Governance Indicators, Aid Allocation and the Millennium Challenge Account by D. Kaufmann and A. Kraay (2002)—DRAFT FOR COMMENTS		Governance and the MCA (71 kb) Abstract (11 kb) MCA Fact Sheet (23 kb) CGD Reactions to the MCA (13 kb) Governance Indicators Dataset Research: Growth Without Governance Rethinking Foreign Aid, Washington Post
Aid works best when it is directed to countries with relatively good institutions and policies. But how should good governance be measured, and how can aid allocation rules be designed in light of the strengths and weaknesses of existing measures? In this brief draft paper (71 kb PDF) Kaufmann and Kraay discuss a number of challenges, motivated by the U.S. government's recent proposal to allocate resources from the new Millennium Challenge Account (MCA)—yet the issues and recommendations apply more broadly. Among others, the authors discuss the implications of margins of error in governance data, the difficulties in measuring trends, and the need to complement existing cross-country indicators with in-depth country diagnostics.   This draft is intended to elicit feedback and reactions on these issues of governance measurement and aid allocation. For that purpose, the MCA fact sheet (23 kb PDF) recently released by the U.S., as well as a note (13 kb PDF) reacting to such fact sheet by Steve Radelet of the Center for Global Development (CGD) are also included here. Furthermore, links to the CGD website as well as to the detailed governance databank and background governance research papers are provided.       Download "Governance Indicators, Aid Allocation, and the Millennium Challenge Account" (71 kb PDF) by D. Kaufmann and A. Kraay, December 2002, draft or one-page abstract (11 kb PDF). Please scroll down to provide feedback and read the feedback received.

Your feedback on these issues would be appreciated, and posted, as part of this discussion. To read the feedback received, scoll down. (view posting guidelines)

From: virachai@ksc15.th.com

I find your document is most informative, and useful for further researching works.

From: c-santiso@dfid.gov.uk

Once again, the WBI's governance team has produced a timely and useful paper on governance quality and aid selectivity. I wanted to commend your work, as it is very useful in performing mine.

One issue that goes beyond, I think, the issue of selectivity relates to the new fad in terms of aid modalities and the increasing use of budget support by bilateral aid agencies. In that context, we need reliable time-series of governance indicators that would allow us to assess trends (rather than arbitrarily measure quality).

I understand that you plan to expand the existing set back to 1995 and produce regular updates of the indicators. How institutionalized is this process?

Once again, thank you and congratulations for the excellent piece of work!

Carlos Santiso
DFID Governanmce Adviser

From: alicht@idc.ac.il

The work of Kaufmann, Kraay, and Zoido-Lobaton has taken the operationlization of social institutions to a new level in terms of both rigor and coverage. Here, Kaufmann and Kraay point out several important issues that should be bourne in mind when considering action based on their governance dataset. While I fully agree with their remarks, I would like to suggest a complementary proposition, namely, that most, if not all of the governance indicators in fact reflect a common underlying social factor. Statistically, these indicators are highly correlated – specifically, the rule of law and non-corruption measures correlate nearly fully – and they exhibit similar correlations with numerous factors. Researchers have thus started to use principal component analysis to extract such an underlying factor statistically. Licht et al. relate three indicators to psychological profiles of nations. So, while there is room for further in-depth elaboration on these indicators (item 6) we should also seek better understanding of their general structure.

From: Theodore R. Breton, t.breton@verizon.net

It seems to me that research, such as LaPorta, et.al.(1999) and Treisman 2000), indicates that income levels are the primary determinant of government corruption. If so, then assisting the poor requires that aid be targeted on the most corrupt countries.

My empirical research indicates that the best way to raise national income (eventually) is to promote universal education. The most corrupt countries are where this is most needed. The World Bank should be figuring out how to raise educational levels in the most corrupt countries.

Theodore R. Breton
George Mason University

From: Clay Wescott, cwescott@adb.org

Dear Dani and Aart,

Thanks for sending your MCA paper and asking for feedback. I think that the MCA is an exciting policy experiment, and agree that all the cautionary issues you raise need to be dealt with. Here are two quick thoughts that you may want to consider.

First of all, there is one other example of possible margin of error in governance indicators that you may want to look at. Under the World Bank’s LICUS initiative, some of the countries identified as exceptionally weak performers (Cambodia and Lao People’s Democratic Republic) are indicated as “green light” countries under your rankings, with Cambodia fully in the top half of performers at the 90% confidence level.

