other variables and factors having economically and statistically significant associations with future inflation. I then compare these to the conditional biases in US household forecast surveys. I find that many of the conditional biases are similar, although households also make errors that reduce their forecast performance compared to feasible empirical alternatives. Therefore a combination of optimal feasible expectations and behavioural errors appear to explain US household inflation forecasts.

In the second chapter I study whether increases in asset price convergence and the quantity of cross-border asset holdings, common measures of financial integration, imply high quality changes in financial integration, i.e. changes that are likely to produce the largest net economic benefits. I use a new methodology based on a Bayesian FAVAR to overcome the econometrically challenging setting and test three aspects of the quality of changes in financial integration measures. I apply this methodology to the EU in the 21st century and find that there is a common factor that drives a wide range of price and quantity integration measures. However the changes in financial integration are primarily cyclical, as long-term cyclicality strongly outweighs deterministic and stochastic trends, and dependent on macroeconomic conditions, as virtually all sign identified economic shocks cause large corresponding effects on financial integration. This suggests that increases in financial integration have not been high quality: they actually appear most closely related to cyclical changes in the underlying risks of European assets and aversion to these risks.

In the third chapter I introduce a new test of whether house prices are always equal to their fundamental value, adjusted to account for contractual rigidities and search frictions, based on the speed of their reaction to monetary shocks. I justify this test with two conceptual frameworks and references to existing empirical work on the transmission mechanisms of monetary policy. I then apply this test to house prices in the US using narrative monetary shocks in a local projections approach. I find that real house prices do not react to monetary shocks when contractual rigidities stop binding, however they have economically and statistically significant reactions at horizons over a year. This result is inconsistent with house prices always being equal to their fundamental value, but is consistent with agents either not fully observing monetary shocks or not incorporating these shocks into their expectations rationally. I also use a sign decomposition based on the conceptual frameworks to identify the relative importance of proximal drivers of house price cycles: I find that consumption demand is the most important driver but asset demand is also relatively important. Therefore housing cycles are likely to arise from the partially behavioural reactions to changes in housing demand.

Description

Date

2020-01-01

Advisors

Holly, Sean

Keywords

Financial Economics, Macroeconomics, Behavioural Economics, Applied Econometrics, Cyclicality and Sustainability, Financial and Economic Expectations

Qualification

Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
Sponsorship
The Cambridge Trust and Girton College, Cambridge