Expected Stock Returns: Explaining Consumption and Household Portfolio Choice
Project Leader: Sule Alan
This project is the first sustained analysis of heterogeneous expectations formation within a life-cycle portfolio choice model.
Empirical evidence strongly suggests that the average historical stock market return is an inadequate proxy for expected returns. One route taken by some researchers is to use direct survey responses to identify expected returns. This literature has produced some interesting results including that the expected returns are procyclical and investors (even professional ones) seem to use some rather heuristic rules in forming expectations regarding future stock returns. In particular, expected returns seem to be extrapolated from near past stock market returns. This finding appears to be quite robust across different data sets. In this project, Dr Alan explored whether this phenomenon manifests itself in the actual portfolio choices made by households, and investigated whether the failure of the standard dynamic portfolio choice model in capturing the life cycle dynamics of risky asset holding can be circumvented once we relax the assumption of fully rational expectations formation.
For this, Dr Alan first parameterised the distribution stock market expectations (likely to be heterogeneous across households) and treated the moments as structural parameters. She estimated these moments and inter-temporal allocation parameters for different cohorts and education groups via Simulated Minimum Distance using micro data on portfolio choice and consumption. Consistent with the experimental evidence and recent survey results, she found that households extrapolate from near-past returns when forming expectations over stock market and make portfolio allocation decisions accordingly. The project shows that incorporating such a simple rule of expectation formation takes us a long way towards reconciling observed consumption and portfolio choice dynamics with predictions of the standard model.