Results interpretation of a Panel estimation model

[Pilot013]

Hello dear colleagues!

I am writing a bachelor's thesis on the topic of Implications of public debt on economic growth. I have a panel of 17 countries, observed over 30 years on several variables. As I performed a Granger causality test, I obtained results that in fact GDP_growth causes debt_gdp fluctuation and not the opposite, as many people hypothesize.

So in that sense, my basic equation would be:

debt_gdp = c gdp_growth + u

debt_gdp values are represented as ratios. For example: 0.35,0.38,0.46.... where this means that that debt/gdp = 35%,38%,46%, etc...
gdp_growth is presented as the growth rate from year to year of GDP. For example: 0.02,0.03,-0.01, etc... gdp_growth = 2%,3%,-1%, etc...

Estimating a simple model for the period 1985-2010, excluding 5 countries because there is some missing data, I obtain the following results:

Regression of debt_gdp on gdp_growth.

Capture1.JPG (125.89 KiB) Viewed 2842 times

I get a bit confused how to interpret correctly ratios and what is the meaning of "-1.51", standing for Beta1. Does that mean that 1 % point increase in gdp growth would lead to a -1.51% point reduction in the debt/gdp ratio or? Furthermore, when I take a look at the country effect, it gets even more confusing for me:

Cross-section Fixed effects

Capture2.JPG (58.23 KiB) Viewed 2842 times

Taking a look for Portugal, it seems that if gdp_growth is 0%, then Portugal's constant would be -0.18, implying that the country should be a NET LENDER to World, lending 18% of its GDP. This doesn't make any sense, as their Debt/GDP now is around 0.90 and growing.

I got totally confused with the interpretation of those results and how to interpret the debt/gdp ratio correctly when used in a panel regression. Furthermore, I have also calculated the growth rate of debt in the following way: log(debt_gdp)-log(debt_gdp(-1)). Perhaps I should also try to predict the growth rate of debt based on economic growth, bb_gdp(budget balance as a ratio to GDP) and some other explanatory variables that may have an effect.

I hope you would be able to share with me your suggestions and ideas.

Thanks in advance!