Recent evidence suggests that the variation in the expected excess returns is predictable and arises from changes in business conditions. Using a multifactor latent variable model with time-varying risk premiums, we decompose excess returns into expected and unexpected excess returns to examine what determines movements in expected excess returns for equity REITs are more predictable than all other assets examined, due in part to cap rates which contain useful information about the general risk condition in the economy. We also find that the conditional risk premiums (expected excess returns) on EREITs move very closely with those of small cap stocks and much less with those of bonds.Recent evidence suggests that the variation in the expected excess returns over time is predictable and is the result of changes in business conditions. 1 We offer further evidence on this issue by extending the previous literature to include real estate, particularly equity real estate investment trusts (EREITs). 2 What is unique about EREITs is that it is traded as a stock on a stock exchange but represents an underlying ownership in a portfolio of real estate. This feature raises the possibility that different variables may be required to capture the time variation in its risk premiums relative to those for bond and non-REIT stocks. Another issue related to the hybrid nature of EREITs is whether EREITs are a hybrid of stocks and bonds and whether the stock component is representative of large cap stocks or small cap stocks. More specifically, the questions addressed in this article include: (1) Do the same variables forecast stocks, bonds, and real estate returns so that the expected returns (conditional risk premiums) on these assets move together? In particular, do cap rates carry information about the conditional risk premium for equity REITs but no other asset class? (2) Is the variation in the expected returns on equity REITs related to business conditions? (3) To what extent do REITs resemble stocks with large capitalizations, stocks with small capitalizations, and bonds?While Mengden and Hartzell (1986), Giliberto (1990), and Corgel and Rogers (1991), among others, have studied the hybrid nature of REITs in the past, none of these studies