| Summary: | The role of liquidity in asset pricing model has attracted much attention in recent financial studies; however there is a paucity of literature with respect to liquidity risk and asset pricing in the direct housing market. The housing market is characterized by costly searching, inelastic supply and short-sale constraints. It is expected that the housing market should incur more significant illiquid effects, since it is much more illiquid than stock market. Motivated by this intuition, this thesis aims to explore 1) whether and to what extent liquidity can explain variations in over-time/crosssectional housing returns; and 2) whether liquidity factor plays a role in explaining the second moment (i.e. volatility) of housing price.
We employ the panel regression and Fama-MacBeth two-stage procedure to investigate over-time and cross-sectional relations between liquidity and housing return. Liquidity asset pricing theory suggests that assets with lower degree of liquidity offer higher expected returns. Consistent with this prediction, the panel regression results suggest that housing return is a decreasing function of liquidity in previous year, while it is positively relative to contemporary liquidity shocks. For the cross-sectional asset pricing tests, housing estate specific betas are estimated using rolling time-series regressions of a three-factor asset pricing model. We investigate the proposition that housing estates with greater return sensitivity to market liquidity earns higher expected return. Using a disaggregate dataset of 55 popular housing estates, we find (1) both market liquidity beta and housing estate specific liquidity risk are significantly priced in the cross-sectional housing estate returns, implying that cross-sectional differences in estate premium partially represent the liquidity premium. (2) The market beta, sentiment beta and market liquidity beta explain 14.36% of variations in cross-sectional estate returns. The results are robust across different specifications. (3) Investors are less willing to bear liquidity risk during the down markets, which shed new light on the positive price-volume correlation. These findings complement the cross-sectional liquidity-return relationship in the financial literature.
Measuring housing price volatility is fundamental to the study of the dynamics of housing price risk. We investigate the effects of liquidity on housing price volatility in different housing classes (classified by size of the housing unit according to the Rating and Valuation Department’s definitions). Housing price volatility is measured as the conditional variance of a GARCH model under the Adaptive Expectations framework. We reveal that volatility transmits from small housing units to large housing units, which indirectly supports the trade-up effect in previous literature.
Besides, the starter and high-end housing classes are extraordinarily sensitive to negative and positive liquidity shocks respectively. Consistent with the friction search theory, we find that the pricing errors are alleviated as the trading volume increases, since the valuated price tends to be more accurate as more information arrives. Lastly, the variance decomposition and impulse response results imply that the positive liquidity shock accounts for a large proportion of variations in housing volatility. === published_or_final_version === Real Estate and Construction === Doctoral === Doctor of Philosophy
|
|---|
