Stay on the Safe Side: Liberalize!

Although there is an abundance of theoretical results in international economics and growth theory predominantly indicating a positive relationship between economic liberalization and economic welfare, confirming this prediction empirically has proven to be elusive. Yet, Andreas Tommaso Nannicini (Department of Economics) and Andreas Billmeier (Ziff Brothers Investments) quest and answer such an over-pending question by means of the state-of-the-art statistical methodologies in Assessing Economic Liberalization Episodes: A Synthetic Control Approach, forthcoming in the Review of Economics and Statistics.

The authors look into the impact of economic liberalization episodes on the pattern of income per capita across the world. An indicator that captures the scope of the market in the economy in terms of openness to international markets is used as a measure of economic liberalization. Consequently, empirical results suggest that in most regions under scrutiny economic liberalization had a positive effect on real GDP per capita. However, this finding bears a certain amount of heterogeneity of the effect across regions and time. More specifically, countries that have experienced liberalization after 1990- many of which are located in Africa- had no significant positive impact on GDP per capita in comparison to similar, but closed economies.

In order to answer such a question, previous studies have relied on cross-country analysis or country specific case studies. However, the authors employ a transparent statistical methodology for data-driven case studies, namely, the synthetic control method. Superiority of this methodology comes from being able to overcome the endogeneity problems present in cross-country estimators and it is more generalizable than individual country episodes. This methodology allows one to compare the post-liberalization income trajectory of open economies with the income trajectory of a combination of similar but closed economies, while controlling for time-varying unobservable variables. An advantage of this unified statistical framework is that it forces one to compare the income performance of open and closed economies that are similar in other covariates and in their past realizations of the GDP variable. This minimizes discretion in the analysis and leaves room for a more appropriate set of comparison units. Another advantage of this statistical framework is that it is able to deal with endogeneity problems stemming from omitted variable bias. It is an improvement upon panel data models such as fixed effects, which can only take into account time-invariant unobservable confounders, since it can account for the presence of time-varying unobservable confounders.

Lastly, the authors conduct two types of experiments on the control groups. In the first experiment, they restrict the control group (closed economies) to eligible countries in the same macro region of the treated (open economies), such as Asia, Latin America, Africa and Middle East. This allows the control group to be similar to the treated group in terms of the geographic factors and provides a more reliable comparison. In a second experiment, on the other hand, the authors make use of all the countries eligible in the control group, and this permits an increased sample size and power of the test.

In conclusion, Andreas Billmeier and Tommaso Nannicini apply a powerful statistical technique to tackle a challenging question and provide convincing evidence that economic liberalization raises income per capita, albeit, this effect is not uniform across regions and time.