In this paper we summarize and respond to some recent critiques of our Worldwide Governance Indicators project (as described in our series of papers Kaufmann, Kraay and Zoido-LobatСѓn (1999a,b) and (2001), and Kaufmann, Kraay and
Mastruzzi (2004, 2005, and 2006)). The latest round of these governance indicators reports on six dimensions of governance every two years since 1996, and annually between 2002 and 2005, for over 200 countries. For brevity we will refer to them here
as the WGI. The WGI are based on the aggregation of perceptions of governance from 31 different data sources provided by 25 different organizations, and so provide a synthesis of the views of a very large and diverse group of stakeholders regarding the
quality of governance across countries. We report the aggregate governance indicators, the underlying individual indicators from all but three of the sources, together with accompanying descriptive papers and a Web-based interactive data tool, at
While this paper is primarily devoted to responding to critiques of the WGI, we think it useful to begin by noting that the WGI have in recent years become among the most widely-used indicators of governance by policymakers and academics.2 The
usefulness of the aggregate indicators in the WGI stems from the fact that (a) they provide very broad country coverage, greater than that provided by any individual data source on governance; (b) by averaging information from many different data sources
they are able to conveniently summarize the wealth of existing information on governance; (c) by averaging they are also able to smooth out some of the inevitable idiosyncracies of individual measures of governance and so be more informative about
the broad notions of governance they are intended to measure than any individual data source; and (d) the estimates of governance are (unusually in this field) accompanied by explicit margins of error that transparently indicate the unavoidable degree of uncertainty associated with measuring governance by any means. Indeed, we think that it is in part because of the widespread use of the WGI that they are increasingly also beginning to attract critiques. We think that this process of discussion and debate of these critiques is very useful in identifying -- but also often discarding -- potential problems that arise in efforts to measure governance.
Here we address eleven specific criticisms of the WGI that are made in four recent papers (Arndt and Oman 2006 (AO), Knack 2006 (K), Kurtz and Shrank 2006 (KS), and Thomas 2006 (T)). AO provide an extensive and very useful survey of the
many different types of governance data available, and in fairness we note that AO have many nice things to say about the WGI, kindly referring to them as "probably the most carefully constructed governance indicators". Here we focus only on addressing their main criticisms, contained in Section 4 of their paper. Similarly, K's focus is on interpreting the available data on trends in corruption in countries in Europe and Central Asia (ECA) between 2002 and 2005, and also contains in our view a very useful and
thorough review of the many different types of data available to measure corruption in these countries. However, along the way he raises several criticisms of the WGI as well as the underlying data with which we disagree. The paper by KS, which is forthcoming
in the Journal of Politics, is primarily focused on critiquing the WGI. We have prepared a fuller response to the issues they raise for publication in the same journal, and we refer the interested reader to that article for details. Finally, T's paper is also a critique specifically of our indicators, and we respond to it here.
We organize the points made in these papers into eleven related critiques, and provide our responses. The first four critiques call into question the usefulness of the Worldwide Governance Indicators for making comparisons of governance over time and
across countries. Critiques 5 and 6 allege various sorts of biases in the individual indicators underlying our aggregate governance indicators. Critiques 7 and 8 concern the independence of the assessments of governance provided by our different data
sources, and the consequences for the aggregate governance indicators. Critique 9 responds specifically to the main thesis of T that the WGI are an "elaborate untested hypothesis" because we fail to provide evidence of "construct validity", a somewhat
obscure term that we define below. Critique 10 deals with concerns primarily of T regarding access to the data used in the WGI. Finally, the 11th critique raised by AO, while not specifically about the WGI, refers to a paper of ours on the causality between
governance and growth that used data from the WGI.3
1818 H Street NW, Washington, DC 20433, , , . The views expressed here are the authors' and do not reflect those of the World Bank, its Executive Directors, or the countries they represent.
For example, the United States Millennium Challenge Account aid program prominently relies on five of the WGI in its procedures for determining country eligibility, see www.mcc.gov for details.
KS also criticize the literature on governance and growth more broadly, and we provide a response, in our forthcoming Journal of Politics article.