Economists look at more than GDP when choosing countries to study
ECONOMIC RESEARCH can reverberate beyond the ivory tower. In 2003 a study of Kenyan schools found that treating intestinal worms improved attendance. After similar work confirmed the policy’s benefits, one author, Michael Kremer, founded an NGO that treats 280m children a year.
Mr Kremer’s work was unusually impactful, but reflects a pattern of research improving policy. One study found that telling Brazilian mayors about the gains from sending reminder letters to taxpayers sharply increased their chances of doing so. Yet many similar countries attract far fewer studies. This can leave policymakers fumbling in the dark (see Free exchange).
To measure this problem, we turned to EconLit, a database curated by the American Economic Association with 910,000 journal articles from 1990-2019. It only tracks papers with abstracts in English, the field’s lingua franca, causing it to under-represent studies intended for non-Anglophone audiences. However, EconLit does include 110,000 papers in other languages with abstracts translated into English.
By far, the best predictor of the amount of research conducted on a country was its GDP. However, economic size leaves many cases unexplained. Kenya gets three times more articles than its GDP suggests; Algeria has one-quarter as many as expected.
Such outliers often cluster in research “oases” or “deserts”. Obie Porteous of Middlebury College notes that studies of Africa are disproportionately concentrated in the continent’s south and east. Expanding this analysis worldwide, we find that the Middle East and parts of Latin America get relatively few papers with English abstracts. China and Russia also seem under-studied.
In contrast, South Asia and some regions in eastern Europe were oases. Like much of southern and eastern Africa, India and Pakistan were colonised by Britain. Today, many authors of articles about them work in Britain or America. Meanwhile, European research gluts seem locally driven. Lots of studies on Slovenia, which has one of EconLit’s highest papers-to-GDP ratios, stem from universities in Maribor and Ljubljana that churn out articles in English.
To adjust for such factors, we built a statistical model to predict a country’s share of studies in each year. GDP remained the most important variable, though it mattered less in oil-rich states. The next-best predictors of popularity in the Anglophone database were listing English as an official language and sending lots of students to American universities (boosting places like China). Variables that capture data availability, such as the number of World Development Indicators a country publishes, also had meaningful effects.
These factors improved the model a lot. They explained most of the difference between Kenya and Algeria, for example. After incorporating them, we found that a country’s spending on universities, form of government and involvement in armed conflicts did not yield additional accuracy.
For policymakers in research deserts who want academic support, that is good news. In the short term, they can do little to boost national GDP significantly. But being more forthcoming with data and fostering links with Western scholars should help. ■
Sources: American Economic Association; World Bank; Encyclopedia Britannica; Institute of International Education; government websites
This article appeared in the Graphic detail section of the print edition under the headline "Starving for knowledge"
- Natural Gas Outlook: LNG Price Action Approaches Long-Term Resistance
- Gold (XAU/USD) and Silver (XAG/USD) at Critical Levels ahead of FOMC
- The Best Performing Women Led Companies
- GBP/USD Bounce Fizzles Out Despite Higher Retail Sales in June
- Coffee Prices at 6-Year High as Extreme Cold Weather Hits Brazil