Measuring high-frequency income risk from low-frequency data

We estimate a monthly income process using annual longitudinal household-level income data, in order to understand the nature of income risk faced by households at high frequency, and to provide an input for models that wish to study household decision-making at higher frequency than available data. At both frequencies, idiosyncratic earnings shocks have a highly persistent component. At monthly frequency, transitory shocks account for most of the earnings variance; at annual frequency, the persistent component is dominant. We apply our estimates in the context of a standard incomplete-market model, and show that decision-making frequency per se makes a small difference.