The dynamics of political corruption: Re-examining the influence of democracy

Democratization

Volumn 13, Issue 1, 2006, Pages 1-14

  Blake, Charles H ^a   Martin, Christopher G ^b  

^a JAMES MADISON UNIVERSITY   (United States)

^b UNIVERSITY OF CHICAGO   (United States)

Author keywords

Accountability; Corruption; Democracy; Governance

Indexed keywords

ACCOUNTABILITY; CORRUPTION; DEMOCRACY; GOVERNANCE APPROACH;

EID: 30944453288     PISSN: 13510347     EISSN: 1743890X     Source Type: Journal    
DOI: 10.1080/13510340500378191     Document Type: Article

References (53)

1. On corruption's implications for legitimacy, see Mitchell Seligson, 'The Impact of Corruption on Regime Legitimacy: A Comparative Study of Four Latin American Countries', Journal of Politics, Vol. 64, No. 2 (2002), pp. 408-33.
2. Regarding the negative effect of corruption on economic growth, see Paolo Mauro, 'Corruption and Growth', The Quarterly Journal of Economics, Vol. 110, No. 3 (1995), pp. 681-712.
3. Arthur Goldsmith, 'Slapping the Grasping Hand: Correlates of Political Corruption in Emerging Markets', American Journal of Economics and Sociology, Vol. 58, No. 4 (1999), pp. 865-83.
4. Wayne Sandholtz and William Koetzle, 'Accounting for Corruption: Economic Structure, Democracy, and Trade', International Studies Quarterly, Vol. 44, No. 1 (2000), pp. 31-50.
5. Daniel Treisman, 'The Causes of Corruption: A Cross-National Study', Journal of Public Economics, Vol. 76, No. 3 (2000), pp. 399-457.
6. Treisman (note 4) emphasizes his WLS results. In those results, nearly all of the variables that found some support in the OLS results found more robust significance in the WLS models - British colonial status, national wealth, federalism, trade and economic liberalization were all consistently significant. In addition, the uninterrupted democracy variable that lacked significance in the OLS analysis was also significant. In our treatment here, we focus on the OLS findings. Our reanalysis of Treisman's data using the same models did not suffer from serious heteroskedasticity, nor did the models we examine here. Accordingly, we do not find it appropriate to use WLS in this instance.
7. For a concise discussion of these issues that provides two specific examples, see Michael Lewis-Beck, Applied Regression: An Introduction (Beverly Hills, CA: Sage, 1980), pp. 56-63.
8. Montinola and Jackman find that contemporary political freedom is quadratically related to levels of corruption observed in two different data sets than those discussed here - the Business International data for 1980-83 and the Corruption Perceptions Index for 1988-92. They also examine logged GDP per capita, logged government size and membership of the Organization of Petroleum Exporting Countries (OPEC). See Gabriella Montinola and Robert Jackman, 'Sources of Corruption: A Cross-Country Study', British Journal of Political Science, Vol. 32, No. 1 (2002), pp. 147-70.
9. Subsequently, Sung also found linear, quadratic and cubic terms for contemporary political freedom to be significant in a pooled (cross-sectional time-series) analysis of the Corruption Perceptions Index for 1995-2000 that controlled only for purchasing power parity, unemployment, and inflation. See Hung-En Sung, 'Democracy and Political Corruption: A Cross-National Comparison', Crime, Law & Social Change, Vol. 41, No. 2 (2004), pp. 179-94. It is unclear whether or not these studies' results would be robust to the inclusion of the other variables consistently confirmed in past research and discussed in our literature review (particularly British colonial heritage, Protestantism and trade). When we present our models here, we focus on the linear relationship discussed in other previous research; a polynomial transformation does not provide a significantly better fit in any of the five years under examination (1996-2000). Our results are robust to the inclusion of the OPEC variable confirmed by Montinola and Jackman (that was not examined in other research) while the coefficient for OPEC membership is not significant in our data set.
10. For example, Susan Rose-Ackerman, Corruption and Government: Causes, Consequences, and Reform (Cambridge: Cambridge University Press, 1999).
11. Gerald E. Caiden, O.P. Dwivedi and Joseph G. Jabbra, Where Corruption Lives (Bloomfield, CT: Kumarian Press, 2001), p. 236.
