Summary: |
In this article our main objective is to study simultaneously the influence of three important risk factors on asset returns, namely the return on financial market (proxy for financial wealth), the consumption growth and the return on real estate wealth. The third factor has been surprisingly neglected by the financial literature: the global financial crisis has highlighted this drawback. More specifically, we develop a linear multifactor pricing model using monthly data of stock returns. For a long period from January 1972 to December 2013, our results shed new light on the relative importance of real estate risk on the 25 portfolios formed on size and book-to-market as computed from the CRSP database. Dividing our full sample in two sub-periods, we compute for each asset an estimate of the beta coefficient of each factor and then the mean risk premia associated. Using our full sample and dividing it in two sub-periods related to the contrasted evolution of the systemic risk (from January 1972 to December 1990 and from January 1991 to December 2013), we present results based on two alternative GMM procedures, the iterated GMM with optimal weighting matrix and the one-step GMM (introduced by Cochrane (1996)). The second GMM specification exhibits clearly, as suggested by Cochrane (1996), estimates with much lower average pricing errors. More interestingly, we report that the return on real estate wealth is a priced risk factor. Thanks to our sub-periods analysis, we are able to show that this neglected factor is useful to explain the cross-sectional variation of returns, especially for the last two decades associated to the so called «real estate bubble». |