Explaining corruption at the village and individual level in India. Findings from a study of the Panchayati Raj reforms

Asian Survey

Volumn 45, Issue 5, 2005, Pages 756-776

  Widmalm, Sten ^a  

^a UPPSALA UNIVERSITY   (Sweden)

Author keywords

[No Author keywords available]

Indexed keywords

CORRUPTION; CULTURE; DECENTRALIZATION; PUBLIC SECTOR;

ASIA; EURASIA; INDIA; SOUTH ASIA;

EID: 31344452682     PISSN: 00044687     EISSN: None     Source Type: Journal    
DOI: 10.1525/as.2005.45.5.756     Document Type: Article

References (58)

1. In this article, corruption is defined as the abuse of public office for private benefit. For further discussion of this definition, see Michael Johnston, "The Search for Definitions: The Vitality of Politics and the Issue of Corruption," International Social Science Journal 48:3 (September 1996), pp. 321-35
2. Vito Tanzi, IMF Staff Papers 45:4 (December 1998) (on the IMF website, ), accessed October 4, 2004.
3. See, for example, the Institute of Social Sciences, George Mathew, ed., Status of Panchayati Raj in the States and Union Territories of India 2000 (New Delhi: Concept Publishing, 2000), pp. 10-11. In the countryside, the reform stipulates that under state government the lowest level unit will be the village panchayat (council of elders), with a number of its elected representatives drawn from the population that the panchayat legally covers. Each (elected) chairperson in the panchayat then becomes a representative at the next level - the intermediary level - most commonly known as the block (tehsil, taluk, or mandal). However, smaller states may opt not to provide an institution at this level. The block panchayats then elect a chairperson who becomes a representative at the district (zilla) level. In effect, this three-tier system below the level of the state reaches out to even the most remote parts of rural India. However, the size of the village panchayats varies quite a lot between states. In Madhya Pradesh, commonly a group of three to four villages with a population of a few thousand each comprise one village panchayat. In Kerala, the word village does not have the same meaning, and houses and settlements can spread out over vast areas not geographically connected the way they are in Madhya Pradesh. In Kerala, a village panchayat can cover a population of 30,000 people.
4. Ibid.
5. Ibid. According to George Mathew and the Institute of Social Sciences, there were 523 district panchayats in 2000, with an average size of two million inhabitants. There were 5,912 block panchayats covering a population of about 200,000, and 231,630 village panchayats, which would include villages with 1,000 to 30,000 inhabitants. The situation in the cities is somewhat different, but in this study we have only surveyed the rural areas. Altogether, about three million members were elected to the panchayat institutions in rural and urban areas in India in 2000. Mathew, Status of Panchayati Raj.
6. Citizens in both Kerala and Madhya Pradesh claim that the reforms have significantly increased the powers of the citizenry. See Sten Widmalm, "What Are Decentralization and Panchayati Raj Reforms, and Who Likes Them?" in Decentralized Governance: Essays in Honour of George Mathew, L. C. Jain, ed. (Hyderabad: Orient Longman, Ltd., forthcoming).
7. Furthermore, a study carried out by Chattopadhyay and Duflo shows that villages where the position of head of the village council was reserved for a woman provided public amenities preferred by women to a higher degree than other villages (Chattopadhyay, Raghabendra, and Esther Duflo, Women as Policy Makers: Evidence from an India-Wide Randomized Policy Experiment, 2003, , accessed June 28, 2004.
8. It should be noted that this should not be confused with the kind of absenteeism caused mainly by a poor infrastructure or very low wages. In both Kerala and Madhya Pradesh, the wages for teachers and health sector personnel are relatively high. Nonetheless, the process of selecting villages that were corrupt was a real challenge. The most common way to approach this research problem is to use the surveys and responses to questions dealing with levels of satisfaction or dissatisfaction with public services to indicate which areas are plagued by corruption. The risk in this method is that citizens may confuse inefficiency with corruption. In this project, we used local contacts and informants who were well known to the organizations that helped carry out the surveys. They were asked to identify villages that had either health workers or teachers who hardly ever turned up for work and were well known to be working privately on the side - for example, running a clinic in a nearby larger village or town, or giving private lessons to students in the cities. As we shall see later, there was in general a correspondence between the survey results and our classification of the villages. However, the survey also showed that we could identify the corrupt health workers fairly well with questions regarding bribes, but those questions would not help us identify villages with corrupt teachers. The methodology is further discussed in Sten Widmalm, Decentralization and Governance in India (forthcoming 2005).
