The World Bank and IMF held their annual meetings last week, and it appears from the agenda that considerable attention was devoted to corruption—an encouraging sign that these organizations continue to treat this problem as both serious and relevant to their work. But does addressing the corruption problem effectively require that these organizations make more of an effort to quantify the problem? In a provocative post last week on Global Financial Integrity’s blog, Tom Cardamone (GFI’s President) and Maureen Heydt (GFI’s Communications Coordinator) argue that the answer is yes. In particular, they argue that the IMF should “undertake two analyses”: First the IMF “should conduct an annual assessment of grand corruption in all countries and publish the dollar value of that analysis.” Second, the IMF “should conduct an opportunity cost analysis of  stolen assets”—calculating, for example, how many hospital beds or vaccines the stolen money could have purchased, or how many school teachers could have been hired.
This second analysis is more straightforward, and dependent on the first—once we know the dollar value of stolen assets (or grand corruption more generally), it’s not too hard to do some simple division to show how that money might otherwise have been spent. So it seems to me that the real question is whether it indeed makes sense for the IMF to produce an annual estimate, for each country, of the total amount stolen or otherwise lost to grand corruption.
I’m skeptical, despite my general enthusiasm for evidence-based policymaking/advocacy generally, and for the need for more and better quantitative data on corruption. The reasons for my skepticism are as follows:
- First, measuring grand corruption (or, more narrowly, stolen assets) is extraordinarily difficult under the best of circumstances. The activities in question are criminal (even if in some countries the perpetrators have de facto impunity) and therefore generally secret. Even a well-resourced law enforcement agency like the US FBI would be hard-pressed to generate a reliable annual estimate of total stolen assets in its home country, let alone some other country. And the IMF is not a law enforcement agency. It cannot issue subpoenas or conduct forensic audits. Without a radical revamp of how the IMF operates—one that the member countries are unlikely to approve—I can’t see how the IMF could engage in the kind of data collection effort that Cardamone and Heydt seem to envision.
- Second, if the idea is not that the IMF will conduct its own investigations into financial crime, but instead use existing public data to estimate grand corruption levels, then I think it’s worth emphasizing that we’ve been down this road before, and the results are not encouraging. The existing data is so poor, and the extrapolation methods so dubious, that these sorts of attempted “quantification” are not much better than wild guesses. (See, for example, here and here.)
- Third, I’m not convinced that numerical quantification (especially if it takes the form of a wild guess) is really essential, or even all that helpful, for either advocacy or policymaking in this area. This is an issue I’ve previously discussed on this blog, and I don’t want to repeat myself at length here, but the basic idea is that there are ways both to demonstrate the magnitude of the problem, and possible causes and consequences, without a precise (perhaps misleadingly precise) dollar value attached to the harm. We already know that corruption is a big problem, and we have ways of plausibly identifying the countries where the problem is likely to be especially bad. While Cardamone and Heydt cite the “business adage” that “if you can’t measure it, you can’t manage it,” this adage—like many such adages—is more pithy than true. This is especially so in those contexts where the goal is to call attention to the urgency of the problem than to fine-tune the managerial response. And sometimes the attempt to produce precise quantitative estimates of what are really wild guesses can do more harm than good.
- Fourth, the attempt to gather this sort of data—for all countries every year—would be extraordinarily resource-intensive, and if attempted might well crowd out resources for other analytic and data-gathering work on corruption’s causes and consequences. It seems to me that the IMF in particular should dedicate its resources to analyzing how to address corruption in countries where misgovernance poses a threat to macroeconomic stability (which, after all, is the IMF’s main mission), rather than, say, trying to calculate the total amount lost to grand corruption in places like Sweden and Canada and Japan each and every year.
To be clear, I am in complete agreement with Cardamone and Heydt’s conclusion that, “Among the tasks at the top of her priority list, the new IMF Managing Director should commit herself to stemming the suffering of people resulting from the avarice of their leaders.” My respectful disagreement is only with the proposition that an annual country-by-country quantification of grand corruption is the best way for the IMF to use its limited resources to achieve this laudable goal.
