Detecting random coefficients in a periodic multiple regression model

This paper presents a pseudo-Gaussian test for detecting random coefficients in a periodic multiple regression model applied to unbalanced panel data. The test is constructed based on uniform local asymptotic normality results. Simulation studies confirm the validity and effectiveness of the proposed test under both symmetric and asymmetric distributions. Furthermore, applications to real data demonstrate the practical advantages of the proposed model over the periodic regression model introduced by Regui et al. [Locally optimal tests against periodic linear regression in short panels. Commun Stat Simul Comput. 2024;54(6):1–15].