Dynamic panels with MIDAS covariates: Nonlinearity, estimation and fit

This paper introduces Mixed Data Sampling (MIDAS) into the panel data context. To address unidentified nuisance parameter problem, we propose to invert model specification tests for inference on MIDAS along with bounds coefficients. Illustrative identification, simulation and empirical analyses are conducted in dynamic GMM framework. Our framework allows departures from i.i.d errors such as clustering specifications. A study an application a of reserve holdings illustrate usefulness proposed methods, more broadly set promising template shrinkage approaches.