Title: Asymptotic behavior of temporal aggregation in mixed-frequency datasets
Authors: Cleiton Guollo Taufemback - Universidad Carlos III de Madrid (Spain) [presenting]
Abstract: It is well known that discarding intermediate data in a time series mixed-frequency dataset, also called temporal aggregation, imply a loss of information. We demonstrate that temporal aggregation results in inconsistent estimates by showing how aliasing affects the linear regression model. We also propose a new method to circumvent this inconsistency. We analyze stationary and non-stationary linear semiparametric/nonparametric models. Monte Carlo simulations illustrate that the proposed method has good finite sample properties. Finally, an empirical application to quarterly GDP and monthly US indicators data highlights the applicability of our approach.