A similar probem is the effort to quantify the proceeds of crime more generally. Better to focus on harms and outcomes rather than a monetary guestimate. But how else to measure the harm of grand corruption?
This is an interesting perspective. Are you arguing only against expending resources into an attempt to put a dollar value on corruption, or more broadly, against quantifying it in other ways? For example, do you think things like Transparency International’s country score are not useful? Another approach I have recently seen is the US State Department’s tier ranking system for human trafficking, with the 2019 report available at https://www.traffickingmatters.com/wp-content/uploads/2019/06/2019-Trafficking-in-Persons-Report.pdf. I wanted to clarify whether this is limited to only the type of monetary quantification that Cardamone and Heydt propose.
Good question. To clarify, the post is intended as a critique only of the proposal that the IMF try to gather this kind of data. I’m not opposed to investing resources in gathering and producing quantitative data more generally. Indeed, I think such efforts are often worthwhile, and on the TI index specifically, I think that this index can definitely be useful if used properly. I do have concerns, though, that the impulse to quantify can sometimes lead us astray, and I think that we need to be alert to the dangers of producing and repeating what are essentially made-up numbers.
It seems that even if full data was cheaply available and crime was fully observable, in some imaginary world, quantifying would still seem highly speculative. It seems that any such attempt would be strongly downward-biased, focusing mainly on direct monetary harm of corruption, missing things like reduced investment incentives and lower market entry, that by definition cannot be observed directly
True, but if taken as a reason not to quantify, I think that version of the objection proves too much. GDP figures don’t tell us the full social value of goods and services produced, and mortality/morbidity statistics don’t account for the full social costs of illness, but it’s still useful to gather quantitative data on those things.
Of course, if your comment is taken as a general reminder that the magnitude of a transaction in monetary terms does not by itself tell us the ultimate social value (or cost) of that transaction, then I wholeheartedly concur.
This post raises some interesting points that I hadn’t considered before. My initial reaction was that more and better data on these kinds of issues may be important for conducting research on other aspects of corruption. To do that well, presumably we would need data on the costs of corruption in Canada and Switzerland as much as in Nigeria and Romania. However, I do agree that the risks of getting bad data are real. If we don’t think that it is possible to get accurate measurements, perhaps it is better to refrain from investing in such projects (given the risk of garbage in, garbage out). However, in terms of the resource question, maybe there are other alternatives. If we think it is possible to get good data, given sufficient resources, perhaps this is worth investing in. Could the IMF, perhaps in partnership with other organizations, delegate such data collection to others, i.e. universities? The result might be more and better data that is useful for a wide array of corruption-related work.
The challenge of quantifying corruption damages is that the phenomenon presents hardly quantifiable aspects. For example, how could the lack of confidence (among people and in relation to the government) caused by corruption be quantified? In Brazil, the Federal Prosecution Service has been charging punitive damages in the most complex corruption cases, apart from disgorgement, restitution, and fines. However, the criteria to establish the amount of anticorruption punitive damages are not settled yet. Possible ideas from the World Bank and IMF would be helpful.
Thank you for this post Prof. and I’d like to second your fourth point, that “the IMF in particular should dedicate its resources to analyzing how to address corruption in countries where misgovernance poses a threat to macroeconomic stability (which, after all, is the IMF’s main mission)”. While the merits and demerits of quantitative data collection and inter country comparative assessments have been discussed, I see this as a broader question of holding IMF accountable to its own mission, i.e. is it choosing to do something that is (relatively) easy for it to do, and not necessary towards its mission.
Direct support to strengthen national statistical organizations within countries and interventions to strengthen their autonomy from the government (so as to protect them from pressure to massage or down play figures) would seem like an obvious priority. Once nations achieve a standardization in data collection and quality, such kinds of assessments can be done much quicker and at less cost. They will also have more credibility as the parent source of data would be disclosures from national governments. This would reduce the likelihood of “guesswork” as as happened in the past.