I am currently looking to study the profitability of banks. My 4 independent variables, which are taken from banks balance sheet, are just financial ratios. My time is 13 Years, and I have 21 banks. My dependent variable is return on equity. I first chose to use an individual and time fixed effects model where my formula is:
where $$s_t=time FE$$ and $$a_i=individualFE$$
My standard errors are biased, but I am wondering should I use a dynamic model with a lagged dependent variable, or stick with my FE model and correct for the SEs, or could you suggest what you believe would be a better model?