Do African resource rents promote rent-seeking at the expense of entrepreneurship?

Survey-based responses from the Global Entrepreneurship Monitor (GEM) provide data on respondents having the intention of starting a business, being in the process of starting a new business, or being engaged in early-stage entrepreneurial activity. However, GEM data does not measure formal and informal entrepreneurship separately and it can also easily overstate the rate of entrepreneurship if some individuals who claim to be in the process of starting a business ultimately fail to do so.

It should be noted that natural resources could also be a curse in other ways related to resource scarcity (see for example Collier and Laroche (2015)) and geographical disadvantage due to inadequate rainfall, poor soil, tropical diseases, etc. While acknowledging that these factors are important, the resource rents measure is used as this paper is more focused on the rent-seeking channel associated with resource rents.

To verify that this is an appropriate and informationally efficient way to aggregate these underlying measures, principle component analysis (PCA) was performed to determine the variance maximizing linear combination of the governance measures. The resulting weights are very similar to the uniform weights from the simple average. Indeed, when the overall institutional quality is calculated both ways (i.e., the simple average and the weighted average according to the PCA weights), the correlation coefficient between the two series is 0.99. Therefore, overall governance/institutions are measured by way of the simple average of the underlying six governance measures.

We would expect the adverse effect of natural resource dependence on entrepreneurship to be more pronounced for point-source resources such as oil, diamonds, and minerals, which are found in dense concentrations and generate large rents that are easily appropriable, than for resources such as agricultural commodities, whose rents are more dispersed within different regions of a country. However, when more disaggregated data on resource rents are used, the sample size is reduced significantly. For example, estimating a separate regression using oil rents only results in a sample size of 10 countries and 86 observations in total. Although the results are consistent with those obtained using the aggregate measure of resource rents, they are no longer significant due to loss of statistical power.