Small sample regression robustness

[eviewsuser2015]

Dear all,

I'm currenly trying to estimate a regression with Y = yeardummy + beta X + e in Eviews 8. My sample is only three years, with n=30, so 90 obs. total. Because the sample is only small (data on Y is limited and the total population is around 200) I am worried about the robustness of my estimates.
What could I do to estimate the equation under more 'relaxed' restrictions to make it more robust?

Thanks in advance!