Tests of asset pricing with time‐varying factor loads

We test the empirical validity of the stochastic discount factor (SDF) theory. We develop a model where the SDF theory holds when the parameters that relate the risk premium and the factors are common across assets. This property is then statistically assessed through slope homogeneity tests for panel data for the mean and quantiles of the assets' risk premium. Applications to different sets of assets confirm the empirical validity of the SDF theory for the average and central quantiles of the risk premium distribution, but not at the tails. These pricing anomalies suggest that assets are priced independently of the rest of the cross-section during boom and bust periods.