Help with Panel Regression model..

#1
Hi all. Kind of a specific question. But I'm not sure where to start, mostly because my knowledge on econometrics sucks.

I'm analyzing effects of 3 diff healthcare insurance enrollment on healthcare inequality in China. For inequality, we're looking at both medical care utilization, and catastrophic health expenditure (both will be dichotomous) between poverty and non-poverty groups (diff definitions here). So:

Insurance 1, compare utilization and expenditure between non-pov and pov groups
Insurance 2, compare similarly
Insurance 3, compare similarly

I have a panel data set which seems relatively robust to handle this analysis. 1991-2011 data, which straddles over the years whenthe insurance was introduced. My questions are:
1) Should I focus the analysis to only years *after* the insurance program was implemented? This would focus analysis only on those enrolled
2) On the other hand, if I focus years before and after, I could compare within these groups poverty and non-pov -- before and after enrolment, to see how enrolment changed their utilization and catastrophic health expenditure?
3) How does using the panel structure factor into this analysis at all, as opposed to just pooling the data?

Thanks.. ergh