This Matlab code allows estimating the parameters of a Panel Smooth Transition Regression (PSTR) model. In this code, the panel can be unbalanced. The user can define the number of location parameters see Gonzalez et al., 2005 for more details) and the maximum number of transition function. The code automatically determines the optimal number of transition functions, by testing the hypothesis of no remaining heterogeneity, using a 5% nominal risk. The parameters (slope parameter and location parameters of the transition function, slopes parameters in each regime for all the explicative variables…) are estimated by NLS. At the end, the individual elasticities for each explicative variable are computed and stored in an excel file.
Scaillet Olivier, Gagliardini Patrick, and Ossola Elisa