A World Bank-initiated effort to reduce corruption in school funding in Uganda is widely, and rightly, celebrated for its results (click here and here for background). In the early nineteen nineties on average 87 percent of the monies the Ugandan central government budgeted for textbooks and other school supplies “leaked out” somewhere between departing the Finance Ministry and arriving at the school house front door. Yearly data revealed that 73 percent of the schools received less than five percent of the monies to which they were entitled, and only ten percent received more than half. The 1996 Bank project had an immediate effect on the rate of losses. By 1999 the government found schools were receiving on average 95 percent of what they were supposed to receive, and a 2002 World Bank study likewise showed a sharp drop in fund leakage.
The dramatic improvement is attributed to the enormous publicity the data on losses garnered. Parents were outraged and the government and donor agencies embarrassed. Within the development community, the Uganda Public Expenditure Tracking Survey, as the work to dig out and publicize the loss data became known, has been enormously influential, the story becoming a parable for how to fight corruption. A Uganda-like PETS project is now routinely prescribed for attacking corruption in public expenditures, and a Google search on “Uganda PETS” yields over 100,000 hits and returns some 20,000 citations on Google scholar.
But for all the attention the effort has generated, there is evidence that it may not have had any impact on the level of corruption in Uganda. It is possible that all it did was force those raking funds off the school fund program to turn elsewhere. The Uganda PETS thus may simply have displaced the corruption in the school funding program rather than ending it.
Evidence from the Uganda PETS pointed to those working in the then 33 Ugandan district governments, the entities that stand between the central government and the individual schools, as primarily responsible for the leaks. The central government sent the money for schools and other social welfare programs to these governments, and officials there divided the money up and passed it along to schools, health care clinics, and other lower level units of government. Although how the stolen funds were used remains a mystery, the best guess is that they supported local patronage networks. As Melissa Thomas, a long-time student of Uganda politics has observed, political power in the districts is wielded by a small circle of individuals connected through common schooling, marriage, friendships, shared ethnicity, and membership in the ruling party. Maintaining the power balance among these local elites is essential for keeping the peace and ensuring their continued loyalty to the party. It is also expensive, and thus it likely that some or all of the embezzled funds helped finance patronage networks.
One test of this conjecture would be what happened when the leak of school funds was plugged. If the stolen money was being spent on vice or luxury goods, one might have observed a sudden recession in the gambling, prostitution, and other industries dependent on discretionary spending as in China when it began to crackdown on corruption. On the other hand, if the money was oiling an essential part of the machinery of government, one would expect those responsible for keeping the machine running would find another funding source.
That appears to be what happened in Uganda. Coincidence with plugging the leaks in the school fund program was a sharp increase in the creation of new district governments, from 33 in 1990 to 80 in 2006. Each new district government provided a wealth of government jobs, of contracts to build and maintain buildings to house the new governments’ offices, and other opportunities for patronage.
So the Uganda PETS story may well hold a second lesson in addition to that of the power of information to curb corruption. It may also teach that it is not enough to identify a corrupt practice and then shut it down. One must also ask: why corruption here? What purpose does it serve? Otherwise, plugging a leak in one place may simply result in a leak springing elsewhere.
To be sure, if that were indeed what happened in Uganda, displacement may still have enhanced national well-being. An expansion in government employment surely is less harmful than denying primary students text books. But in any event the Uganda case still provides a bit of caution to corruption fighters: the importance of a thorough appreciation of what motivates corruption when devising a strategy to combat it.
GAB | The Global Anticorruption Blog 
There is a very nice article that provides greater context to the Uganda PETS narrative by Paul Hubbard, here: http://www.cgdev.org/files/15050_file_Uganda.pdf. Hubbard points out that the level of corruption actually didn’t change; it was smaller as a percentage because donors poured more money into the sector. In addition, donors responded to the results of the PETS by imposing stricter monitoring and conditions on their aid. As far as I know, there has been no evidence advanced that the publication of school budgets generated any action by parents, much less that this action reduced corruption. This was never more than speculation, which got unusual traction because it made such a good story.
But the question Rick raises is a very important one. I argue that many governments do not govern by relying on the legitimacy that comes from the provision of public goods and services and the rule of law, but through older, cheaper strategies, including patronage and repression. The poorest governments can’t afford to provide public goods and services and the rule of law, which means that they must rely on those older, cheaper strategies if they are to govern at all. What President Wolfensohn of the World Bank called “the cancer of corruption” may actually be the patient. In those cases, we should be asking how to make patronage systems more inclusive, more stable, more fair, more predictable, or more productive instead of asking how to destroy them. See my book, “Govern Like Us,” http://www.amazon.com/Govern-Like-Us-Expectations-Countries/dp/023117120X.
Rick,
This is a very useful, if somewhat depressing, post. I’ll own up to being one of the people who frequently cites to the Uganda PETS experience — in both casual and formal talks — as an example of a successful intervention to fight corruption. I was not aware of the displacement you describe. And your larger points, both about the risks of displacement and about the need to understand the functions that corruption may be serving in different contexts, are important ones which I fully embrace.
All that said, I do want to emphasize the point you make in your final paragraph, which to some extent undercuts your earlier claim that the PETS intervention “may not have had any impact on the level of corruption in Uganda.” Two points here:
First, even if there was some displacement, it’s still quite possible that the total level of corruption (broadly defined) went down. People regularly, in my view, exaggerate displacement arguments of the sort you’re making, implying that the substitution is complete. Whenever you prohibit, or raise the cost, of X, there is likely to be some displacement to substitute Y–but there’s usually an overall decrease in consumption as well. If I raise the cost of hamburgers, people might eat more hot dogs–but total meat consumption may well still decrease, and indeed in most cases (though not all) it will. Likewise, if I raise the cost of securing patronage funds by stealing them from school budgets, local elites might raise patronage funds from other sources–but the total amount of money going to patronage activities might still decrease.
Second, as you rightly note in your last paragraph, the substitute activities might not be nearly as destructive, and indeed some of them might not be “corrupt” at all (though this gets us into the tedious business of debating the definition of corruption, something I’d rather avoid). Over-employment in the public sector might not be ideal, but it may not be nearly as destructive as, say, looting all the money intended for elementary education, public health, infrastructure, etc. One might generalize the point by saying that in pretty much every political system, politicians try to find ways to do favors for their constituents, but in some systems this is done through means we’d view as corruption (as usually understood), while in others it’s done through means that, while perhaps tawdry or inefficient, are not corrupt (in the narrow sense), and are not nearly as damaging.
Maybe I’m just straining to find a justification for continuing to cite the Uganda intervention as a (partial) success, but I think it’s worth bringing out these points lest we become too pessimistic about the ability of any realistic intervention to do anything.
Your comment reflects an argument SOAS Professor Mustaq Khan has made. In some countries, he asserts, corruption is the glue that holds the country together and is thus a necessary evil.
In such cases, what anticorruption advocates should be doing is finding ways governments can prevent civil war or separatist movements legally. That is, turn “illegal corruption” into some form of pork barrel spending or patronage. The Uganda example, assuming the creation of additional district governments was legal, would be an illustration of how to accomplish this.
An alternative approach to testing the effectiveness of PETS, I suppose, is to have a cost benefit analysis of PETS itself. What is the cost of undertaking PETS compared to the amount of leakages it being able to identify or prevent?
Indeed. There may be cases where the cost exceeds the benefits. Although as I suggest in my reply to Matthew, “benefits” may need to be widely interpreted.