Secondly, I agree fully with your issue 2, that margins of error can be reduced by looking at more sources of information. You may want to take this point on board in your discussion of issue 6: in-depth country diagnostics. As you know, the World Bank isn’t the only agency carrying out such work. ADB, for example, has completed country governance assessments (CGAs) in Bhutan, Cambodia, Lao People’s Democratic Republic, People’s Republic of China, Thailand and Vietnam. CGAs that have been conducted but not yet finalized include Bangladesh, Fiji, Indonesia, Kazakhstan, Kyrgyz Republic, Federated States of Micronesia, Mongolia, Nepal, Pakistan, Papua New Guinea, Samoa, Tonga and Vanuatu.  Recently initiated CGAs include India, Philippines, and Sri Lanka. As these are completed, they are being published, and made available on http://www.adb.org/Governance/govpub.asp. Other MDBs and bilateral agencies are also carrying out comparable diagnostic studies, which can usefully complement your own GAC surveys.

I look forward to reviewing the revised version of your paper.

Best regards,
Clay Wescott

From: jvandervossen@imf.org

I applaud the attempt to analyze the degree to which countries meet standards of good governance. This is a very difficult area, and the authors show courage in addressing these issues in a systematic way. They commendably try to reduce the subjectivity of most discussions of these issues into more concrete information - at least as much as possible.

However, while I appreciate that the study is geared towards eligibility for funding from the MCA, and therefore needs to take MCA criteria as a point of departure, the draft should, in my view, discuss at greater length and more objectively the good governance criteria chosen by the MCA. How relevant are they, how do they hang together, how does one assess observance? The study seems too willing to accept these criteria \"as is\". It would be interesting to hear to what extent the authors would agree with the criteria chosen by MCA. What other criteria could be developed and applied, for instance, and on what grounds?

In brief: even if the MCA criteria are accepted, more explanation and justification of their rationale and cohesion would be needed.

Has there been any form of country feedback on this study? Or have countries provided informational input?

Is this study to be published further? Especially if the countries are to be assessed based on the MCA criteria taken at face value, this issue should be reviewed very carefully. Once countries have been listed, say in the bottom 25 percent, and taking into account the non-trivial margins of error, there needs to be some possibility for countries to get \"de-listed\", or \"re-rated\".

The experience of countries with the Financial Stability Forum list of weak offshore financial centers comes to mind. This list had no pre-designed exit for countries which had been listed, but subsequently improved their act, or which felt they had been assessed unfairly (the FSF survey was also perception-based). This caused much frustration in the countries concerned, and reduced the authority of the list.

From: fwitthans@usaid.gov

The statement that margins of error can be reduced by relying on more sources of information may not be true if in including more sources of information less reliable data is included. Perhaps some sources of data should be excluded if an examination of the yearly data for an indicator by that source shows great variance from year to year. Does each source employ an acceptable methodology in collecting data?

From: Roy Smith, roy.smith@pwgsc.gc.ca

The papers written by Mr. Kaufmann and Mr. Kraay are interesting in that they look at six governance indicators in the selection, retention and ability to govern as well as construct policies in a given area, which in turn influences the general populace's ability to adopt these policies. This explains in some sense some of the underlying reasons why aid is successful in one area and not in another. There are other affiliate reasons such as extent of freedom of speech, and freedom of assembly as well as the journalistic freedom that affects the immediate impact of foreign aid to stimulate growth. Prosperity is a relative term that takes on many different signs depending on the society where it is being measured.  The yardstick used in Europe or North America is not the same that can be used  in other societies. It is like comparing these societies of the present day with that which existed here 100 years ago. The trends we see today with demonstrations of socialism and greater compassion for the weaker units in society in the North American context are there because we can \"afford\" to be generous. In order for the masses to benefit the elite must feel comfortable and unthreatened. Corruption and deceit flourish when shrouded with the veil of success, a high level of materialism and when there is no polarization of the higher values that one would look for in a society.

The discussions generated by these papers will provide new impetus into this conundrum. Thank you.

Roy Smith

From: fwitthans@usaid.gov

The measure \"Control of Corruption\" is confined to measuring the exercise of public power for private gain. Clearly, the more corruption the greater are transaction costs and uncertainty and, hence, the fewer the number of private projects that will be viewed as profitable. However, is it not true that private rent-seeking activities like theft, piracy, litigation, protection, and extortion also increase transaction costs and uncertainty, and also result in fewer private projects that will be viewed as profitable?  Should not an indicator for private rent seeking also be measured? The \"Rule of Law\" indicator includes perceptions of the incidence of both violent and non-violent crime. Measures of private rent seeking may be included in that measure but it will be watered down by all the other components comprising that index. For some countries, extortion and other Mafia type activities may be as harmful for productive activities as corruption.

Response: D. Kaufmann and A. Kraay

Thanks to those of you who have posted comments and reactions to our recent note “Governance Indicators, Aid Allocation, and the Millennium Challenge Account”. These comments raise a number of interesting issues, and we respond to a few of these here.

