Time-Aggregation Effects on Estimating Asset Pricing Models

The main goal of the paper is to characterize the impact of temporal aggregation on the estimation of the preference’s parameters of a representative agent in a consumption based asset pricing model (CCAPM). We assume that the true economy is at a high frequency (say monthly), that is we model the endowments and the preference of representative agent and then we characterize the asset prices at the high frequency. We then assume that the econometrician observes data at a lower frequency (say annual) and he postulates a model for the endowments and the preferences at the high frequency and he uses the data (implied by the high frequency economy) to estimate the parameters of the model at the low frequency. We are then able to characterize the mapping between the preference’s parameters at low frequency with those of the high frequency. This mapping is analytical in many examples. The model is misspecified at a low frequency, therefore the estimated parameters will depend on the statistical method and the moments that are matched. We therefore follow the CCAPM literature by focusing on matching the means of the risk free rate and the excess returns.