Re-testing the Resource Curse Hypothesis Using Panel Data and an Improved Measure of Resource Intensity

Abstract

Most empirical studies of the Resource Curse Hypothesis (RCH) find evidence of a strong negative relationship between a country's natural resource abundance and economic growth. We question the reliability of these findings in relation to the definitions and measures used for both resource intensity and economic growth, and the econometric testing which we consider deficient. We use an alternative, per capita resource rents measure of resource intensity that excludes renewable resources and avoids the circularity and bias of other output-related measures. Using Cluster Analysis, we compare the grouping of countries on the basis of three resource intensity definitions; viz. Sachs and Warner's (1995), Gylfason and Zoega's (2002) and our per capita rents, and find substantial differences in the resulting clustering. We then re-test the RCH econometrically using panel data and the corresponding econometric technique, as opposed to a single period-average data cross-sectional model. We test the sensitivity of our results by estimating the same model with the three different definitions of resource intensity. Previous Sach's and Warner's findings still hold when using their measure of resource intensity in our different modeling framework, while those of Gylfason and Zoega do not. Using our per capita rents measure of resource-intensity we find evidence of a positive relationship between natural resource abundance and economic growth. We conclude that the results of the test for the RCH are dependent on both the definition of resource-intensity and whether the growth is modeled with cross-sectional or panel data.