Federal Reserve Bank at Kansas City

Under linear approximations for asset prices and the assumption of indepen-dence between expected consumption growth and time-varying volatility, long-run risks models imply constant market prices of risks and often generate counterfac-tual results about asset return and cash flow predictability. We develop and esti-mate a nonlinear equilibrium asset pricing model with recursive preferences and a flexible econometric specification of cash flow processes. While in many long-run risks models time-varying volatility influences only risk premium but not expected cash flows, in our model a common set of risk factors drive both expected cash flow and risk premium dynamics. This feature helps the model to overcome two main criticisms against long-run risk models following Bansal and Yaron (2004): the over-predictability of cash flows by asset prices and the tight relation between time-varying risk premia and growth volatility. Our model extends the approach in Le and Singleton (2010) to a setting with multiple cash flows. We estimate the model using the long-run historical data in the U.S. and find that the model with generalized market prices of risks produces cash flow and return predictability that are more consistent with the data. Key Words: Long-run consumption risks; Time-varying risk premium; Recursive preference