The Millennium Challenge Corporation’s New Country Scorecards Discriminate Against Development Impact

One of the unique features of the US Millennium Challenge Corporation is its scorecard selection model, which uses a range of development indicators as a first step to decide if a country should be considered for MCC financing. The assumption is the finance will have a bigger impact in countries that pass the indicator “scorecard.” The system has long included flawed measures around governance and corruption, but reforms this year have made the impact of these indicators even worse, ensuring scorecards systematically discriminate against the poorest countries where MCC’s support can do the most good. That’s the antithesis of the idea behind the scorecard system. MCC’s supporters in Congress and beyond should cry foul.

For most of the 22 indicators in this year’s scorecard, a country is considered to pass that indicator if it performs better than the median score in its income group. But the scorecard has three “threshold” indicators, where a country must pass a certain set level, and which is the same regardless of income group. These are (i) an inflation rate below 15 percent, (ii) government accountability, which is a combination of the Political Transformation index of the Bertelsmann Stiftung’s Transformation Index (BTI) and the TRACE risk score and; (iii) personal freedom, which is the Voice and Accountability indicator from the World Bank’s Worldwide Governance Indicators (WGI).

This year, a country is considered to pass the full scorecard screening process if it: (i) "passes" at least half of the indicators (as previously); (ii) "passes" the (threshold) Personal Freedom indicator; and (iii) "passes" either the (relative to income group) Control of Corruption indicator or the (threshold) Government Accountability indicator. Stages (ii) and (iii) replace the previous system of passing one of two Freedom House indicators of civil and political rights (both threshold) and passing the control of corruption indicator (relative to income group).

Back in 2014, I wrote a paper with Casey Dunning and Jonathan Karver, “Hating on the Hurdle: Reforming the Millennium Challenge Corporation’s Approach to Corruption.” The reasons we didn’t like the control of corruption hard hurdle then in place was “(1) the control of corruption indicator reflects broad perceptions of governance with some noise, risking considerable errors of inclusion and exclusion; (2) the control of corruption indicator is not strongly related to progress in development outcomes, nor are country-level governance indicators strong determinants of aid project performance; and (3) the control of corruption indicator changes slowly over time, with an opaque relationship to reform efforts.”

The new threshold hurdle indicators share all of those bad features. The Political Transformation index of the BTI is a mashup index purely made up of perceptions, the TRACE risk score is a mashup index largely made up of perceptions (which partially draws on the BTI), and the Voice and Accountability indicator from the WGI is a mashup index largely made up of perceptions (which also draws from the BTI).

To be fair, the two Freedom House indicators dropped from the hurdle requirement were also threshold indicators and perceptions-based. But using the three replacement indicators to set thresholds regardless of income group makes their impact even worse. These indices, particularly the Voice and Accountability indicator, are strongly correlated with GNI per capita. Set an absolute threshold value on them and you are sure to fail more poor countries. (Average inflation, the third threshold measure, is also historically higher in poorer countries.)

Figures. The relationship between MCC threshold hurdle indicator components and income

MCC has three income brackets of countries in which it works: $2,155 and below (we’ll call this low income), $2,165–$4,495 (mid), and $4,496–$7,855 (high). In part because of the threshold hurdle indicators and their association with income, pass rates both for the hurdles and for the measure of outperforming the median on half of all scorecard indicators are significantly higher in richer candidate country pools.

As it turns out, removing the (relative to income) hard hurdle based on corruption also favors richer countries, The countries that pass the half scorecard test that don’t get rejected thanks to the old hurdle are: Bolivia (mid income on MCC measure), Ecuador (high), El Salvador (high), Honduras (mid), Lebanon (mid), Mongolia (high), Papua New Guinea (mid), Paraguay (high), Peru (high), Thailand (high), and Ukraine (high). Of these countries, Bolivia, Ecuador, El Salvador, Honduras, Mongolia, Paraguay, Peru, Thailand, and Ukraine fully pass the scorecard (i.e. they also clear the accountability and freedom measures).

The “hard hurdles” have a far greater impact on removing potential candidates in the low-income countries (a 17 percentage point drop) than high-income countries (4 percentage points). Add in the fact that the threshold indicators all count towards passing half the scorecard and MCC pass rates by income bracket are almost twice as high in MCC’s high income category than in its low-income category: 62 percent of countries as compared to 33 percent.

Table. MCC scorecard pass rates by measure and MCC income bracket

MCC was originally designed to work only in the poorest countries (the current low-income category), where it would have the greatest impact. Over time, it has expanded eligibility towards ever-wealthier countries, recently expanding to countries that are as rich as Peru and Thailand. There are very good reasons to worry about this shift: a dollar goes far further in improving outcomes in the poorest countries home to far larger shares of people living in extreme poverty and dying prematurely, and the impact of a similarly sized MCC program that might be transformative in a poor country will be hardly noticeable in one with ten times the income per capita. The evidence suggests each dollar of MCC finance would have a far greater impact if focused on the countries it was originally designed to support.

But the new scorecards push MCC even further in the direction of abandoning those very economies where it can make a difference. And policy decisions after passing the scorecard are making this problem worse. MCC recently abandoned a number of compacts and threshold agreements covering Cabo Verde, Senegal, Lesotho, Malawi, Timor-Leste, Gambia, Togo, Kenya, Mauritania, and Tanzania. That’s one MCC high-income country and nine low-income countries.

The one silver lining: while all of this is a shift in the wrong direction, it suggests (contrary to previous belief in MCC and beyond) the corruption hard hurdle is not sacrosanct. it points to the possibility that in future years MCC could make reforms in the right direction, ones that would allow for more compacts in the places that would benefit the most from MCC support.