This paper provides a time-series and cross-sectional investigation of the conditional and unconditional capital asset pricing model (CAPM). The unconditional CAPM fails, but the conditional CAPM with dynamic conditional correlations (DCC) succeeds in generating a significantly positive risk-return tradeoff. The conditional alpha estimates indicate that the time-varying conditional covariances explain the industry, size and value premiums, but the momentum profits cannot be explained by the conditional CAPM. The multivariate GARCH-in-mean model with DCC provides an accurate characterization of the conditional betas that significantly covary with the expected market risk premium to explain stock market anomalies. The paper also examines the significance of intertemporal hedging demand identified by the covariation of portfolio returns with the innovations in macroeconomic variables. The results show that only inflation and dividend related shocks have significant risk premia.