Corruption: Good news and bad news

1. Corruption: Good news and bad news Jakob de Haan Head of Research DNB Professor of Political Economy, University of Groningen

2. Corruption • Corruption is an act in which the power of public office is used for personal gain in a manner that contravenes the rules of the game (Jain, 2001). • Three conditions: • 1. discretionary power • 2. economic rents • 3. weak institutions

3. Effects of corruption • There is strong evidence that corruption has detrimental effects on economic growth and other economic outcomes. • Although it is sometimes argued that corruption ‘greases the wheels’ of poor institutions, this is generally only true as a second-best outcome; the first-best outcome is to improve the institutions.

4. Persistence of corruption • Theoretical models of corruption generally predict that corruption is immutable. • When there are many corrupt individuals, it is optimal to be corrupt; thus once corruption is widespread, it will be persistent. • In countries where corruption is pervasive, there are no incentives for individuals to fight corruption, even though they would be better off without corruption.

5. New research with Harry Seldadyo • Using the International Country Risk Guide (ICRG) data for the period 1984-2008 for 101 countries, we find strong evidence that corruption is not persistent. • Good news: Many corrupt countries were able to reduce their level of corruption. • Bad news: many clean countries have become less clean over the same period.

6. ICRG Data • The ICRG data is based on perceived corruption by a panel of experts. The level of corruption is expressed on a scale between zero and six, but for the current analysis the index is rescaled from one to seven where a higher score means less corruption. • We use the ICRG data because this is the only indicator of corruption that is available for a long time period on a consistent basis.

7. Losers • Total 69 losers, i.e., countries with an increase in corruption between 1984-2003. Zimbabwe has the highest relative change in this group. It drops from a relatively moderate level (3.92) to the lowest level (1.00). The change in corruption does not only occur in countries categorized as moderately or highly corrupt in 1984, but also in clean countries, like Canada, Ireland, Singapore, Switzerland, and the United Kingdom fall into this category.

8. Winners • There are 23 winners, i.e., countries with a declining level of corruption over our sample period. Most of them were in the low and mid positions in the corruption ranking in 1984. Examples are South Korea, with the smallest positive relative change, as well as Bangladesh, the country with the biggest improvement. Also some African countries saw their level of corruption decline.

9. Other ways to show convergence • We calculate the correlation between the level of corruption in 1984—the starting point of our sample period—and corruption in subsequent years. In case of corruption persistence, the correlation coefficients should be stable over time (-persistence). • Similarly, we also run a series of cross-country auto regressions between corruption in year t and its lagged values—with lags of 1, 2, …, 24 years. If corruption were persistent, the constant and the slope should be zero and one, respectively.

11. Distribution over time

12. Cont.

13. Conclusions (1) • Is corruption really persistent? In contrast to what seems to be conventional wisdom, we argue that it is not. We find that corruption does not change much from year to year. However, viewed from a longer-term perspective, many countries see their level of corruption decrease or increase over a long- term horizon. The good news is that there are more countries with a decline than countries with a rise in their level of corruption. The bad news is that many clean countries have become less clean over time.

14. Conclusions (2) • Our evidence points to a clear convergence process. The correlation between the levels of corruption in two subsequent years is high, but the larger the distance between the two years becomes, the lower is the correlation. Not only do corrupt countries become less corrupt, clean countries tend to become less clean over time. This pattern is confirmed in our analysis of the dynamics of the distribution of the data. There is a significant modality shift in the corruption distribution over time from a bimodal to a unimodal distribution.