One comment noted the good performance of Cambodia and Laos in the control of corruption indicator. These two countries fall in the top quarter of potential MCA-eligible countries. This comment is interesting because it underscores a point we made in our note – the importance of relying on as many sources of information as possible. Both these countries are part of only a small set of countries for which we have data from only one of our underlying sources of corruption perceptions. In contrast, the typical MCA-eligible country appears in four sources, and for the world as a whole, the typical country appears in five sources. In the updated indicators for 2002 that we will be releasing shortly, we will have additional sources of information for these countries which may affect their relative positions.

Two comments noted the strong complementarities between control of corruption and rule of law. It is true that these two tend to be highly correlated in the data we have – countries in which corruption is widespread are also often countries where the state is unable and/or unwilling to protect private property rights. It is interesting to note however that not all dimensions of governance that we measure are so highly correlated. This is especially true among the set of potential MCA-eligible countries. For example, in this set of countries, the correlation between voice and accountability and control of corruption is nearly zero.

A further comment raised the view that income is a primary determinant of corruption, and measures that raise income are likely to be most effective in fighting corruption. We do not fully agree with this interpretation. It is true that corruption and income levels are strongly correlated across countries. In principle this could reflect the beneficial effects of higher income on corruption. But it also reflects strong causal effects in the opposite direction – less corruption leads to higher income levels. In a recent paper “Growth Without Governance”, we try to sort out these two channels and find that the latter one overwhelmingly dominates the former. In fact, if anything the data suggest that higher incomes could actually worsen corruption, which is just what one might expect in countries where the fruits of growth accrue to entrenched elites. This research strongly suggests that growth alone will not be a very effective tool in reducing corruption.

This leads naturally to another point that was raised – how are trends in corruption measured over time and can countries that are initially in the bottom half of the sample eventually become MCA eligible. The recently-released MCA factsheet does not address this issue in any detail, and we cannot speculate as to what the eventual decision of the US government on this point will be. In our note, we do however discuss some of the difficulties in measuring trends in corruption over time, while noting that there are countries that, even over a relatively short period of three or four years, have shown substantial upward and downward trends.

Finally, one comment correctly noted that margins of error need not improve if additional data is of low quality. This is certainly correct. Moreover, and attractive feature of our methodology is that these sources will also receive less weight in the overall scores.

From: topholt@ofir.dk

Dear Sirs

First of all, I would like to emphasise that I\'m very impressed with the quality of research, the World Bank Governance initiative have generate.

But, and there is always a \'but\'(!), I have a number of elaborative questions concerning the second - quite remarcable -main finding of this particular study: The weak and maybe even negative causal effect running from income to quality of governance.

Firstly, this result seems to conflict with D. Treisman\'s conclusion in his seminal article: The Causes of Corruption: A Cross-National Study’, Journal of Public Economics, Vol. 76, 2000.

In his article, Treisman concludes that the strong association between higher income and lower levels of corruption runs from income to corruption in terms of causality. I.e. as countries increase their income, corruption declines.

Treisman e.g. uses countries\' latitudinal distances from the Equator as an instrumental variable for income to reach this conclusion.

Secondly, I take it that quality of governance and level of corruption are or at least should be highly correlated. Certainly, high quality and high corruption government hardly makes sense!

Thirdly, studies on the effects or consequences of corruption do indicate that corruption has detrimental effects on economic growth. However, these studies only seem to be able to account for a small part of the strong income-corruption association.

Fourthly, I know you have criticised the perception indicators regarding corruption, which provide the main part of the empirical foundation of Treisman\'s study, in other articles. But you also seem to include some of the sources on which these indicators are based in the governance indicators? Could you please, provide a not to technical answer to this particular question concerning methodology and evidence?

Fifthly, the supposedly false causal relation from income to lower corruption (better governance) could be (more or less) easily explained by functional theoris of public institutions according to which, public institutions are established and maintained to the extent the benefits of doing so exceed the costs for citizens in total. Related to this, principal-agent theories would point to less scarcity in higher income countries thus reducing incentives among public officials and citizens to engage in corruption or dysfunctionality inducing activities, so to speak, to explain the effect from higher income to lower corruption!

How would you explain the opposite effect in terms of theories on government?

Im perfectly aware that an exhaustive answer to these questions would propably require another article! But in any case, I think a comprehensive approach to the seemingly growing volume of \'quality of governance\'- related literature and studies is highly required. Especially, as these studies provide necessary input to real-life and concrete policy initiatives and hence make quite a difference to a large number of fellow human beings troughout the world. This is, in other words, more than just an academic dispute, so to speak.

Looking forward to your answer.

Sincerely yours,
Carsten Topholt Larsen