I have the following problem:

y= intercept + y(-1)+ x1 factors + x2 factors which the literature models using dynamic panel models (since y depends on its lagged value)

where

y(-1) is the lagged y variable

x1 factors are firm specific

x2 factors are macroeconomic factor

I want to investigate the effects of y falling below a certain value on x1 variables. In other words assume that a benchmark=100 when y<100 I want to see how the firm reacts and changes its x1s.

How can i model this problem? which regression or series of regressions should I use?