12. A. S. Banks, A. J. Day and T. C. Muller (eds), Political Handbook of the World 1997 (Binghamton, NY: Center for Social Analysis Publications, 1997).
13. Michael Alvarez, José Antonio Cheibub, Fernando Limongi and Adam Przeworski, 'Classifying Political Regimes', Studies in Comparative International Development, Vol. 31, No. 2 (1996), pp. 3-36.
14. For a more complete discussion of this measure, see Adam Przeworski, Michael Alvarez, José Antonio Cheibub and Fernando Limongi, Democracy and Development: Political Institutions and Well-Being in the World, 1950-1990 (Cambridge: Cambridge University Press, 2000).
15. A more recently published quantitative study of corruption dynamics similarly employs this operationalization of consolidated democracy as uninterrupted electoral democracy, but the authors code additional years to cover the time period from 1900 to 1995; see John Gerring and Strom C. Thacker, 'Political Institutions and Corruption: The Role of Unitarism and Parliamentarism', British Journal of Political Science, Vol. 34, No. 2 (2004), pp. 295-331. In an examination of the 1997 CPI data using OLS regression, Gerring and Thacker find a statistically significant negative relationship between years of uninterrupted democracy (between 1900 and 1995) and corruption; while their analysis controls for British colonial heritage and Protestantism, it does not control for trade or economic freedom.
16. For a complete discussion, see Freedom House, 'Rating system for political rights and civil liberties'. Retrieved 24 March 2003 from .
17. Some may ask: Why draw the line for consolidation at 20 years? We believe that 20 years is sufficient time to create the expectation in most citizens' minds that democratic principles are likely to continue to frame governance. After 20 years, an entire cohort of young adults has grown up knowing nothing but democracy in their personal experiences. Meanwhile, older adults - even those potentially supportive of non-democratic politics - have had time to adjust their expectations about the durability and the relevance of democratic institutions. For another study that uses a two-decade cutpoint for democratic consolidation, see Arend Lijphart, Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries (New Haven, CT: Yale University Press, 1999), especially pp. 48-55. As a check on the empirical implications of this choice of a 20-year cutpoint, the empirical analysis examines a variety of alternative cutpoints in the regression analysis. We also examine the results using a continuous measure of uninterrupted democracy.
18. Swamy et al. (below) argue that gender relations affect corruption. They find that female participation in the workforce, in the legislature and in ministerial posts are each related with the CPI in their examination of the 'graft index' (Kaufman et al., 1999) and of the CPI; they report a high correlation between these two corruption indicators (r = 0.98). See Anand Swamy, Stephen Knack, Young Lee and Omar Azfar, 'Gender and Corruption', Journal of Development Economics, Vol. 64, No. 1 (2001), pp. 25-55. When these measures of gender relations are added to our model, they are not statistically significant. We have chosen not to report this variable in our main results because the support provided for a microtheoretical foundation is limited. Swamy et al. assert that data from the World Values Survey for the early 1990s demonstrate that women are less tolerant of corruption than men; however, the size of the bivariate association in the WVS data they emphasize is very small (V = 0.059) and remains small in the mid-1990s WVS (V = 0.041). Nevertheless, we control for this variable in note 28 and its inclusion would have no effect on the results.
19. There is a considerable amount of theoretical and empirical work regarding the control variables in this model. For more on the role of national wealth in limiting corruption, see John Sahr Kpundeh, Politics and Corruption in Africa: A Case Study of Sierra Leone (Lanham, MD: University Press of America, 1995);
20. see also Robert Williams, Political Corruption in Africa (Aldershot: Gower, 1987).
21. Regarding the possibility that market-oriented economic practices limit opportunities for corruption, see Bela Balassa, Gerardo Bueno, Pedro-Pablo Kuczynski and M. H. Simonsen, Toward Renewed Economic Growth in Latin America (Washington, DC: Institute for International Economics, 1986);
