I am regressing real loss triangles and would like to examine the possibility of discrete changes in both the inflation and decay parameters. I am not quite sure how this model should be set up.
Option 1: 2 independent dummy regressors, each interacting with only 1 of the quantitative explanatory variables.
Option 2: 3 dummy regressors representing 4 mutually exclusive domains, 2 of which interact respectively with 1 or the other of the quantitative explanatory variables and the 3rd of which interacts with both.