A statistical approach to identifying poorly performing countries

This paper asks whether it is possible to identify, using purely statistical criteria on widely available quantitative data, a set of developing countries that can be classified as poor performers. We restrict attention to two performance indicators, economic growth and infant mortality, over two periods 1980–90 and 1990–2000, and use four different statistical criteria to identify poor performance. The main finding is that very few countries consistently appear as poor performers: those that perform poorly on one indicator, or in one period, typically do not perform poorly on/in the other. A similar result is obtained in the context of identifying, on statistical grounds, good performers. The research cautions against labelling countries as poor performers without careful qualification.