9. The random sample drew from election rolls covering the areas selected. The surveys were carried out in 2001 with help from the Madhya Pradesh Institute of Social Science Research, in Ujjain; Samarthan, in Bhopal; and Health Action by People and the University of Kerala, in Thiruvanantapuram. The average response rate was about 70% for the mass survey, including all 24 panchayats in the study. The survey personnel received detailed information about its background and the ideas guiding the design of the questionnaires. The main study was preceded by a pilot study undertaken in 2000 with the support of the Institute of Social Sciences in New Delhi and the organization Debate in Madhya Pradesh.
10. For a discussion of the problem of selecting cases through dependent variables, see, for example, Gary King, Robert O. Keohane, and Sidney Verba, Designing Social Inquiry (Princeton: Princeton University Press, 1994), ch. 4.
11. Naturally, the diversity and size of these states made it hard to meet the latter criteria in several respects. For example, the caste composition in Madhya Pradesh varied so much that we mainly had to make sure that the relative dominance of Rajputs was the same in most places we surveyed. So, in a strict sense, the sample can only be said to represent the 24 villages selected, and that the extent that we can generalize from these villages to the state level cannot be known exactly. However, if we are satisfied that this study is an exploratory project where the villages are chosen on the best available information indicating their representativeness, then the conclusions reached here can at least be viewed as useful until larger-scale surveys become available.
12. Bimaru is an acronym for Bihar, Madhya Pradesh, Orissa, and Uttar Pradesh and resembles the word bimar (ill or sick, in Hindi).
13. This is so at least as measured in the criteria of Human Development Indicators. For example, the literacy rate in Madhya Pradesh was 44% in 1991, 58% for men, and 29% for women.
14. Susan Rose-Ackerman, Corruption and Government: Causes, Consequences, and Reform (Cambridge: Cambridge University Press, 1999).
15. We can look back at social science studies in the 1960s that provided deep insights into the varieties of meaning in transfers of wealth (for example, M. D. Sahlins, Tribesmen (Englewood Cliffs: Prentice-Hall, 1968) and see that in these studies it was common to focus on the problems of transferring Western-rooted concepts to other societies, such as the concept of corruption. For example, David H. Bayley's statement from 1966 in an article about corruption in India is illustrative: "The man who in many non-Western countries is corrupt in Webster's sense is not condemned at all by his own society. Indeed, he may be conforming to a pattern of behaviour his peers, family, and friends strongly support and applaud." David H. Bayley, "The Effects of Corruption in a Developing Nation," Western Political Quarterly 19:4 (1966), pp. 719-32.
16. Rose-Ackerman, Corruption and Government, p. 104.
17. Ibid.
18. Akhil Gupta, "Blurred Boundaries: The Discourse of Corruption, the Culture of Politics, and the Imagined State," American Ethnologist 22:2 (1995), p. 393.
19. See, for example, Atul Kohli, Democracy and Discontent: India's Growing Crisis of Governability (New York: Cambridge University Press, 1990).
20. Akhil Gupta, "Blurred Boundaries," pp. 375-402.
21. See Robert D. Putnam, Making Democracy Work (Princeton: Princeton University Press, 1993), especially pp. 171-76. The reference to Granovetter is found on p. 175.
22. Rose-Ackerman, Corruption and Government, pp. 162-74. Here, Rose-Ackerman explores important routes for collective or individual complaints and the role of protection for those who complain.
23. Susanne Hoeber Rudolph, "Civil Society and the Realm of Freedom," Economic and Political Weekly 35:20, pp. 1762-69, see p. 1762.
