Anticorruption advocates often argue that the fight against corruption is not just about strengthening systems for detecting and punishing corrupt behavior, but about implementing broader systemic reforms to policies and institutions that create the conditions in which systemic corruption is more likely to take hold. That advice is sound as far as it goes—but the challenge then becomes identifying those policies and institutions that have this “corruptogenic” character. One prominent hypothesis in this vein is that corruption thrives in environments where there is a lack of “economic freedom”—where the government plays an outsize role in the economy, imposes lots of burdensome regulations on private enterprise, does not provide effective protection for private property and contract rights, and generally restricts economic activity. This idea (which is perhaps especially attractive to those who favor a limited government role in the economy for other reasons) is certainly plausible. But is it true?
Proponents of the idea that a lack of economic freedom leads to more extensive corruption can point to a substantial body of cross-country research that purports to find a strong negative correlation between economic freedom and corruption. Most of this research measures (perceived) corruption using one of the familiar international indexes, most commonly Transparency Internationals’ Corruption Perceptions Index (CPI). The research in this vein also measures “economic freedom” using indexes produced by NGOs—the most widely-used of which is the Heritage Foundation’s Index of Economic Freedom (IEF), which aggregates a number of variables thought to be related to economic freedom, grouped into four different categories (rule of law, limited government, regulatory efficiency, and open markets). Numerous studies have found a strong and statistically significant correlation between the IEF and the CPI, and treated this as strong evidence that a lack of economic freedom is at the very least associated with, and most likely causes, more widespread corruption (see here, here, here, here, here, and here).
Unfortunately, these results tell us precisely nothing. Put aside the standard admonition that we can’t infer causation from correlation. Put aside the concern that “economic freedom” may not be a coherent concept, and that the Heritage IEF aggregates a large number of disparate factors. And put aside worries about whether these studies control for potential additional variables that might influence both corruption and economic freedom. The fatal flaw in drawing any inferences at all from the correlation between the IEF and the CPI is in fact much more straightforward:
The flaw is simply that the Heritage IEF uses the CPI as one of its components. As part of the “rule of law” category, the IEF includes “freedom from corruption,” which is measured simply by the CPI. (Earlier versions of the IEF confusingly labeled this component “black market activity” or the “informal market,” but still used the CPI as the primary measure.) In other words, the studies showing a correlation between the CPI and the IEF are merely showing a correlation between the CPI (on its own) and the CPI (averaged with a bunch of other variables). It would be shocking if there weren’t a statistically significant correlation between these variables—that would only happen if the other variables thrown into the IEF were had a sufficiently strong inverse correlation with the CPI (either systematically or through random error).
Of course, given that the IEF includes scores in different subcategories, it would be possible to see if these individual subcategories (other than the CPI) are correlated with the CPI—or, if the correlation between the CPI and the aggregate IEF score holds up when the latter is recalculated with the CPI excluded from the set of components. I haven’t had time to do this myself, and relatively few of the existing empirical studies have done so. For what it’s worth, the few that have seem to reach much more equivocal results—sometimes finding statistically significant correlations with the CPI, but often not (see here, here, here, and here).
Oh, and by the way, the other major aggregate index of economic freedom—the Fraser Institute’s “Economic Freedom in the World” (EFW) project—suffers from a similar problem. Although the EFW index does not include the CPI as one of its components, it does include (as part of the “business regulation” sub-category) data from the World Economic Forum’s Global Competitiveness Report survey on the frequency and impact of bribery. So, although studies that show a correlation between the Fraser Institute EFW and the CPI (or other measures of perceived corruption) don’t have the obvious circularity problem of studies that correlate the CPI with the Heritage IEF, they still, in effect, show us that (one measure of) corruption is correlated with (another measure of) corruption (averaged together with a panoply of other variables). So the numerous studies that find a correlation between the EFW and CPI (see here, here, here, and here) actually tell us precious little about whether economic freedom is in fact associated with higher perceived corruption. And the handful of studies that have explored the association between perceived corruption and subcomponents of the EFW (subcomponents that do not include the bribery measure) have tended not to find much in the way of statistically significant correlations between the measures that capture government intervention in the economy and perceived corruption (see here and here).
None of this necessarily means that economic freedom (or at least some of the various components grouped together in these economic freedom indexes) is unrelated to corruption. But it does mean that we can and should completely ignore the correlations between the main international economic freedom indexes (the Heritage IEF and Fraser EFW) and the main perceived corruption indexes.