Interpreting Decade Dummy Data in a Fixed Effects Panel Data Regression

[Thomas Lawrence]

Hello,

I am researching the impact of growth of education on the growth of productivity. I have a sample of 90 countries from 1970-2019. I am running a panel data regression where I regress growth_e (growth of education) and growth_c (growth of capital stock) on growth_tfpw (growth of productivity), I have added 4 dummy variables, 1980, 1990, 2000 and 2010.

I appreciate the coefficients are relative to the ommited fixed effect (1970s) however I would like some insight into how I interpret these coefficents. As I have two independent variables surely I cannot understand the values to be the growth of education's impact on productivity. What do these coefficients then mean?

Moreover, only the 1990s and 2000s are statistically significant, does this mean the other decade coefficients should be disregarded?

Thank you for any help

[Carlo Lazzaro]

Thiomas:
the first advice is to gather all your decades in a single -timevar-.
Generally speaking, the -timevar- effect explains the contribution of time to within-panel variation, when adjusted for the other predictors.
More substantively, it can be that you do not have a panel-data effect, as the two sigmas overlap.