The World Bank’s Integrity Vice President (“INT”), responsible for investigating corruption and fraud in World Bank projects, recently released its Fiscal Year 2015 Annual Update. INT had a busy year, opening 323 preliminary investigations, of which 99 were selected for full investigation, and closing 81 investigations, with three-quarters finding evidence of sanctionable conduct. (A primer on how INT conducts external investigations is here.) Some of INT’s recent cases, such as those brought against Alstom SA and SNC-Lavalin, involve large companies. Yet despite these examples, the data in the Annual Report raises questions about whether INT is sufficiently effective in uncovering corruption and fraud by large companies. The evidence suggests not: The firms debarred in FY 2015 are mostly small- and medium-sized enterprises—minnows, not sharks. The longest debarment leveled was for thirteen years on N.C. Sanitors and Service Corporation, essentially for paying public officials in Liberia and falsely claiming it collected trash that it never picked up. The challenged contract was worth about $350,000—not exactly a break-the-bank amount, especially considering the largest contracts the World Bank awarded last year were worth $438 million, $98 million, and $53 million (excluding government-awarded contracts funded by World Bank loans).

Perhaps large corporations with World Bank contracts and governments officials administering large World Bank loans are not engaging in corruption—but I doubt it. It’s much more likely that INT does not have the information that it would need to investigate and seek to sanction large companies. According to people familiar with INT’s intake system, while INT gets thousands of tips a year through its phone and online tip lines, many of which prove valuable (either individually or when aggregated), relatively few tips relate to large contracts where the amount of money at stake enhances the harm from corruption and bribery. INT should therefore develop methods to get actionable information on fraud and corruption related to large projects. My suggestion: pay for information.

One reason why INT may receive few tips about large contracts is that INT currently only offers confidentiality to protect whistleblowers. When it comes to large contracts, the likelihood that a whistleblower will face repercussions if her tip is revealed increases, changing the cost-benefit analysis of reporting. Some potential whistleblowers with actionable information might need some sort of additional material incentive to offset the potential risks. A well-structured system using payments to induce reporting might therefore increase the amount of actionable information INT receives about large-contract corruption.

What would such a system look like? How should it be designed? While this is not the place to lay out the proposal in all its details, the essential elements might work as follows:

• Qualifying Contracts. First, only information about fraud and corruption related to large loans and contracts should qualify for the payment program. To keep the program relatively small at the outset, the threshold should be set fairly high: For World Bank loans to governments, a starting benchmark around $500 million or more seems reasonable (according to IBRD and IDA data, there appear to be fewer than 150 such loans currently outstanding); for World Bank contracts, $10 million seems reasonable (the Bank awarded 55 such contracts over the past three fiscal years). Although arbitrary, both cut-off points limit the universe of eligible loans and contracts so that INT can determine the usefulness of whistleblower rewards while decreasing the risk that INT staff will be flooded with useless tips that “crowd out” useful information.

• Qualifying Information. INT should further limit the scope of the rewards program by paying only those whistleblowers whose tips lead to cases that proceed to the full investigation stage. According to the Annual Update, INT closed 81 full investigations last year, and in 74% of those, the claims were “substantiated.” Using full investigations as the benchmark for rewards should still allow whistleblowers to receive compensation even if the investigation results in a Negotiated Resolution Agreement. It is worth mentioning two potential concerns with the “full investigation” benchmark. First, as I mentioned in a previous post, there’s a risk of creating perverse incentives for investigators to decline to proceed to full investigation in marginal cases, so as to avoid the need to pay the whistleblowers. That risk is mitigated in a program like this one, given the relatively small universe of eligible contracts, and the relative success rate in past full investigations. Second, setting the reward benchmark relatively late in the investigation process might reduce the number of whistleblowers, because the odds of a given tip reaching that stage is lower (and the whistleblower may not have a good sense, at the moment she must decide whether to contact INT, how likely her tip is to prompt a full investigation). Although this may marginally decrease the amount of corruption being reported, funding constraints (discussed next) necessitate limiting the number of recipients.

• Reward Size. The appropriate size of the reward may be the toughest question to answer. I think $10,000 (approximately the median household income globally in 2013) is a reasonable place to start (this figure could be scaled down for information on smaller contracts if INT expanded the program to cover smaller contracts). Unfortunately, INT probably cannot adopt a system like the False Claims Act or the Dodd-Frank Section 922 Whistleblower Program, under which whistleblowers receive a percentage of the money recovered, primarily because the World Bank sanctions and debarment regime rarely results in restitution payments (as discussed in this World Bank overview of its Sanctions Regime). (It is an interesting question whether the World Bank could or should redirect some restitution from the government borrower to a qualifying whistleblower, but that is beyond the scope of this post.) Since money recovered from debarred entities probably cannot fund the program, rewards for whistleblowers on large contracts that reach the full investigation stage might have to come from INT’s budget. Unfortunately, that is not a deep well. INT’s funding has fallen each of the last four years, to $18.6 million last year. Any dollar put toward rewards would be a dollar taken from investigations, perhaps making this proposal a non-starter for INT officials. That said, $10,000 is just 0.05% of INT’s FY 2015 budget, so even if INT got 10 tips per year that triggered a full investigation of corruption at a large company, this would only be 0.5% of INT’s budget, which actually seems like a pretty good return on investment.

There are, to be sure, numerous other questions that would need to be addressed if INT were to adopt a system like this (for example, what happens if multiple people report the same information). And there are reasonable concerns that paying for information might lead to a flood of worthless information, that paying whistleblowers could “crowd out” the intrinsic moral motivation to report bad acts, and that a new program will just add more red tape to an already slow-moving organization and lengthy investigative process (which averaged about two years according to this Office of Suspension and Debarment report on 2007-2013). Some of these concerns were discussed in the comments to my earlier post.

More generally, while these and other concerns are serious and must be addressed, INT should nevertheless explore a program like the one sketched above, and could experiment with the program design, making adjustments and improvements based on its knowledge and experience. INT needs more and better information about how World Bank money is being used, especially in large contracts involving large companies, and paying for information may be one of the most effective ways to achieve this goal.