Dynamic Analysis of the Housing Market Risk Premium Using a CAPM-Based EGARCH-in-Mean Model

This study empirically analyzes the dynamic characteristics of systematic risk and risk premiums in the Korean housing market using a capital asset pricing model (CAPM)-based exponential generalized autoregressive conditional heteroskedasticity (EGARCH)-in-Mean model. The market portfolio is constructed by weighting financial and housing assets according to household asset shares, thereby reflecting the actual asset composition of Korean households. Using monthly apartment sale price indices from November 2003 to December 2025, the empirical results show that market beta β is positive and statistically significant across all regions, confirming that the Korean apartment market is structurally linked to the financial market. The risk premium is significantly associated with volatility and exhibits time-varying dynamics. While the nationwide and metropolitan housing markets show a positive relationship between volatility and the risk premium, consistent with the conventional risk-return trade-off, the Seoul market exhibits an asymmetric response, with both volatility and the risk premium increasing following positive (price increase) shocks but declining after negative (price decrease) shocks. This pattern reflects a structural feature of the Seoul market, where contractions in transaction activity and adjustments in market expectations tend to precede price corrections. In the forecasting analysis for Seoul, the EGARCH-in-Mean model outperformed both the Random Walk and conventional CAPM models. Overall, these findings suggest that the Korean housing market exhibits characteristics of a financial asset market closely integrated with the broader financial market, while regional markets display heterogeneous risk structures and price adjustment mechanisms. The results further imply that housing market stabilization policies should consider not only price levels but also the time-varying nature of volatility and risk premiums.