Endogenous Extrapolation: Implications f

Why are Economies Unstable? Research Project

Endogenous Extrapolation: Implications for Boom-Bust Cycles and Macroeconomic Policy

Scroll Down

Principal Investigator: Michael Hatcher

Dr Michael Hatcher is Lecturer in Economics within Economic, Social and Political Science at the University of Southampton.

Michael joined Southampton as a Lecturer in September 2014. He holds a PhD in Economics from Cardiff University and was ESRC Postdoctoral Fellow at the University of Glasgow. Prior to that, he taught economics for a year at the University of Oxford. Michael’s research interests are in the area of macroeconomic policy in general equilibrium models, with a focus on monetary policy, government debt, pensions and housing markets.

Project Summary

In this project I have developed models for understanding asset price booms and busts and their implications for the macroeconomy. The project makes two main contributions:

(1) Endogenous extrapolation: Investors have cognitive limitations that prevent them from having rational expectations, but they do attempt to maximize returns by switching to heuristics that performed well in the recent past. Differently to current approaches, investors may choose between multiple extrapolative predictors associated with different degrees of optimism about future prices.

(2) Wealth channel: Asset prices affect the macroeconomy through a wealth channel in aggregate demand. Higher asset prices increase aggregate wealth and consumption, raising output. A fall in asset prices has the opposite effect. In the main paper of my project I focus on the housing market, but I have also studied the stock market. The wealth channel encompasses both as a general mechanism applicable to asset prices.

The main findings of the project are as follows:

· Endogenous extrapolation helps the model to generate a tight correlation between investor optimism and house price returns, as empirically supported in U.S. data on house price returns and survey expectations. The model does well on this score because it generates waves of optimism that roughly coincide with house price peaks and troughs.

· The model with endogenous extrapolation does not behave like a predictable weighted average of models with the individual extrapolative predictors. Instead, its behaviour is complex and highly non-linear; for example, the incidence of instability is increased and there are `tipping points’ in behavioural parameters which are otherwise absent.

· The long run impact of a house price boom and bust on macro variables depends on wealth at the start of the boom, being expansionary if wealth is sufficiently high, but contractionary if most investors are forced to borrow heavily to buy housing. Hence, the circumstances that precede a housing boom matter, with wealth and credit availability being key.

· Monetary policy that `leans against the wind’ of past house price increases is highly risky. If the interest rate response is too strong, it leads to recurrent fluctuations or a collapse in house prices. Macroprudential regulation of credit-wealth ratios is also problematic because it can leads to highly unstable dynamics. However, limits on house purchases per investor are an effective (and equitable) stabilization tool.

Future work could assess the empirical performance of endogenous extrapolation in greater detail; for example, by estimating the model or extending the study to a wider range of house price data and stylized facts. The housing-macro model itself also requires further development.

Finally, one could extend the model with a network, as `word of mouth’ is known to be important in asset markets – and hence learning of investors will be social as well as performance based. I am currently pursuing this avenue in joint work with my colleague Tim Hellmann.