I wanted to estimate a difference-in-differences model using Stata looking at the effects of a trade shock (in 2007) on households' income. I have a repeated cross-sectional data for years 1995-2015. So, I estimated this model:
reg income Treat##Post i.year
which Treat is a dummy variable (1 for treated group and 0 for the control group), Post is a dummy variable (1 for years after the shock and 0 for years before the shock). I included year fix effects (i.year) to control for time-varying macroeconomic changes. Treated households used to be richer than the control group before the shock, however, their income trends were parallel (so, their income differences are not zero before the shock). The coefficient of interest (Treat*Post) is significantly negative.

I am also interested to find the effect of the shock for each year because I believe that the effect of shock has decreased over time. So, I estimated this model:
reg income ib2006.year##i.Treat

I have two questions regarding defining the base year:
(1) I define the year before the shock (2006) as the base year. It is assumed there is no difference between these two groups in term of income in the year 2006 which is not correct, because as I said before treated households used to be richer than the control group before the shock, so their income differences are not zero before the shock.

(2) Although Treat*Post coefficient is significant in the first model, Treat*year2007-Treat*year2015 are not significant in the second model. Why?
(if I change the base year to 2007, the coefficients will be significant because all values shift down.)