Estimating MIDAS Regressions via OLS with Polynomial Parameter Profiling

Abstract

A typical MIDAS regression involves estimating parameters via nonlinear least squares, unless U-MIDAS is applied - which involves OLS - the latter being appealing when the sampling frequency differences are small. In this paper we propose to use OLS estimation of the MIDAS regression slope and intercept parameters combined with profiling the polynomial weighting scheme parameter(s). The use of Beta polynomials is particularly attractive for such an approach. The new procedure shares many of the desirable features of U-MIDAS, while it is not restricted to small sampling frequency differences.

Keywords: Mixed frequency data, MIDAS regressions, profile likelihood

JEL Classification: C13, C22, C52, C53