24. When Hardin discusses ethnic mobilization, he argues that the reason that ethnic groups can co-operate effectively and overcome the collective action problem is because individual and group interests coincide. And when such cooperation occurs: "the result is often appalling." Russel Hardin, One for All (Princeton: Princeton University Press, 1995), p. 5.
25. Hans Blomkvist and Ashok Swain, "Investigating Social Capital and Democracy in India," Economic and Political Weekly (February 24, 2001), pp. 639-43
26. Ashutosh Varshney, Ethnic Conflict and Civic Life (New Haven: Yale University Press, 2002);
27. Robert Putnam, Bowling Alone (New York: Simon and Schuster, 2000).
28. Also see Susanne Rudolph's essay "Is Civil Society the Answer?" in which she emphasizes the need to make clearer distinctions between various types of associations. In the field of social sciences, such distinctions can be traced back to the 1940s and were highlighted in the early 1980s by Robert Dahl.
29. See Robert Dahl, Dilemmas of Pluralist Democracy: Autonomy vs. Control (New Haven: Yale University Press, 1982).
30. Mutz has written an excellent article on this topic, which also shows the contributions of the social sciences in the 1940s and the 1950s by Lazarfeld among others. Diane Mutz, "The Consequences of Cross-Cutting Networks for Political Participation," American Journal of Political Science 46:4 (2002), pp. 838-55.
31. Most studies generally involve attitude surveys in areas generally recognized as corrupt, but they seldom compare with non-corrupt areas. One study of particular interest here was carried out by Miller et al. in the Czech Republic, Slovakia, Bulgaria, and Ukraine. The study seeks to determine whether citizens should be seen as sources of corruption, victims, or accomplices (William L. Miller, Åse B. Grödeland, and Tatyana Y. Koshechkina, "Are the People Victims or Accomplices?" in Crime, Law, and Social Change 29 (1998), pp. 273-318.
32. Ronald Inglehart and Christian Welzel, "Political Culture and Democracy: Analyzing Cross-Level Linkages," Journal of Comparative Politics 36:1 (October 2003).
33. Note also that when the respondents were asked to what extent they thought the ideal view was held by public sector personnel (see Figure 1, Question 27), we received some unexpected answers. Overall, about 40% of the respondents said they thought less than half of public sector employees held such a view. However, in Kerala, which most observers of politics in India would agree is a better-governed state than Madhya Pradesh, 44% answered that they thought less than half of all public servants held views in accordance with the ideal bureaucrat. In Madhya Pradesh, the corresponding figure was 35%. In other words, the better-educated Keralites are more skeptical about their bureaucrats than the citizens included in this survey from Madhya Pradesh. Consequently, this shows that we would have drawn the wrong conclusion if we had used quite general questions about bureaucrats as an indicator of performance at the state level.
34. All the scales that were presented to the interviewees were graded zero to four. Interviewers were instructed to illustrate each step in terms of a meaningful quartile. This worked well, especially in Madhya Pradesh where literacy levels are low, because it is common to refer to measurements using the old "anna" coin denomination for comparison. Until 1957 (when the rupee was decimalized), 16 annas would be equivalent to one rupee. Therefore, people still commonly say "four annas" when they want to say 25%, or one-quarter. So, after zero, we told people that the next step on the scale was equivalent to "four annas," the next "eight annas," and so on.
35. Although there is a significant correlation between support for the ideal bureaucrat and what we could call the acceptance of bribery, we can see that the influence these two variables have on each other is not very strong (n = 1,093; B = -.06; S.E. = .027, significant at 95% level of confidence). Moving one step from left to right on the scale for support for the ideal bureaucrat will only, on average, have the effect of moving 6% of one step from right to left on the scale for attitudes to bribes. Interestingly, in the survey of elites that was carried out, a clearer connection is visible between these two variables (n = 121; B = -.161; S.E. = .055, significant at 95% level of confidence).
36. Inglehart and Welzel, "Political Culture and Democracy."
37. For example, we asked, "How do you think the public (sarkari) services are financed?" The options available were as follows: "Public (sarkari) services are: 1 = financed mainly by fees you pay when you use them; 2 = financed mainly by taxes; 3 = financed mainly by donations." Although all questions would not necessarily measure the capacity to distinguish between the private and the public sphere in a direct way, the test results are at least useful as a proxy for that capacity.
