Analysis of the Corruption Perception Index by Transparency International

Introduction

Over the last decade, much attention has been placed on corruption all around the world through the work of Transparency International (particularly their Corruption Perceptions Index, their Global Corruption Report, and their Bribe Payers’ Index) as well as through other organizations.

While organizations have conducted research, results have not always been conclusive due the ‘tabooness’ of collecting the data – some folks just do not want to talk about, reveal their organization’s impact by, or their organization’s participation in corruption. Because corruption is a form of illicit and illegal activity that may or may not be made open, the perceptual measures are the next best thing available to gauge the extent of its existence other than the all but impossible to obtain objective actual quantity.[i] This paper aims to critically analyze and assess the Transparency International Corruption Perception Index.

Overview

Transparency International, a corruption-fighting organization based in Berlin, Germany publishes the Corruption Perceptions Index (CPI) annually since 1995. Transparency International (TI) formed an Index Advisory Committee (IAC) in 1996 to offer advice on its global corruption measurement tools. The role of the committee is to provide technical expertise and advice in the development and strengthening of the methodologies used by TI to measure corruption and governance. Members of the committee are economists, statisticians, and social and political scientists.

The CPI indicates the degree to which corruption is perceived to exist among public officials and politicians in the countries included in the report. The CPI 2008 ranks 180 countries from the least perceived to the most perceived corrupt. The CPI is a composite survey, reflecting the perceptions of business people and country analysts, both resident and non-resident. It draws data from 14 different surveys from 10 independent institutions. For a country to be included, it must be featured in at least three of the 14 surveys. As a result, a number of countries, including some which could be among the most corrupt, are missing because not enough survey data is available.

More than two-thirds of the 180 nations in the 2008 CPI, including many of the world’s most poverty stricken, score fewer than 5 out of a clean score of 10. Corruption is perceived to be rampant in Bangladesh, Chad, Democratic Republic of Congo, Sudan, Guinea, Iraq, Myanmar, and Somalia, countries with a score of 2 or fewer in the 2008 index. Countries with a score of 9 or higher, with very low levels of perceived corruption, are predominantly wealthy countries, namely Denmark, New Zealand, Sweden, Singapore, Finland, Iceland, and Switzerland.[iv] Peter Eigen, the former Chairman of Transparency International said in his interview, ‘Poverty is no excuse for tolerating corruption, but neither can wealth protect a country against suffering from it in the future. Constant vigilance is essential everywhere’.[v]

Table 1. 2008 Corruption Perceptions Index [vi]

The CPI continues to indicate a vicious circle of poverty and corruption. Corrupt political elites in the developing world, working hand in hand with unscrupulous investors, are putting private gain before the cornerstone of capitalism, the welfare of citizens and the economic development of their countries. The world’s poorest peoples are the greatest victims of corruption.[vii]

Measurement and Validity

The CPI provides data on extensive perceptions of corruption within countries. It ranks countries on a one to ten scale; a perfect 10.00 would be a totally corruption-free country. The CPI is a composite index that makes use of surveys of businesspeople and assessments by country analysts. This index is based on a weighted average of approximately ten surveys of varying coverage. For example, 14 sources could be included in the 2008 CPI, originating from 12 independent institutions[viii].

Therefore, the CPI is an aggregate perception indicator based on single perception indexes computed from surveys of business people, local citizens, and experts’ opinions. All sources employ a homogenous definition of the ‘extent of corruption’, in which this term equally reflects the frequency of bribes and the total value of bribes paid. The definition of corruption generally is ‘the misuse of public power for private benefit, such as bribing of public officials, kickbacks in public procurement, or embezzlement of public funds. Each of the sources also assesses the ‘extent’ of corruption among public officials and politicians in the countries in question.’[ix]

It’s worth mentioning that a source must provide a ranking of nations to be included in the CPI sources. For example, if a source conducts surveys in a variety of countries but with varying methodologies, the condition is not met because comparison from one country to another is infeasible in this case. Moreover, the sources must measure the overall extent of corruption. This condition is not met if aspects of corruption are mixed with issues other than corruption such as political instability or nationalism or if measures are changes instead of the levels of corruption.[x]

Additionally, comparisons to the CPI for different years should be based on a country’s score and not its rank since a country’s rank can change simply because new countries enter the index and others drop out. A higher score suggests that respondents provided better ratings, whereas a lower score suggests that respondents revised their perceptions downward. Nevertheless, year-to-year changes in a country’s score could result from a changing sample and methodology, as well as from a changing perception of a country’s performance. Old sources drop out of a recent CPI index and new sources enter, which disturbs the consistency of the assessment.[xi]

