Many people believe that one of the most important root causes of public corruption is “big government.” This view was perhaps captured most famously and most succinctly by Gary Becker, the late Nobel Laureate economist, who declared (in a couple of memorable op-ed headlines), “If you want to cut corruption, cut government” and “to root out corruption, boot out big government.” Professor Becker was not what you would call cautious or circumspect in advancing this claim: He insisted that “instituting large cuts in the scope of government is the only surefire way to reduce corruption,” and that without such cuts even the most well-intentioned anticorruption reforms and crackdowns would fail, because “corruption always reemerges wherever governments have a major impact on economic conditions.” Though Professor Becker was perhaps the most blunt (and famous) advocate for this view, many others have taken this position. (See here, here, here, and here.) Indeed, a while back I attended an anticorruption conference at which a former senior minister of a European country (whose identity I cannot disclose due to the conference’s confidentiality rules) declared that the key to reducing corruption in his country was the decision to drastically shrink the public sector, slashing taxes, public spending, and the overall size of government–and this ex-official called on other countries to follow that advice as well.

But before we go charging ahead advising countries that the only way that they can get their corruption problem under control is to cut their governments, it might make sense to assess whether the available empirical evidence actually supports Becker’s hypothesis. Is it true that (all else equal) countries with larger governments have more corruption, compared to countries with smaller governments?

The answer is no. If anything, the evidence cuts in the opposite direction.

There’s actually been a fair amount of cross-country empirical research on this, and the weight of the evidence suggests that government size—typically measured by the ratio of government spending to GDP—actually seems to have a negative correlation with (perceived) corruption, as measured by the various international indexes, such as the Corruption Perception Index (CPI) and Worldwide Governance Indicators (WGI) graft index (see, for example, here, here, here, here, and here). That is, those countries with a higher government spending/GDP ratios tend to get better scores, not worse scores, on the CPI, WGI, or similar corruption indexes. Other studies measure government size with the Fraser Institute’s “government size” index—which includes not only government spending (in general and in particular categories, such as transfers and subsidies) but also the role of state enterprises in the economy, state investment, and top marginal tax rates—and similarly find, contra Becker, that countries with “worse” Fraser “government size” scores (that is, larger governments) tend to have lower perceived corruption (see here and here).

To be sure, there are questions about the robustness of this negative correlation: some studies that find the negative correlation between government spending/GDP is sensitive to the particular corruption index used, or the particular set of control variables (see here, here, and here), and a few studies do find some (weak) evidence of a positive correlation between government spending/GDP and perceived corruption (see here and here). Moreover, a number of other studies that find no statistically significant correlation at all between government spending/GDP and perceived corruption (see, for example, here). Yet while the evidence is not entirely uniform, and all of the studies here have their limitations and methodological difficulties. overall it seems fair to conclude that the weight of the evidence seems to support the view that, on the whole, larger government—again, measured principally by government spending as a percentage of GDP—is correlated with lower perceived corruption, not higher perceived corruption.

If this is indeed the case, what’s the explanation? The short answer is that I’m not sure, but I’ll offer some tentative speculations in my next post.