1 Validity of the CPI, Passau 23-25 June 2005 DIES OECONOMICUS VI 23-25 June 2005 University of Passau Organized by Transparency International Germany and the Passau University On the occasion of 10 years of Corruption Perception Index (CPI). [At the end I placed a reaction that I received from prof. Lambsdorff and the comments placed on both my paper and the Lambsdorff reaction by prof. Austin Murphy] The Validity of the CPI by Dr. Michel van Hulten, Buitenplaats 49, 8212AB Lelystad, the Netherlands Tel. +31-320 221279, e-mail: michelvanhulten@planet.nl, www.corruptie.org 1995 1995 is the year in which the first Corruption Index was published. The winner was New Zealand followed by Denmark, Singapore and Finland. The ranking ended with number 41, Indonesia. The corruption score was still calculated with a precision of two figures behind the comma: New Zealand 9.55 and Indonesia 1.94 on a scale from 1 to 10. Since then, the list was renamed as Corruption Perception Index (CPI) and the precision is now with only one figure behind the comma. This list lifted Transparency International, already being developed as an organization since 1991 under the initial name of Bad Business Practices, and formally established in 1993 in Berlin, to international fame. It created for its reason of existence, corruption, immediately in 1995 and in all following years a high level of attention. The CPI attracted and attracts broad media coverage. Corruption came on the political agenda all over the world. The CPI made corruption understandable for the broad public and created the awareness that it is everywhere. It was a splendid tool to begin the work on corruption, although it was sometimes wrongly understood and misused. 2005 Time to celebrate. Corruption has become known. Fighting corruption has become governmental business in many countries. However, it is also time to reconsider the work done, the flaws in the methodology, available data and published results, in order to improve on what we have. We cannot be satisfied with what was very original and thoroughly done in those early years. With a widening circle of studies contributing to the CPI and more and more countries being listed, we have to grow in the collection of data and their processing in order to get ever improving results.
2 I try below to list some elements about which we will have to think. The conference in Passau could add new elements of thinking and analysis and calculation, and new observations around the world in order to improve our fight against corruption and our promotion of integrity and accountability in public life. More and more corruption? The total value of bribes paid each year as published by the World Bank rose between 1994 and 2004 from 50 billion US dollars to 1.000 billion (one trillion) US dollars. The 50 billion was repeatedly used by the World Bank and by Transparency International (TI) in the late 90s. Early in the first decade of this century 400 billion was used as a reliable estimate or approximation. The 1.000 billion US dollars was published for the first time on the internet website of the Bank in the monthly Newsletter dated 8/3/2004, under the title the costs of corruption. The journalistic form chosen for the publication of this latter figure was an interview with Dani Kaufmann, the World Bank Institute s director for governance. That Newsletter also mentioned an article by Peter Eigen, Chairman of TI, titled Grease Trade, in which the same figure of one trillion dollars was used. Understandably, I was intrigued by the quick rise of the estimated or calculated total value of bribes paid worldwide. I asked by e-mail both authors what study was made and published, justifying this higher figure. TI pointed to the Bank as the source of this information and Kaufmann informed me by e-mail of 1 st of December 2004, that a full fledged report/study is still underway as background to the estimates given in these two articles, which we should expect to have at least in draft soon. He provided me also with the full text of the article from which the Newsletter had quoted and he wrote: The estimation approach is simply based on extrapolating from thousands of survey responses from firms and households on bribes paid, which implies that we get an estimate for procurement bribes from firms and also other types of bribes, from firms and households. The trillion dollar estimate is obviously subject to a margin of error, it could be somewhat less or more. An interesting background figure was given on the internet WB-website: the $1 trillion figure, calculated using 2001-02 economic data, compares with an estimated size of the world economy at that time of just over US$30 trillion, and does not include embezzlement of public funds or theft of public assets. 1 Only 3 percent of the world economy paid in the form of bribes, it sounds rather likely. The report/study mentioned by Kaufmann still is not available. This is regrettable in the light of another remark he made in his 1 st of December e-mail written to me, reading: 1 (see also: http://web.worldbank.org/wbsite/external/news/0,,contentmdk:20190187%7e menupk:34457%7epagepk:34370%7epipk:34424%7ethesitepk:4607,00.html)