22. Robert Klitgaard, Controlling Corruption (Berkeley, CA: University of California Press, 1988);
23. and Robin Theobald, Corruption, Development, and Underdevelopment (Durham, NC: Duke University Press, 1990).
24. On the role of trade linkages, see Jennifer L. McCoy and Heather Heckel, 'The Emergence of a Global Anti-corruption Norm', International Politics, Vol. 38, No. 1 (2001), pp. 65-90.
25. Regarding Protestantism as a brake on corruption, see Ronald Inglehart, Modernization and Postmodernization: Cultural, Economic, and Political Change in 43 Societies (Princeton, NJ: Princeton University Press, 1997), pp. 218-24.
26. Treisman (note 4) argues that British colonialism also left an attitudinal legacy that limits corruption.
27. For theories that federalism expands the prospects for corruption, see Frank Anechiarico and James Jacobs, The Pursuit of Absolute Integrity (Chicago, IL: University of Chicago Press, 1996);
28. Rémy Prud'homme, 'On the Dangers of Decentralization', World Bank Observer, Vol. 10, No. 2 (1995), pp. 201-20;
29. and Vito Tanzi, 'Fiscal Federalism and Decentralization: A Review of Some Efficiency and Macroeconomic Aspects', 1995 Annual World Bank Conference on Development Economics, Washington, DC (1995).
30. Finally, regarding the role of contemporary democracy in limiting corruption, see Michael Johnston, 'Party Systems, Competition, and Political Checks against Corruption' in Arnold Heidenheimer and Michael Johnston (eds), Political Corruption, 3rd edn (New Brunswick, NJ: Transaction Publishers, 2002), pp. 777-94
31. and Rose-Ackerman (note 8).
32. For a thorough discussion of how the CPI is calculated, see Transparency International, 'Frequently Asked Questions About the TI Corruptions Perceptions Index (CPI) 2002'. Retrieved 16 January 2003 from .
33. Political corruption, given its illegal nature, is difficult to measure systematically; for a useful discussion of measurement issues, see Lancaster and Montinola, 'Comparative Political Corruption: Issues of Operationalization and Measurement'
34. and Michael Johnston, 'Measuring the New Corruption Rankings: Implications for Analysis and Reform', both in Arnold Heidenheimer and Michael Johnston (eds), Political Corruption, 3rd edn (New Brunswick, NJ: Transaction Publishers, 2002), pp. 865-84.
35. Although the different surveys used in the CPI may emphasize different meanings as to the perceived degree of corruption, none of the sources witness any correlative problems between themselves, leading one to infer that such differences have little impact on the outcome of the study at present. For more discussion and examination of correlation across years of the CPI, see Treisman, 'The Causes of Corruption: A Cross-National Study', pp. 409-12. In the pursuit of caution, some might contend that the CPI is not always and everywhere a valid and reliable proxy measure for the extent of corruption. Regardless, even if one has such doubts, the level of perceived corruption is important in its own right. As Goldsmith (note 2, p. 875) observes, 'With corruption, however, what people think they see may be as important as what actually happens'. The effects of corruption on political and economic behaviour are crucially mediated through perceptions. Perceived corruption is an indirect measure of corrupt behavior but perhaps a more direct measure of its societal relevance for the sorts of political and economic outcomes we highlighted at the outset of this article.
36. GDP per capita data for 1996 come from World Bank, 1999 World Development Indicators on CD-ROM (Washington, DC: World Bank, 1999). Data for the years 1997-2000 come from World Bank, World Development Indicators Online. Retrieved 14 November 2002 from .
37. In two prior cross-national studies, this variable was operationalized using Freedom House's 15-point scale of economic freedom in which higher scores indicate greater degrees of economic liberalization. For details consult R. E. Messick (ed.), World Survey of Economic Freedom 1995-96 (New York: Freedom House, 1996). We changed measures in pursuit of a similar measure that was tracking the variable over the entire period under study. The Freedom House data's correlation with the Heritage Foundation data we employ here is high (r = 0.756) and statistically significant at the 0.01 level.
