Following Sachs and Warner (1995, 1997), the relationship between natural resources and growth has attracted extensive attention from economists and policymakers. One challenge in terms of understanding the relationship between natural resources and economic growth is that the literature is large and reaches different conclusions about the existence of a curse and, for papers that find a curse, the conditions under which a curse may exist. This paper uses new state-level panel datasets spanning 1880-2012 to investigate the relationship between natural resources and growth in the context of the American states.
The paper has four main findings. First, the relationship between growth and natural resources varies across types of natural resources – agriculture, fossil fuels, and other minerals – and over time. In some periods the relationship is negative and in other periods it is positive or not statistically significantly different from zero. Second, the effect of resources on growth differs depending on whether the change is an increase or a decrease in the resource and whether the economy is in a period of low growth or not. Third, for the period 1980-2000, a period that is widely studied, whether one finds evidence of a resource curse is highly sensitive to specification. Time series results differ depending on time intervals (decadal or annual), the type of natural resource, whether changes in resources are measured in total value or value added, and whether effects are allowed to differ in the South and non-South. Fourth, we can replicate many of the findings from previous studies of the resource curse in the United States. The divergent findings are largely due to the use of different dependent variables, measures of resources, estimation techniques, and time frames.