Time-Varying Risk Premium in Large Cross-Sectional Equity Datasets

We develop an econometric methodology to infer the path of risk premia from large unbalancedpanel of individual stock returns. We estimate the time-varying risk premia implied by conditional linearasset pricing models where the conditioning includes instruments common to all assets and asset specificinstruments. The estimator uses simple weighted two-pass cross-sectional regressions, and we show itsconsistency and asymptotic normality under increasing cross-sectional and time series dimensions. Weaddress consistent estimation of the asymptotic variance, and testing for asset pricing restrictions inducedby the no-arbitrage assumption in large economies. The empirical illustration on returns for about tenthousands US stocks from July 1964 to December 2009 shows that conditional risk premia are large andvolatile in crisis periods. They exhibit large positive and negative strays from standard unconditionalestimates and follow the macroeconomic cycles. The asset pricing restrictions are rejected for the usualunconditional four-factor model capturing market, size, value and momentum effects.[...]