Estimating the conditional CAPM with overlapping data inference

Asset pricing models such as the conditional CAPM are typically estimated with MLE using a monthly or quarterly horizon with data sampled to match the horizon even though daily data are available. We develop an overlapping data inference methodology (ODIN) that uses all of the data while maintaining the monthly or quarterly forecastingperiod, and we apply it to the conditional CAPM. Our approach recognizes that the first order conditions of MLE can be used as orthogonality conditions of GMM. We simulate from GARCH and MIDAS models and examine the substantial reductions in standarderrors and increases in power that arise from our methodology. Using historical data, we find considerable differences in the estimates from the non-overlapping samples that begin on different days. Using our overlapping data inference, we find a significant risk-return trade-off in the monthly data from 1955 to 2011 with a symmetric GARCHmodel.