To the extent that changes can be traced to a change in assessments provided by individual sources, trends can be identified. For example, countries whose 2008 CPI decreased by at least 0.3 relative to the 2005 CPI, where this drop is not a result of technical factors, are Barbados, Belarus, Costa Rica, Gabon, Nepal, Papua New Guinea, Russia, Seychelles, Sri Lanka, Suriname, Trinidad and Tobago, and Uruguay. It should be noted that these countries are in democratization development process now. Improvements of at least 0.3 can be observed for Argentina, Austria, Bolivia, Estonia, France, Guatemala, Honduras, Hong Kong, Japan, Jordan, Kazakhstan, Lebanon, Moldova, Nigeria, Qatar, Slovakia, South Korea, Taiwan, Turkey, Ukraine, and Yemen.[xii]

Countries whose 2008 CPI decreased by at least 0.3 relative to the 2005 CPI, where this drop is not a result of technical factors, are Brazil, Cuba, Israel, Jordan, Laos, Seychelles, Trinidad and Tobago, Tunisia, and USA. Improvements of at least 0.3 can be observed for Algeria, Czech Republic, India, Japan, Latvia, Lebanon, Mauritius, Paraguay, Slovenia, Turkey, Turkmenistan, and Uruguay.[xiii]

On the other hand, there is high correlation between different sources as shown by the standard Pearson correlation and Kendall’s rank correlation, suggesting that the sources do not differ considerably in their assessment of the levels of corruption. The standard Pearson correlation and Kendall’s rank correlation are, on average, 0.87 and 0.72 respectively, averaged over different pairs of sources.

Since each source uses its own scaling system, the data should be standardized before each country’s mean value can be determined. This is done by using matching percentiles; the percentile rank of countries – not the score – is the only information processed from each source. Another standardization is done in a second step by beta-transformation, a monotonic transformation to increase the standard deviation to the previous year’s value, while preserving the range from 0 to 10.

Paul G. Wilhelm, a professor in the School of Business at the University of Texas, conducted a study to examine the validity of the CPI. He used one criterion (macro-economic outcome) in his validation analyses and found a highly significant correlation (p < 0.001; r = 0.86) between the CPI and real gross domestic product per capita thereby supporting validity of the CPI.[xiv] This highly significant correlation supports the validity of the CPI.

On the other side, critics point out that the CPI is imprecise since its components often do not measure the same thing. The components themselves are often imprecise, and the accuracy of the CPI will depend on the accuracies of the components in a particular year. In the same way, the type of data used to create the CPI varies from one year to the next, and comparing the CPI ranking of countries from one year to the next demands care. In addition, the CPI relies primarily on the perceptions of a handful of country experts. These perceptions are distorted by a variety of factors, including media coverage, culture, and personal experience/interests. Similarly, Razafindrakoto and Roubaud argue that experts systematically overestimate the level of corruption suffered by the citizens.[xv]

In their study, the authors coordinated two types of surveys on the same subject, the petty bureaucratic corruption experienced by the population in their interactions with government officials, in eight sub-Saharan African countries (Benin, Burkina Faso, Cote d’Ivoire, Madagascar, Mali, Niger, Senegal, and Togo). Their study does not address large-scale corruption. The first type of survey covering a sample of over 35,000 people takes an objective measure of the frequency of petty bureaucratic corruption and its characteristics. The second type (the mirror survey) reports on 350 experts’ opinions on the same matter. They also found that experts who contribute to the CPI overestimated four or fivefold the extent to which households in some francophone African countries experienced corruption, compared to survey of households evidence.[xvi]

Discussion and Conclusion

The CPI is a universally accepted measure of international corruption. The lack of a widely agreed upon definition of what counts as corruption is an obstacle to measuring it. Likewise, it is virtually impossible to come up with precise objective measures of corruption since corruption is clandestine. However, tracking perceptions about corruption can be a useful way of measuring it and monitoring the success of governments’ anti-corruption strategies. However, averaging over several numbers improvers the accuracy.

The indices reflect people’s self-reported perception rather than objective measures. Yet, this perception can be different from reality and from one source to another. However, Wei argues that the pairwise correlations among the indices are high, despite the different sources of the surveys. For example, the correlation between the BI and TI indices and that between BI and GCR indices are 0.88 and 0.77 respectively.[xvii] These high correlations suggest that statistical inference on the causes and consequences of corruption is not very responsive to the choice of corruption index.

On the other hand, Donchev and Ujhelyi provide evidence from the International Crime Victimization Survey that reported corruption perception may be weak predictor of actual corruption experience, and they suggest that corruption perception indices measure corruption perception and there is little compelling evidence that they measure corruption experience.[xviii] Corruption experience might require more objective measures of corruption. They also argue that the respondent assessment could be attitudinal if the respondent did not have any personal experience with corruption. These attitudes may also be affected by individual characteristics.

