Housing consumption-based asset pricing and residential mortgage default risk

Housing is a macro asset category and has significant impact on the whole economy. In recent years, some consumption-based asset pricing (CCAPM) literature states that the optimal consumption-saving/investment decision depends not only on aggregated consumption but also on composition between housing and non-housing. This study adopts an Epstein-Zin recursive utility specification to set up a housing consumption-based capital asset pricing model (HCCAPM) which models the housing both as an asset and as consumption good, to study the impact of housing consumption and long-run consumption risks on asset pricing. The study introduces an equilibrium asset pricing model with housing and presents long-run risks and cointegration between expenditure share uncertainty and economic growth as other two factors that drive asset prices. The model reveals that the household not only concerns with uncertainty on overall consumption and consumption composition between non-housing and housing service, but also concerns with the long-run uncertainty of the consumption streams. This study analytically solves the pricing kernel in terms of state variables and derives equilibrium solutions for two systematic determinants of mortgage credit risk---the risk free rate of return and the real return of housing asset.In empirical part, the study utilizes the Unobserved Components Method (UCM) to decompose the unobserved exogenous state variables and GMM approach to estimate the preference parameters in the pricing kernel. Based on the estimation results, the study analyzes the mutual interactions between the interest rate and housing return in terms of state variables. This dissertation finds that households are not willing to substitute consumptions over time. It also provides potential explanations about why housing is on average more uncertain than other risky assets. Compared with the homeownership ratio, I argue that the housing quality index is an important alternative indicator of well-being the shield provides. The fact that the housing consumption per agent had declined since 1991 to 2008 raises questions about the sustainability of the U.S. homeownership policy.In the application part, I incorporate these two mutual interacted factors into an option-based proportional hazard model to forecast risks of FHA-insured mortgage default and prepayment. I find that this model could deliver quite accurate predictions of the impacts of macroeconomic dynamics on credit risks than the hazard model that assumes the risk free rate of return and housing prices are independent processes.Key Words: Housing consumption, asset pricing, interest rate, housing return, long-run risks.