Estimating and Testing Long-Run Risk Models: International Evidence

Abstract

We estimate and test long-run risk models using international macroeconomic and financial data. The benchmark model features a representative agent who has recursive preferences with a time preference shock, a persistent component in expected consumption growth, and stochastic volatility in fundamentals characterized by an autoregressive Gamma process. We construct a comprehensive dataset with quarterly frequency in the post-war period for ten developed countries and employ an efficient likelihood-based Bayesian method that exploits up-to-date sequential Monte Carlo methods to make full econometric inference. Our estimation provides international evidence in support of long-run risks, time-varying preference shocks, and countercyclicality of the stochastic discount factor.

Keywords: Consumption-based Asset Pricing, Long-Run Risks, Stochastic Discount Factor, Equity Premium Puzzle, Autoregressive Gamma Process, Projection Methods, Sequential Monte Carlo

JEL Classification: C11, C32, C58, E44, G12