For instance, a more educated respondent living in an urban area might be more knowledgeable about politics and the operation of the bureaucracy, and might be more critical of certain behaviors. It is also likely that he/she might have heard of more concrete examples of corruption from personal contact or the media, and this might cause him/her to report higher corruption perception. Generally, specific, well-publicized events might have a large impact on the respondents’ perception of corruption. In addition, someone who benefits from a corrupt climate, such as an entrepreneur with political ties, may be reluctant to label or view corrupt acts as corruption. Attitudes will also be influenced by country characteristics, which include the norms of behavior of political leaders or officials, and the political culture in general.

Therefore, corruption perception can be different than corruption experience, and thus in order to understand the causes and determinants of actual corruption and to test theories about actual corruption, better measures of actual corruption might be needed. Corruption perception indices might have to be regarded as measuring corruption perception, but not necessarily corruption experience, but this is no way diminishes their importance or usefulness. In their opinion, many studies using corruption perception indices might be usefully rethought as telling us something about the determinants and implications of corruption perceptions, and political trust more generally.

In a similar vein, Olken examines the empirical relationship between beliefs about corruption and a more objective measure of it in the context of a road-building program in rural Indonesia.[xix] He suggests that perceptions data should be used for empirical research on the determinants of corruption with considerable caution, and that there is little alternative to continuing to collect more objective measures of corruption, difficult though that may be.

The bottom line is that one should undertake a careful analysis of the validity of corruption indicators. In addition, there is need for further collection and analysis of quantitative and qualitative survey information from firms and households to help in improving our understanding of how corruption impacts different social groups in various countries, which will also contribute to more effective design and implementation of anti-corruption strategies. Furthermore, those measuring corruption should make efforts both to minimize measurement error and to be transparent about the error that inevitably remains. On the other hand, imprecision does not mean that indicators are unreliable. Rather, explicit margins of error allow users to be clear about the conclusions that can and cannot be made with confidence based on a particular measure.

Peter Eigen said on the release of the 2005 CPI, ‘Corruption is a major cause of poverty as well as a barrier to overcoming it. The two scourges feed off each other locking their populations in a cycle of misery. Corruption must be vigorously addressed if aid is to make a real difference in freeing people from poverty’.[xx] Transparency International continues to point out that good governance and transparency are essential to sustainable development. Political leaders of the world must set the framework for investment such that the rule of law is applied and enforced fairly so that industries are sustainable for the development of the local economy.[xxi]

Endnotes

• [i]  L Soon, ‘Macro-economic Outcomes of Corruption: A Longitudinal Empirical Study’, Singapore Management Review, Vol. 28, no. 1, 2004, pp. 63-72.
•  Transparency International. Corruption Perceptions Index 2008, retrieved April 2009, <http://www.transparency.org/news_room/in_focus/2008/cpi2008>
• TI CPI, 2008
•  Ibid.
• P Eigen, ‘Clean Sheet: Transparency International’s New Chapter’, OECD Observer, 252/253, 2005, pp.16-17.
• TI CPI, 2008
• Ibid.
• JG Lambsdorff, ‘The Methodology of the Corruption Perceptions Index 2007,’ Transparency International (TI) and University of Passau, September 2007, retrieved April 2009, <http://www.transparency.org/content/download/23965/358196>
• TI CPI, 2008
• P Wilhelm, ‘International Validation of the Corruption Perceptions Index: Implications for Business Ethics and Entrepreneurship Education,’ Journal of Business Ethics, No. 35, 2002,  pp. 177-189
• M Razafindrakoto and F Roubaud, ‘Are International Databases on Corruption Reliable? A Comparison of Expert Opinion Surveys and Household Surveys in Sub-Saharan Africa,’ Document de Travel Dial, DT/2006-17, 2006, retrieved April 2009, <http://www.dial.prd.fr/dialjublications/PDF/Doctravail/2006-17_english.pdf>
• S Wei, ‘Why is Corruption so Much More Taxing than Tax, Arbitrariness Skills,’ NBER Working Papers No. W6255, 1997
• D Donchev and G Ujhelyi, “Do Corruption Indices Measure Corruption?,” Economics Department, Harvard University, 2007, retrieved April 2009, <http://www.people.fas.harvard.edu/~ujhelyi/corrmeasures.pdf>
• B Olken, ‘Corruption Perception vs. Corruption Reality,’ Center for Economic Policy Research (CEPR), Discussion Paper 6272, 2007
• Transparency International. Corruption Perceptions Index 2005, retrieved April 2009, <http://www.transparency.org/news_room/in_focus/2005/cpi_2005>
• Transparency International. Corruption Perceptions Index 2006, retrieved April 2009, <http://www.transparency.org/news_room/in_focus/2006/cpi_2006>