38. n = 1,139; B = -.0.208; S.E. = .031, significant at 99% level of confidence.
39. n = 1,136; B = -.0.065; S.E. = .013, significant at 99% level of confidence. The respondents were asked, "What is your level of education?" and on our scale the number 1 corresponds to the answer: none and illiterate; 2 = none, but can write name; 3 = none, but literate; 4 = primary school; 5 = middle school; 6 = secondary or technical school; 7 = college; 8 = university or polytechnic.
40. For a discussion on the relationship between the level of education and corruption, see, for example, M. S. Alam, "A Theory of Limits on Corruption and Some Applications," Kyklos, 48:3, pp. 419-35.
41. The respondents could indicate their living standard by answering the question, "In what kind of house do you live?" where the options were 1 = pucca (solid brick); 2 = semi-pucca (asbestos, tile); 3 = kutcha (hut or non-brick). The reason for using housing standards as a proxy for income is that most people in the survey answered this question, while the number of people who refused to give information about their income was about 43%.
42. See, for example, Vito Tanzi, "Corruption around the World: Causes, Consequences, Scope, and Cures," IMF Staff Papers 45:4 (December 1998), pp. 559-94.
43. Naturally, several other models were tested but yielded no results that change the conclusions drawn here.
44. This relationship is described in fn. 32.
45. This is so at least according to the survey's set of test questions, which tried to measure this either directly or by proxies.
46. According to Lewis-Beck, "Even beginners know that OLS is simply 'not done' or at least not published, when Y is binary." (See the "Series Editor's Introduction," in Fred C. Pampel, Logistic Regression: A Primer [Thousand Oaks: Sage Publications, 2000], p. v.) Nevertheless, OLS will be used here for the following reasons.
47. First, the sample size is smaller than is recommended for using logit. Second, OLS is known to be robust against violations of assumptions, for example, as discussed in D. N. Gujarati, Basic Econometrics (New York: McGraw-Hill, 1995). Moreover, the models for explanation here were tested using logit. However, none of the results would indicate that the results from the OLS regressions were incorrect.
48. We could imagine that there is some sort of threshold effect where villages below a certain average are always more likely to be low-performing, while villages above the threshold remain unpredictable in outcomes.
49. Note: N = 24; B = .665; S.E. = .347, significant at 90% level of confidence. In other words, moving one step from left to right on the scale gauging opinion about public servants will have the effect of moving almost 70% on the scale for performance.
50. The extent to which the survey could indicate performance is discussed further in fn. 6.
51. Compare this to Inglehart and Welzel's argument that individual support for democratic values may only be "weakly linked to societal-level democracy." See Inglehart and Welzel, "Political Culture and Democracy."
52. Given the way the model for the investigation is constructed, we cannot control for the variables stated here.
53. Ashutosh Varshney, Ethnic Conflict and Civic Life (New Haven: Yale University Press, 2002).
54. The questions about whom one can count on for help are used as indirect indicators or proxies of trust. Other questions regarding trust were also asked but indicated no strong relationship to performance at the village level.
55. Karolina Hulterström, "The Dilemma of Personal and Private Social Capital: A Study of the Creation of Social Capital in Bangladesh," C-level Thesis, Department of Government, Uppsala University (1996). Gibson and Gouws came up with results that were similar to Hulterström's but in South Africa.
56. See James L. Gibson, and Amanda Gouws, "Social Identities and Political Intolerance: Linkages within the South African Mass Public," American Journal of Political Science 44:2 (2000).
57. The relationship between inter-group and intra-group trust is fairly strong (n = 1,155; B = .483; S.E. = .021, significant at 99% level of confidence). The results tell us that moving one step on the inter-group trust scale corresponds on average to moving half a step on the intra-group scale.
58. Anirudh Krishna, Active Social Capital: Tracing the Roots of Development and Democracy (New York: Columbia University Press, 2002).