In this paper, I have adopted some methods applicable to the panel data analysis. The methods are suggested by Wooldridge and Dougherty. Therefore, some methods in panel data analysis such as pooled OLS and Fixed-Effect estimations are employed in this paper.
Pooled OLS Estimation
The pooled OLS estimation in this paper is used to estimate the baseline regression model to find the overall effect of the observed variables on economic performance of the countries in this sample. The model specification from Dougherty (2006:411) can be written as follows: yit = b0 + b jxitj + ykZkt + uit
In equation, y can be supported for member i at time t by j number of observable variables x and k number of unobservable variables z and an idiosyncratic error term uit.. If z is supposed to be fixed over time, then the unobservable effect will capture z.
The second estimation method adopted here is the fixed-effect model since I attempt to find the truth whether the effect of foreign trade on economic growth has changed over time. If the unobserved effect ai is supposed to be correlated with one of the independent variable the term Cov(xit,ai) = 0 is violated, and the fixed-effect estimation should be employed; Cov(xitj,at) □ 0. Following Wooldridge (2009:482), the model follows that:
ytt – y = □ (bk (xijt – xik )) + urt – u
If we subtract the time average in equation, the time invariant a, is naturally dropped from the equation as ai – a. In this paper, Hausman Test is also used to test the model suitability before the final model is decided.
The Model Specification
The Baseline Regression Model in this paper will look like follows:
log(GDP) = b0 + b log(labor) + b2 log(capital) + b3 log(trade) + b4 Institutionalstocks + b5totaldomesticcredit+uit
where: Institutional Stocks here includes the total stock of Rule of Law, Government
Effectiveness, Accountability and Political Stability
Total Domestic Credit includes Domestic Credit Provided by Banking Sector plus Domestic Credit to Private Sector in economy.
To define the roles of Institutions in economic performance, the interaction terms of the Pooled OLS are presented. Suggested by Jansen and Nordas (n.d.), the roles of institutional indicators should be defined separately in the regression. So, I decide to run an individual regression for each institutional indicator. Four institutional indicators are covered in this paper, namely, Rule of Law, Government Effectiveness, Accountability and Political Stability.
* The Role of Rule of Law in Economic Development while controlling other variables.
log(GDP) = b0 + bi log(labor) + b2 log(capital) + b3 log(trade) + b4totaldomesticcredit + b5ruleoflaw+ b6 int eractionterm +uit
* The Role of Government Effectiveness
log(GDP) = b0 + b1 log(labor) + b2 log(capital) + b3 log(trade) + b4totaldomesticcredit + b5governmenteffectiveness+ b6 int eractionterm +uit
* The Role of Accountability and Voice
log(GDP) = b0 + b1 log(labor) + b2 log(capital) + b3 log(trade)+ b4totaldomesticcredit + b5accountability+ b6 int eractionterm +uit
*The Role of Political Stability
log(GDP) = b0 + b1 log(labor) + b2 log(capital) + b3 log(trade) + b4totaldomesticcredit + b5politicalstability + b6 int eractionterm +uit