Hello, I am working on a project right now where I have delta_u as the dependent variable (it is the difference of unemployment rate in t+1 and t) and delta_y as the independent variable (it is the real gdp growth rate). I am instrumenting delta_y with gsgro (real government spending growth rate), pop (indicator variable, = 1 if the president at that time was a democrat), and 2 lags of delta_y.

What I have to explore now is see if changes in delta_u are asymmetric to change in delta_y (i.e. delta_u rises more quickly during recessions that it falls in expansions). I know that an indicator variable (grdir =1 if the growth rate was positive) should be used here but I couldn't find anywhere how it is to be used in the context on a 2sls estimation.

Thank you.