Analysis casts doubt on Bank scorecard: CPIA numbers made public for first time

11 September 2006

versión en español

Updated 7 November, 2006

In June, the World Bank disclosed for the first time details of its Country Policy and Institutional Assessments (CPIA), but analysis conducted for the Bretton Woods Project suggests that this costly and influential exercise is just another way to force borrowers into adopting the model of economic development supported by the Bank.

suspicions will remain that the CPIA score is based more on the implementation of preferred policies

The CPIA is made up of 16 indicators, covering four clusters: economic management, structural policies, policies for social inclusion and public sector management and institutions (see Update 43). Each criterion is given a score on a scale from one to six. The ratings, undertaken since 1997, are prepared annually in all countries by Bank country teams and then subjected to a process of internal review. In 2000, the Bank began disclosing the ratings, but only in an aggregated format – countries were ranked and placed into one of five groups from best to worst, referred to as ‘quintiles’. In 2004, after calls from both the board and an external review panel, Bank management agreed to make the detailed scores for the 2005 ranking available for low-income countries. The ranking is particularly important for low-income countries as it plays a central role in the Bank’s allocation of grants and low-interest loans.

The secrecy surrounding the CPIA exercise explains why there is so little literature on the topic despite the crucial role it plays. Barry Herman, former senior official of the UN financing for development office, has highlighted the inability of the CPIA to discriminate between countries or over time. He concludes that the CPIA, together with other similar indicators, should claim to be “no more than windows into a partial and clouded picture of development”. A report on the “uses and abuses of governance indicators” from the OECD concurs, concluding that problems associated with the CPIA’s construction or use include: “i) the likelihood of correlation of errors among the 37 sources from which the composite World Bank Institute indicators are constructed, which significantly limits the statistical legitimacy of using them to compare countries’ scores; ii) their lack of comparability over time; iii) sample bias; and iv) insufficient transparency.”

Other researchers have questioned the underlying objective of the assessments. Researcher at the School of Oriental and African Studies, Elisa Van Waeyenberge, believes Bank assistance remains conditional on a “core set of neo-liberal policies” . She points out how, as the need for greater selectivity in the allocation of aid became stronger within the Bank, loans became conditional on what had been achieved beforehand rather than success in the use of funds. Citizen’s Network on Essential Services has highlighted how the CPIA ratings violate the PRSP’s declared aim of strengthening country ownership because they, rather than stakeholders, set the framework for the writing of the PRSP.

Analysis for the Bretton Woods Project compared the detailed 2005 CPIA scores and trends in previous aggregate scores with relevant indicators produced by other agencies. This comparison revealed a number of discrepancies. A note of caution is warranted in interpreting these discrepancies – efforts to match time periods were restricted by data availability.

Comparison between the overall CPIA scores and growth in Gross Domestic Product (GDP) in the same year showed that many CPIA low performers are growing faster than countries that score well on the CPIA. Together with the world’s fastest growing economy oil-producing Angola, other economies like the Democratic Republic of Congo, Ethiopia, Sudan, Mozambique, Sierra Leone, Niger, Cambodia, Uzbekistan and Laos, all have GDP growth rates higher than Tanzania, Bhutan, Cape Verde, Senegal, Sri Lanka, Ghana, Uganda, Indonesia and other countries which score better in the CPIA.

The tenuous relationship between the ‘quality’ of policies and institutions as rated by the CPIA and growth is further highlighted by looking at the longer-term trend using previous years’ aggregate CPIA data. Samoa and Honduras – which have been ranked in the overall CPIA first quintile since 2002 – showed an average growth in the period 2001-2005 that is much lower than the average GDP growth rates among other countries belonging to the first quintile. The same discrepancies between quintiles’ average growth rates and growth rates of specific countries within the quintile (for the 2001-2005 period) are shown by Bolivia, Grenada, Benin, and Kenya in the second quintile; Madagascar, Moldova, Mozambique, Ethiopia, Tajikistan, Niger in the third quintile; Mauritania, Nigeria, Sierra Leone, Cambodia and Laos in the fourth quintile; and by Chad, Democratic Republic of Congo, Sudan and Angola in the fifth quintile.

Comparison of the overall 2005 CPIA scores with the UNDP’s Human Poverty Index (HPI), reveals several striking discrepancies. CPIA low-performer Sudan is ranked in a much higher position on the HPI than Senegal, Burkina Faso and Tanzania. In contrast, Burkina Faso, Mali, Lesotho and Mozambique – all of which scored high on the overall CPIA index – scored poorly on the HPI. Benin, Haiti, Eritrea, Nigeria and Togo also showed big discrepancies in the scores obtained in the two indicators.

Discrepancies were also revealed between the CPIA’s ‘transparency, accountability and corruption’ index and the Corruption Perception Index 2005 (CPI) produced by NGO Transparency International (TI). Georgia, for example, scored very well in both the overall CPIA score and the specific transparency and accountability index. However, in the TI-produced CPI, Georgia was ranked amongst the bottom countries, after Mongolia, Nepal, Sierra Leone, Eritrea and Gambia (all of which scored much worse than Georgia on the CPIA corruption index). Georgia’s notable improvement seems more related to what was reported in the World Bank’s Doing Business 2006 report, which trumpeted Georgia’s recent introduction of major reforms to its labour laws, removing restrictions on working hours and dismissal procedures, and thereby lowering firing costs to some of the lowest levels in the world. Manana Kochladze, regional coordinator for the Caucasus for the Central and Eastern European Bankwatch network, reports that “due to its absurdity, the introduction of the new labour code in Georgia caused protests by the entire population including businessmen. Even from a neo-liberal point of view, the new code does not bring a lot of gains to employers and it creates instability for employees.”

While these preliminary comparisons are certainly not enough to dismiss CPIA scores, they cast doubt on both the methodology of the assessments and the confidence with which they should be used as a basis for vital aid allocation decisions. Publishing the full ranking was a welcome step towards more transparency but it is still not sufficient. Until more detailed information on how each score is obtained are released, suspicions will remain that the CPIA score is based more on the implementation of preferred policies than on an impartial assessment of countries’ economic and political policies and institutions.