ates its undiversi able market risk exposure. Past research has consis- tently shown that REITs have low betas with respect to market proxies that are often equity indices, such as S&P 500 and CRSP indices. As a result, investors perceive REITs as a less risky investment vehicle that ra- tionalizes the low or moderate compensation in terms of expected returns. However, using equity indices as market proxy leaves part of the market risk to remain diversi able as the market portfolio becomes more com- plete that incorporates various assets besides equity. A natural question that follows is will using a more diversi ed market proxy matter for the estimation of required rate of return that compensates the systematic risk exposure of REITs . In this paper, we construct a market portfolio in the spirit of Roll (1977) that consists of equity, xed-income securities, and real estate, and test if the REITs' risk premium estimated using equity index alone is robust to the misspeci cation of the market portfolio. Our results show that REITs betas are signi cantly increased as more com- plete market proxy is used. However, the market risk premium estimation of REITs does not vary signi cantly as the market proxy becomes more complete over the whole sample period from 1990 to 2008. There is con- sistent evidence that, on average, using equity market proxy understates the riskiness of REITs by less than 3 basis points on a monthly basis over the 2001-2008 sub period. Our results stand after using a survivor-bias free sample of REITs