Independence Status of Territories and the Estimated Trade Effects of Regional Trade Agreements

Almost 150 new sovereignties came into existence since 1945, which is about two times the amount of independent countries by then. New individuals are included by panel datasets based on country level data. In this paper I investigate the impact of panel unbalancedness related to independence status of territories on the estimation of trade effects of regional trade agreements. I first find that the inclusion or exclusion of newly independent countries affects estimation of interest significantly, so future empirical studies using country level data are recommended to verify more seriously about why including some newly independent countries or dependent territories but not others in their samples. Second, although changing independence status of territories raises up the question of interpreting trade data as “domestic” or “international” to (previous) governing countries (Greaves, 1954), I find that both ways of interpreting trade data yield similar estimates of most regional trade agreements. Third, it is shown that unless we are particularly interested in the trade effects of regional trade agreements formed before 1990, we should not include observations before this time into the sample. At last, the Monte-Carlo simulation shows that panel unbalancedness caused by emergence of new countries has different impact on the estimation of interest from missing at random or missing at minimum values.