This code allows estimating the parameters of a Panel Threshold Regression (PTR) model with one or two thresholds parameters. The model does not exactly correspond to that proposed by Hansen (1999). All the slope parameters are affected by the regime. In this model, you can consider contemporary exogenous variable. The code does not automatically introduce some lags on the threshold variable and the explicative variables. If you want to introduce such lags, you have to introduce lagged data in the form. The results display the estimated slope parameters and thresholds parameters. The F-tests F1 and/or F2 (test on the number of regimes) are also displayed. The corresponding Boostrap pvalues are displayed if the chosen number of simulations is greater than 0.
International Journal of Research in Marketing (2018)
Paas Leonard, Dolnicar Sara, and Karlsson Logi