3 We are obtaining new enterprise data to try and have an estimate subject to less margins of error (yet they will remain) - - the sense is that an estimate ranging between 500bn -1.2 trillion may be appropriate given data so far, but as we obtain more it may be possible to narrow the range further and thus have a better estimate, which is why we will finalize a draft only in the (near) future. A margin of error as indicated here by Kaufmann, is in my view not justifiably qualified as somewhat less or more, when you publish a figure of 1 trillion as your estimate for world bribes paid. However, I will not concentrate now on this aspect of the new figures of the World Bank. Discussion of the CPI For the discussion of the Corruption Perception Index (CPI) as yearly calculated, ranked and published by professor Johann Graf Lambsdorff and the TI Secretariat in Berlin, it is more interesting to look at the growth between 1995 and 2005 of the estimates for the sum total of bribes paid worldwide, and to the fact that obviously this does not cause a major change in the CPI rank order of countries. It seems to me that this can only be true if not only the total value of bribes paid is growing 20-fold, but also the distribution of this growth over the countries of the world is evenly spread. This seems to me is highly unlikely. This adds to earlier critical remarks I made about the validity of the CPI. 2 In short: - I doubt the methodology used by Lambsdoff as he claims to increase the reliability of each individual score by combining data in order to balance erratic findings from one source by the inclusion of at least two other sources (Lambsdorff s words in the Global Corruption Report 2003). How does he know what findings have been erratic, and why should erratic findings be improved by adding other findings of which cannot be established whether they are erratic? - The CPI has a remarkable characteristic: the level of corruptness in particular economies and societies is - according to the CPI - rather stable: once you are rich, you rank high (corruption-free), and once you are poor you are low on the list (very corrupt). Not a single country moved over the years from the first quartile of the list to the last one. What mixing there is, it is among those in the middle. Does it matter to have anti-corruption programs? - As I see it, the CPI reflects the views of observers ( their perceptions ) all of whom belong to the culture of the Western financial and economic spheres. Their yardsticks are culturally determined. This does not mean that I disregard Lambdorff s constant endeavours to include and give more voice to people from less-developed countries. My point is that also the observers from the lessdeveloped world, sharing their corruption perceptions with us, do study, work and live in a culture dominated so to say by the Harvard Business School and the Davos Economic Summit. It is not a matter of skin-colour but of culture. 2 See my contribution to the Global Forum III on Fighting Corruption and Safeguarding Integrity, held in Seoul 29-31 May 2003, with the title Corruption, Unknown, Unloved, Omnipresent, 14 pp, full text on my website www.corruptie.org.
4 - Remarkable is that seemingly the rank position of Germany and France on the CPI list did not change over the years while in recent years quite some corruption cases involving high level politicians have come up. How is it possible that these did not change the reported perceptions? - The fact that TI started a second listing ranking the bribe-payers, is also an indication that the CPI does not really portray the full extent of corruption. At most the CPI represents a part of the picture (and regretfully, I do not see a way in which the CPI and the Bribe-payers Index could be integrated into one index and list). - The CPI reflects and includes the outcomes of a great number of underlying studies and reviews. That is good. The problem is that perceptions collected are the result of contacts between companies, businessmen and clients, and that these are translated into a perception about a country. Is that acceptable as a methodology? In my view, it is not. Some after-thoughts In the Netherlands we have now a splendid (sic) example of how you can look in two ways at corruption. Our late Prince Bernhard confirmed just before his death that he took indeed a big bribe paid by Lockheed. He was despised because of that in 1976 and he lost some of his privileges. However, in 1951, he was sent by the Dutch Government to Argentina to collect business orders for Dutch industry, weak after the losses of the Second World War. We know that he carried enough governmental funds with him to bribe Argentine officials so that his mission could become a success. He was hailed for the orders that came. After that he headed a great number of other Dutch trade-missions to the great satisfaction of