Hostname: page-component-586b7cd67f-dlnhk Total loading time: 0 Render date: 2024-11-28T01:34:45.542Z Has data issue: false hasContentIssue false
  • English
  • Français

Inequality and Discontent: A Nonlinear Hypothesis

Published online by Cambridge University Press:  18 July 2011

Jack Nagel
Jack Nagel
Affiliation:
Assistant Professor of Political Science and Public Policy Analysis at the University of Pennsylvania
Get access

Extract

At least since Aristotle, theorists have believed that political dis-content and its consequents—protest, instability, violence, revolution—depend not only on the absolute level of economic well-being, but also on the distribution of wealth. Contemporary political analysts have tried to test this ancient assumption using modern statistical methods. Their results are distressingly confusing. One cross-national investigation finds the commonsensical positive linear relation: the more the inequality, the greater the instability. A second study purports to show the opposite relation in the important case of South Vietnam: the greater the inequality, the less the support for revolution. And a third analysis, also of South Vietnam, detects no relation at all between inequality and rebellion.

Type
Research Article
Information
World Politics , Volume 26 , Issue 4 , July 1974 , pp. 453 - 472
Copyright
Copyright © Trustees of Princeton University 1974

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 Russett, Bruce M., “Inequality and Instability: The Relation of Land Tenure to Politics,” World Politics, XVI (April 1964), 442–54.CrossRefGoogle Scholar

2 Mitchell, Edward. J., “Inequality and Insurgency: A Statistical Study of South Vietnam,” World Politics, x (April 1968), 421–38.CrossRefGoogle Scholar

3 Russo, Anthony J. Jr, “Economic and Social Correlates of Government Control in South Vietnam,” in Feierabend, I. K., Feierabend, R. L., and Gurr, T. R., eds., Anger, Violence, and Politics (Englewood Cliffs, N.J. 1972), 314–24.Google Scholar

4 Mitchell (fn. 2).

5 On the Gini index and other measures of inequality, see Alker, Hayward Jr, Mathematics and Politics (New York 1965)Google Scholar, chap. 3; and Atkinson, Anthony B., “On the Measurement of Inequality,” journal of Economic Theory, II (September 1970), 244–63.CrossRefGoogle Scholar Use of the aggregate Gini statistic in a theory based on propositions about interindividual comparisons can be questioned. The leap will be justified here only by intuition and by the obvious fact that it works in the two-person case. Mathematical justification for the n–person case will be left for a later, more theoretical paper. The need for a detailed derivation is accentuated by Atkinson's account of possible ordinal differences among inequality measures.

6 Festinger, Leon, “A Theory of Social Comparison Processes,” Human Relations, VII (May 1954), 117–40.CrossRefGoogle Scholar

7 See Schachter, Stanley S., The Psychology of Affiliation (Stanford 1959)Google Scholar and various articles in Latane, Bibb, ed., “Studies in Social Comparison,” special supplement to the Journal of Experimental Social Psychology, II (September 1966).Google Scholar

8 Runciman, W. G., Relative Deprivation and Social Justice (Berkeley 1966)Google Scholar, esp chap. x. Similar evidence from France is cited ibid., 26–27.

9 The maximum occurs when the first derivative of D, D', equals zero:

The second-order condition for a maximum is satisfied if a2 > 0 and a1 > 0; the latter must be given the Festinger postulate.

The level of discontent at the maximum point is computed by inserting g = 1/2a1 into Equation (5). This yields.

In Figure 1, all variables are assumed normalized so that diey vary between 0 and 1. a2 has been set at .9 and a1 at .8.

10 Mitchell (fn. 2).

11 Mitchell's four indicators were: the percentage of owner-operated land, the coefficient of variation of the distribution of land holdings, the percentage of land in formerly Vietnamese-owned estates subject to transfer under the Diem land reform program, and the percentage of land formerly French-owned and subject to transfer under the same program. All four measures have been criticized by Paranzino, Dennis, “Inequality and Insurgency in Vietnam: A Further Reanalysis,” World Politics, XXIV (July 1972), 565–78.CrossRefGoogle Scholar

12 Mitchell's data of Saigon control came from a map published by the Los Angeles Times, December 26, 1965, and reportedly obtained from U.S. Government sources.

13 Mitchell, (fn. 2), 437–38.Google Scholar

14 Paige, Jeffery M., “Inequality and Insurgency in Vietnam: A Reanalysis,” World Politics, XXIII (October 1970), 24–37CrossRefGoogle Scholar; Sansom, Robert L., The Economics of Insurgency in the Mekong Delta in Vietnam (Cambridge, Mass. 1970), 230–32Google Scholar; Paranzino (fn. 11); Russo (fn. 3).

15 Mitchell's R2 is .68, but Paranzino reports that the equation is incorrectly estimated and should yield an R2 of .55. Russo reports an R2 of .82, but my own replication of his regression yields an R2 of .83 and an R2 of .79. R2 is the conventional symbol for the coefficient of determination adjusted for degrees of freedom. For the formula relating the two quantities, see Goldberger, Arthur S., Econometric Theory (New York 1964), 217.Google Scholar

16 The correlations are weak, however. Average income and die Gini index correlate .19; the correlation between the Gini and die ratio of average landholdings to subsistence areas is only .05.

