What The Cable Gets Wrong About MCC’s Upcoming Selection Process

Yesterday, Foreign Policy’s The Cable came out with an article warning that—with the upcoming board meeting to determine which countries will be eligible for MCC funding for FY14—MCC is essentially on the verge of pouring money into the hands of corrupt regimes.  

A U.S. government-funded foreign aid organization is considering sending hundreds of millions of dollars in grants to countries whose flawed, corrupt or undemocratic governments should almost certainly be ineligible for the money according to the agency's own internal guidelines.

If the MCC bets on the wrong countries, by contrast, huge amounts of American money could disappear into the pockets of corrupt government officials or business leaders.

That sounds pretty bad.  The reality, however, is far more complex.

As a bit of background, MCC publishes scorecards showing developing countries’ performance on 20 policy indicators.  Indicator performance is one of the MCC board’s primary considerations for deciding compact or threshold program eligibility.  To pass MCC’s indicators, a country must score better than at least half of its income-level peers (for most indicators) or above a fixed threshold (for a couple of indicators) on at least ten of the 20 indicators.  There are two “hard hurdles”: a country must pass the control of corruption indicator and at least one democracy indicator (either political rights and/or civil liberties).

This year MCC’s board faces the difficult task of deciding whether to re-select four countries—Benin, Liberia, Morocco, and Sierra Leone—that don’t meet MCC’s indicator criteria.  All four countries were initially selected as compact eligible in previous years and are now working with MCC to develop a compact.  They’ll need to be re-selected this year, and each year until a compact is signed, in order to continue the process.  As the article points out, this is the first time that so many countries that are up for re-selection don’t meet the indicator criteria.  Benin and Sierra Leone miss the mark on corruption, falling just on or just below the median on the control of corruption indicator.  For Liberia and Morocco corruption is not the issue; both still pass the hurdle but got tripped up in other areas--eco-region protection for Liberia and gender equality and rural land rights for Morocco.

There is no question that the possibility of re-selecting four countries that don’t pass the indicators is, at a minimum, an optics problem for MCC.  But determining what to do in this situation is not as simple as black and white (or green and red, in the colors of the scorecard).  MCC is right to base its selection of countries on a transparent, data-based tool.  But it is just one tool (MCC also relies on supplemental information), and it would be wrong to use that tool exclusively and rigidly.  On balance, the indicators are pretty good at distinguishing high performers from low performers, but they’re imperfect proxies for policy performance, not the “truth”.  And they’re not particularly sensitive to small differences.  Small changes in score from year to year, even changes in pass/fail status do not necessarily mean there have been actual, meaningful changes in a country’s policy environment.  The Board deemed Benin, Liberia, Morocco, and Sierra Leone to be “good enough” to partner with MCC when they were initially selected.  The data now raise some questions.  It is MCC’s job to figure out if the indicator changes are reflecting a real deterioration in policy or if something like the following has occurred:

• The indicator’s methodology changed: Sometimes the data sources change how they measure a particular policy issue.  This can result in a change in score, even if there is no change on the ground.  You can see this affecting Morocco's indicators this year.
• The peer group median increased: MCC’s scorecards assess countries relative to one another.  Therefore, if the average performance of the group improves, a country may bump from pass to fail, even if its own score is largely unchanged.  This affects Morocco this year.
• Better data became available: Data quality can be weaker in poor countries, so sometimes, when better data become available, a country’s score can change, even without a change in actual policy.  This affects Liberia this year.
• A small decline occurred but was insignificant:  Small changes in score are often just noise.  For instance, the Control of Corruption indicator is an index of nearly two dozen possible sub-sources.  Sometimes declines in a few of these sub-sources (which may actually be accompanied by increases in other sub-sources) can trigger a small decline in overall score, but the change is not significant, neither statistically nor on substance (statistically significant changes from one year to the next basically do not occur).  This affects Sierra Leone and has an impact on Benin this year.  Let’s think about it—Benin just fails the control of corruption indicator because it is the median score.  Would there be such concern about Benin if it scored in the 51st percentile?  I suspect not, even though there would be pretty much exactly zero difference in the quality of the anti-corruption policy environment.

If Benin, Liberia, Morocco, and Sierra Leone were considered “good enough” when the board first selected them (and as far as I know, there was little protest about any of them the year they were first picked), if they have had no material change in policy environment since then (which it falls to MCC to determine), they should be re-selected.  This is especially true since these countries have spent the last year developing a partnership with MCC.  The US Government has the responsibility to be a reasonable and rational partner; if MCC is going to cut off an ongoing relationship with a country it must be able to point to a concrete policy deterioration, not just a color change on a scorecard.

It’s also worth pointing out that performance on the control of corruption indicator tells us very little about whether an MCC project will be subject to corruption.  Quite frankly, there’s a risk of corruption in pretty much all countries with which MCC might work, including those that pass the control of corruption indicator.  Whether or not American money is siphoned off depends much more on the specific projects chosen, individual actors involved in the project, and how well MCC’s (quite rigorous) oversight mechanisms are employed.  If one is really worried about misuse of MCC resources (vs. just concerned that MCC will reward the “wrong” countries), it would be better to assess MCC’s operational policies and practices rather than its selection decisions.

All in all, the Cable’s “Exclusive” on what’s brewing for MCC’s selection decisions this year doesn’t really uncover anything particularly scandalous; the less exciting reality is more about the challenges of noisy data and imperfect measurement tools.

As a post script, there are several factual inaccuracies in the article:

• “Countries that want a piece of the MCC's roughly $2 billion in annual funding....”  MCC would probably like to have $2 billion in annual funding, but its budget has not been above $1 billion for several years.
• “The second requirement is that applicants notch scores of at least 51 percent on a subset of specific measurements of their levels of corruption, rule of law, civil liberties and other democratic rights. Benin and Sierra Leone failed to hit those targets.”  True, Benin and Sierra Leone do not pass the control of corruption indicator, but they do meet the criteria on rule of law, civil liberties, political rights, and freedom of information (in fact, they’re among the top performers on the democracy indicators)
• “The question the MCC's leadership will face next month is whether to continue their work with those governments despite their failures to meet the MCC's internal guidelines.”  Actually, MCC’s guidelines on selection—which they make public—clearly state that MCC takes into account supplemental information in addition to the scorecards, and that partner countries may not meet the indicator criteria from time to time due to data-related factors unrelated to an actual policy reversal.
• “That means that [MCC] does not provide general aid to countries like Israel or fund large-scale infrastructure projects in places like Afghanistan or Mali.”  MCC did, in fact, have a compact with Mali, though the compact was terminated early as a result of the coup.