successive Dutch Governments although often bribe-paying was part of the deal. Which ones of these corruption cases added to the perception of the Netherlands as a corrupt country? The Prince as payer or as receiver? A Finnish university professor told me recently that it is common practice in Finland to give a present at any business deal. Nevertheless Finland is at the top of the CPI. If true: are these gifts not perceived as part of the bribing seen elsewhere? One of the participants in the Passau-Conference wrote me: There is an enormous amount of unreported corruption in the USA-- just for instance, a few years ago the Wall Street Journal reported in numerous articles that all retailers in the USA are heavily bribed by manufacturers and other suppliers-- the only exception stated was Wal-Mart, which most recently had one of its own execs caught in some corruption. If unreported, can it contribute to the creation of perceptions? *****
5 20 June 2005 reaction received from Johann Lambsdorff: quote Concerning the term "erratic": Whenever you add data that randomly distribute around a common mean, by aggregation the variance of the mean is lower than that of the individual observations as soon as the error terms are not perfectly correlated. My descriptions thus relate to a standard statistical feature. unquote Reaction received 29 June 2005 from Austin Murphy, professor of Finance at Oakland University, Rochester, USA, also participant in the Passau seminar of June 2005, reacting to the Lambsdorff remark with regard to my 'Validity'-paper. quote Hello Michel, I enjoyed meeting you, and I liked your analysis/discussion. With respect to Lambsdorff's statements on statistics that you wrote: [20 June 2005 reaction received from Johann Lambsdorff: Concerning the term "erratic": Whenever you add data that randomly distribute around a common mean, by aggregation the variance of the mean is lower than that of the individual observations as soon as the error terms are not perfectly correlated. My descriptions thus relate to a standard statistical feature]. He is indeed correct if each of the measures is unbiased. However, as you point out, there probably are substantial biases given the ratings are largely by Westerners (I assume they are dominated by surveys of Western firms?). Adding biased ratings to biased ratings can actually increase the bias if the added ratings are more biased. In fact, the entire concept of measuring only bribes of government officials is incredibly biased. Countries with bigger government sectors (and regulation) will of course show up as more corrupt using that measure just because relatively less of the trade (the businessbusiness sector) is excluded from the measure. China itself may score low on the CPI because its firms are perceived as government firms (because of majority government control), and so more of the bribery there is included in the CPI surveys, whereas the substantial bribery of retailers in the USA by manufacturers is excluded from the CPI. Also surveying businesses allows a bias to begin with. Businesses that are happy with the bribery system are not likely to complain. However, companies that are kept from doing things in countries because of strict regulators who are not bribable are likely to rate such countries as corrupt (and possibly even claim that the amount of the bribery needed to get anything done was just plain excessive).
6 Also, for countries that have made many forms of bribery legal probably would not show up negatively even in a fairer survey (kind of like countries that legalize murder through public lynchings, wars of aggression, etc. would not show up as negatively in mass murder statistics, about which I have written in one of my books, a proposed second edition of which I'll probably offer to send electronically for comments). Then there's the problem of both the CPI and the Schattenwirtschaft indicators ignore illegal tax cheating by rich investors, so there's a bias for the rich there as well. For instance, Germany would probably score 2nd highest on investor tax cheating, with 85% of all the taxes on interest income being illegally avoided by rich German individuals through secret Swiss bank accounts (Germany is exceeded by Italy, which scored 98% cheaters according to one study). I certainly agree with all the studies indicating corruption hurts economic growth, but Peter Eigen's statement at the end that corruption causes poverty (and not vice versus) is utterly absurd. They affect each other. Peter Eigen's use of the USA as an example of a country that grew more rapidly than Argentina over the last century doesn't support his argument at all-- the USA was reputed to be the most corrupt country in the world during much of that time period (it actually grew rapidly because of strong protectionism, imperialism, and large cheap labor imports). Do let me know if you would like to see a copy of the proposed second edition of my 2000 book THE TRIUMPH OF EVIL: THE REALITY OF THE USA'S COLD WAR VICTORY. Greetings, Austin unquote