Wealth Shocks And Relative Risk Aversion

In order to explain problems in asset pricing models, modern literature relaxes the assumption of constant relative risk aversion. This implies that there exists wealth effect in relative risk aversion.This paper tests the relationship using U.S. quarterly macroeconomic data for the period1952-2008. We divide the total wealth into two parts to investigate their wealth effects on relative risk aversion. One is financial wealth, and the other is housing wealth. We find that both of them are significant, with financial wealth effect being negative and housing wealth effect positive. Also, we find that magnitude of the financial wealth effect is5times as large as that of housing wealth effect.We compute wealth shocks by using ARMA models to analyze the impact of wealth shocks on relative risk aversion. We find that the effect of wealth shock on relative risk aversion is negatively significant, while housing wealth shock does not help to explain changes in portfolio composition. Based on this result, the financial shock is estimated to be7times as large as the housing wealth shock.Our results concerning wealth effects are in support of Giuseppe Cappelletti (2012), however, they are unlike Brunnermeier and Nagel (2008). Compared with Ricardo M. Sousa (2007), we find that wealth shocks come from the shock of financial wealth, which is an improvement of his finding.By using the Standard&Poor’s (S&P500) Composite Index to measure financial prices, we find the change in financial prices has strong power in explaining the changes in risky assets ratio, which is quite different from results based on micro-data. However, it does not have predicting power. We argue that every household can reallocate their wealth between risky assets and riskless assets when facing price changes. However, the total value of risky assets varies with the change of financial prices. We use after-tax income to capture human resources, and find that changes in after-tax income have no power in explaining variation of relative risk aversion. Finally using consumption-wealth ratio and labor income-consumption ratio, we find that relative fluctuations in consumption have no power in explaining the changes in risky assets ratio.