38. As with the GDP data, trade data for 1996 come from World Bank, 1999 World Development Indicators on CD-ROM and trade data for the years 1997-2000 come from World Bank, World Development Indicators Online.
39. Rafael La Porta, Florencio López-de-Silanes, Andrei Shliefer and Robert Vishny, 'The Quality of Government', Journal of Law, Economics, and Organization, Vol. 15, No. 1 (1999), pp. 222-79.
40. Treisman (note 4).
41. Ibid.
42. Freedom House, 'Spreadsheet with Freedom in the World data for 1972-2000'. Retrieved 4 March 2002 from .
43. We have chosen to code values of 6 and 7 as reflective of a robust democracy because that is how Freedom House itself designed and interpreted this measure. Freedom House terms countries in that range as 'free' while countries on the two points on the other end of the seven-point scale are termed 'not free'. All countries in the intermediate range of the scale are regarded as 'partly free' by Freedom House.
44. The full data set is available online at .
45. One reader of a previous version of this article recommended that we add a correlation table to our existing discussion of collinearity diagnostics. We do not believe that it is desirable to do so because our diagnostics capture collinearity within pairs and within sets of three or more variables; a correlation table only highlights problems within pairs of variables. As noted in note 26, we have posted the data publicly on the Internet; any readers interested in the bivariate correlations can generate correlation tables for all years under analysis.
46. 2 for this revised model is slightly lower. The five independent variables confirmed in full model remain significant in this truncated model - consolidated democracy, national wealth, British colonial heritage, Protestantism and trade. Finally, as discussed earlier, the full model examined here is robust to the inclusion of other control variables occasionally confirmed in past research - female participation in government, female participation in the workforce, OPEC membership and a polynomial transformation of contemporary political freedom.
47. Consolidated democracy, national wealth, British colonial heritage and Protestantism remain significant in all years. The results for federalism and economic freedom are also identical to those for the full data set; both are significant at the 0.05 level in only one year. Contemporary political freedom was only significant in the full data set in the year 2000; once high leverage cases are removed, it would be significant in 2000 only at the 0.10 level (p = 0.078). The trade variable, however, is markedly affected by the exclusion of the high leverage case of Singapore; when Singapore is removed from the analysis, trade is only statistically significant in 1997.
48. Shifting the cutpoint to the various intermediate options between 15 and 20 years also frequently reduces the significance levels of this dummy variable for the years 1997-2000.
49. Miriam Golden and Eric Chang, 'Competitive Corruption: Factional Conflict and Political Malfeasance in Postwar Italian Christian Democracy', World Politics, Vol. 53, No. 4 (2001), p. 621.
50. Silvia Colazingari and Susan Rose-Ackerman, 'Corruption in a Paternalistic Democracy: Lessons from Italy for Latin America', Political Science Quarterly, Vol. 113, No. 3 (1998), p. 448.
51. For a recent effort to develop hypotheses to explain the resilience of corruption in Italy and Japan, see Donatella della Porta, 'Political Parties and Corruption: Ten Hypotheses on Five Vicious Circles', Crime, Law and Social Change,Vol. 42, No. 1 (2004), pp. 35-60.
52. For example, see David Chandler, 'Anti-Corruption Strategies and Democratization in Bosnia-Herzegovina', Democratization,Vol. 9, No. 2 (2002), pp. 101-20. Chandler details the difficulties involved in implementing a successful anti-corruption programme in the newly formed democratic regime in Bosnia-Herzegovina. He contends that anti-corruption laws are less likely to succeed when they are imposed from abroad (rather than emerging from domestic political dynamics).
53. For a thought-provoking examination of a potential vicious cycle between new democracies and corruption, see Seligson (note 1). He demonstrates that corruption undermines legitimacy in four Latin American countries - thereby endangering the consolidation of democratic practices that, over time, might help to reduce the prevalence of corruption.