Corruption, democracy and Asia-Pacific countries

Abstract

This paper argues that the relationship between democracy and corruption is non-monotonic. When a country shifts from autocratic rule to highly imperfect democracy (an ‘electoral democracy’) it is frequently perceived that the level of corruption increases. Conversely, when the democracy level is already relatively high (approaching ‘mature democracy’) an increase in the level of democracy is typically expected to decrease the level of corruption. To assist with our discussion of these issues, before going on to the empirical part of the paper, we look specifically at the case of South Korea to illustrate how corruption responded to an increasing level of democracy. Using panel data, we find strong empirical support for the non-monotonic relationship. For Asia-Pacific countries, we find that the democracy-corruption relationship becomes negative, at a surprisingly high level of democracy. Moreover we also find that the South Asian region is the most corrupt.

Keywords:

• corruption
• democracy
• Asia-Pacific countries
• panel data

JEL classification:

• C23
• D73
• K42
• O50
• O53

Acknowledgements

The authors are grateful for comments both from the referees and from participants in the Globalisation and Development Centre workshop “Economic and Policy Developments in East Asia” held in October of 2011 at Bond University.

Notes

1. See Moran (2001), Montinola and Jackman (2002) and Sung (2004).

2. We define ‘electoral democracy’ as where elections are sufficiently free and fair that there is a realistic possibility that there can be a change in government, but where normal standards of due process do not hold and institutions that are fundamental to liberal democracy, such as an independent judiciary, are not functioning correctly. See Diamond (1996).

3. The degree to which corruption is unpopular is clearly going to differ between countries. Anti-corruption measures could be quite unpopular where a large proportion of the population are poorly paid petty officials. However, in general, petty corruption generates more ill-feeling amongst the population than high-level political corruption.

4. See Sen (1999).

5. A recent example of a murder, where there seems little prospect of those responsible being brought to justice, is the shooting of Indian anti-corruption activist Shehla Masood, see Chamberlain (2011).

6. In a report by the International Crisis Group (2008, p. 16) it is argued that the Bangladeshi military has attempted to extend its influence by using dubious corruption charges to remove officials and replace them with people who they have handpicked.

7. We obtained these historical Transparency International CPI figures from You (2005, p. 14). Transparency International only publishes CPI data going back to 1995.

8. For details see the Transparency International entries in Table A1. Here corruption is defined as the abuse of entrusted power (public sector power) for private gain.

9. See for example, Ades and Di Tella (1997); Johnson, Kaufmann, and Zoido-Lobaton (1998); Sandholtz and Koetzle (2000); Treisman (2000), Montinola and Jackman (2002); Gupta, Davoodi, and Alonso-Terme (2002); You and Khagram (2005); and Emerson (2006).

10. For details see the Freedom House entries in Table A1.

11. See Nelson and Singh (1998), Ades and Di Tella (1999), Sandholtz and Koetzle (2000), Bohara, Mitchell, and Mittendorff (2004) and Sung (2004) for details.

12. Following Montinola and Jackman (2002), we have utilised government share of GDP for government size. It is argued that higher government expenditure is due to the higher public sector wages which may in turn reduce corruption.

13. Two-way fixed effects estimate the effects inserting the dummies for each country and each period which reduces the degrees of freedom of estimation. Instead, period fixed effects with regional dummies incorporate dummies for each region and each period that increases the degrees of freedom and improves the estimation results. A missing data problem prohibits performing random effects estimation.

14. Dynamic panel estimation results are also consistent with these results although not reported will be available upon request.

15. As is to be expected we find a highly significant negative relationship between corruption and the rule-of-law. It should be noted here that rule-of-law and our democracy indices are not as closely related as one might think. India is a prime example of a country that does well with regard to democracy, but does poorly with regard to the rule-of-law. In contrast, Singapore exhibits a strong rule-of-law with moderate democracy.

16. It is noted that the estimated coefficient of Africa is negative and significant in column (3). Examination of CPI scores confirms that Africa, Asia, Eastern Europe, Latin America and the Middle East countries are all perceived to be more corrupt than the Organisation for Economic Co-operation and Economic Development (OECD) countries. The estimation results of the regional dummies, excluding the control variables, show that African countries are perceived to be significantly more corrupt than Asia, Latin America, Middle East and the OECD countries (not reported here). However, after controlling for economic development, only Asia and East European countries remain more corrupt than OECD countries.

17. It is interesting to note that the coefficient of log (RGDP) is positive and significant which attracts special attention towards the Asia-Pacific countries. The result suggests that an increase in income per capita increases corruption which challenges the existing theory. We feel that it is something of a feature of this region that countries with fairly moderate to high levels of corruption also have experienced strong GDP growth.