Is There a "Consensus" towards Transparency International's Corruption Perceptions Index?
Saha, Shrabani, Gounder, Rukmani, Su, Jen-Je, International Journal of Business Studies
Given the clandestine nature, corruption is intrinsically a complex phenomenon and hard to measure. This paper examines whether Transparency International's corruption perception index converges towards consensus over time? Furthermore, we estimate the speed of adjustment towards general agreement. The results indicate differences in the degree of concordance, i.e. high level of agreement for the mostlyclean and most-corrupt countries but disagreement remains high for the mediumcorrupt countries. The speed of converge is high for the most-corrupt and mostlyclean countries and a decline for the medium corrupt countries.
Keywords: corruption, Perception Index, panel data
(ProQuest: ... denotes formula omitted.)
Studies on corruption have been cautious in interpreting their empirical results given the nature of the definition of corruption and measurement of such corrupt behaviours and activities. While the studies have noted harmful effects of corruption on economic growth, development and its detrimental nature to all societies, but the difficulties in the measures have queried the risk of analysing individual organisations.1 Until recently, research on corruption has been more illustrative than empirical due to the difficulty in measuring relative corruption across countries. Given its clandestine nature, corruption is intrinsically a complex phenomenon and hard to measure due to its actual and perceived notions.
The actual data on corruption to a large extent depend on the effectiveness and capacity of a country's judiciary system in prosecuting corrupt behaviours. Moreover, the objective data of corruption mostly reflects the success of anti-corruption initiatives rather than the actual levels of corruption.2 In order to address measurement problems several organisations (e.g. business risks analysts, polling organisations) have computed the level of corruption based on various perceptions. The perceived corruption indices have been constructed on the basis of survey responses of business people, academics and local residents. Studies have utilised the perception-based-indices as a quantitative measure of corruption, amongst that Transparency International's (TI) corruption perceptions index (CPI) is the most widely used measure in economic studies.3 TI's corruption perception index is highly correlated with other perceived measures of corruption (e.g. World Bank's control of corruption index and International Country Risk Guide's corruption index). This paper analyses an intriguing stylised fact about Transparency International's corruption perceptions index using the standard deviation of CPI rankings and links this to the concordance of the perception over time.
The paper is structure as follows: the next section discusses the Transparency International's corruption perceptions index and section III presents the trends in the standard deviation of the CPI rankings. Section IV explains the findings in terms of the mostly-clean countries, the medium corrupt countries and the mostly-corrupt countries. We examine two issues, first, whether the CPI scores converge towards a degree of concordance, i.e. does the degree of disagreement among the polls decline over time? Second, we estimate the speed of adjustment towards general agreement. The results obtained for 180 countries from 1995 to 2008 suggest that CPI tends towards a higher degree of concordance over time and the speed of convergence is high for the mostly-clean and most-corrupt countries but the degree of disagreement remains high for the middle-range countries (i.e. medium corrupt countries). We attempt some possible explanations for these results. The final section presents the conclusion.
II. CORRUPTION PERCEPTIONS INDEX (CPI): AN OVERVIEW
Transparency International's CPI is a composite index based on individual surveys from various sources. The strength of CPI is the use of multiple sources of data and multi-year averages which increases the reliability of the index (Seligson, 2006). …