Idiosyncratic Risk and Private Real Estate Returns

The theoretical model of Merton (1987) predicts a positive relation between idiosyncratic risk and returns, for investors who are not fully diversified. Investors in the private real estate market hold particularly undiversified portfolios due to lack of information, transaction costs, liquidity requirements, taxes, etc.. Therefore, it is especially important to see whether private real estate returns are significantly related to idiosyncratic risk. The lack of research in the private real estate market due to the lack of high frequency data needed to construct measures of idiosyncratic risk. To overcome this problem we use the cross sectional variance (CSV) as our measure of idiosyncratic risk, as it is calculably at any frequency and is model free. Using monthly data for 35 real estate market segments over the period 1987:1 to 2019:12 the results indicate that CSV is highly correlated with idiosyncratic risk measured by the average variance of errors from the market model. Therefore, we consider CSV a good proxy for idiosyncratic risk in the private real estate market. Then using quantile regression methodology we find that there is a positive relationship in the higher quantiles but an insignificant negative effect in the low quantiles for average market returns 1, 3, 6, 9 and 12 months ahead. Lastly, we find high idiosyncratic risk portfolios produce significantly higher returns than low idiosyncratic risk portfolios. The results indicate that idiosyncratic risk significantly affects private real estate returns. The study therefore provides important implications for investors and fund managers, as well as researchers.