17 Russo, (fn. 3), 322.Google Scholar

18 Goldberger, (fn. 15), 213–18.Google Scholar

19 Russo (fn. 3), Table 1. Original sources are as follows: I: Stroup, Robert H., “Rural Income and Expenditure Sample Survey: Preliminary Report,” U.S. Agency for International Development, 1965Google Scholar; g and L/L8: “Report on the Agricultural Census of Vietnam,” Department of Rural Affairs, Republic of Vietnam, 1961; HH: “The Religions of Vietnam in Faith and Fact,” Southeast Asia Religious Project, Fleet Marine Force Pacific/IMAC. Dennis Paranzino generously supplied the data cards.

20 See fn. 15 above. Comparable results occurred when Mitchell's chief inequality measure, the coefficient of variation, replaced the Gini coefficient This finding may ease doubts raised by fn. 5.

21 Mitchell, (fn. 2), 433Google Scholar and Figures 1, 2, 3, and 4.

22 Specifically, the control value C' is calculated as follows:

The residual term e was computed for each province by inserting the actual values of all variables in die right side of Equation (6) and subtracting the result from die actual C value for the province.

23 Provinces are: 1, An Giang; 2, An Xuyen; 3, Ba Xuyen; 4, Bien Hoa; 5, Binh Dinh; 6, Binh Duong; 7, Binh Thuan; 8, Dinh Tuong; 9, Knanh Hoa; 10, Kian Giang; 11, Kien Hoa; 12, Kien Phong; 13, Kien Tuong; 14, Long An; 15, Long Khanh; 16, Ninh Thuan; 17, Phong Dinh; 18, Phuoc Tuy; 19, Phu Yen; 20, Quang Nam; 21, Quang Ngai; 22, Quang Tri; 23, Tay Ninh; 24, Thua Thien; 25, Vinh Binh; 26, Vinh Long.

24 The values for g and a1 are derived as follows: I assume Saigon control is a linear function of discontent and other variables:

where b1 and b2 are unknown parameters. From Equation (5):

(For the purpose at hand, it makes no difference whedier the non-Gini terms in Equation 6 affect C through D or directly.) To find g*, the level of g at which discontent is a maximum (and Saigon control is a minimum), we set the partial derivative of G with respect to g equal to zero:

From Equation (6) we take b2a2 = 467.5 and b2a1a2 = 415.9. Substituting in (9) and solving for g*, we obtain g* = .56. Inserting this value in (9), it is easy to compute a1 = .89. Unfortunately, it is not possible to solve for a2 without knowledge of the unknown b2.

25 Paranzino, (fn. 11), 574–77.Google Scholar

26 Goldberger, (fn. 15), 215.Google Scholar

27 See Johnston, J., Econometric Methods, 2nd ed. (New York 1972), 159–64Google Scholar; Feldstein, Martin S., “Multicollinearity and Mean Square Error of Alternative Estimators,” Harvard Institute of Economic Research Discussion Paper No. 142 (Cambridge, Mass. 1970)Google Scholar; and Raduchel, William J., “Multicollinearity Once Again,” unpub. paper (Harvard University 1971).Google Scholar I am indebted to Robert P. Inman for the latter two references and for advice on this question.

28 Sansom, (fn. 14), 232Google Scholar; emphasis in original.

29 The N.L.F. General Uprising campaign began July 20, 1960. Ibid., 236.

30 The Mu Gia Pass, heavily bombed by the U.S. until the 1973 treaty, is usually depicted as the starting point of the Ho Chi Minh Trail network—for instance, on a map in the New York Times edition of The Pentagon Papers (New York 1971), viii. Distances to province capitals were measured on a straight-line basis using a steel tape measure and the National Geographic Society map of Vietnam, Cambodia, Laos, and Thailand (Washington, February 1967), the scale of which is 30 miles to the inch. Province capitals are from the House Committee on Armed Services edition of the Pentagon Papers: US-Vietnam Relations, 1945–67 (Washington 1971), Book 2, IV.A.5, map on p. 3.

31 This finding indirectly helps refute the 1965 White Paper of the U.S. State Department, Aggression from the North, which purported to show the revolt in the South “inspired, directed, supplied, and controlled by the Communist regime in Hanoi.” On the White Paper, see Kahin, George McT. and Lewis, John W., The United States in Vietnam (New York 1967).Google Scholar

32 There does exist, however, an alternative substantive explanation for the curvilinear relation of land-tenure inequality and control by the Saigon government. It can be argued that discontent varies direcdy with economic inequality, while the power of the discontented varies inversely with the same factor. If one further assumes that revolutionary success varies as the product of the oppressed class's discontent and its power, then this theory also implies a relation like that depicted in Figure 1. (The argument is an extension of that presented by Gamson, William A., Power and Discontent [Homewood, Ill. 1968]Google Scholar chap. 7.) The data used in this paper do not permit a test of the two interpretations, but circumstances can be imagined in which they would make different predictions—for example, when the arrival of a revolutionary army reduces the normal power disparity between landlords and peasants.

33 In particular, Sansom's historical account of the Mekong Delta describes Davies' J-curve pattern in classic purity. A major J-curve developed between 1868 and 1945, culminating in die Vietminh revolution. After a long era of rapid expansion, die riceland fronder closed in 1930. As population continued to rise, the rice area cultivated per capita fell. As it did, “a rather starding decline in real incomes” occurred. A secondary J-pattern can be traced from 1945 to 1960. The peasants] lot improved in 1945–54, as the Vietminh reduced rents and interest, and redistributed land. But “during the interregnum of 1954–59, the social-institutional accomplishments of the Viet Minh were held in abeyance or (often) reversed.” In 1960 die National Liberation Front instituted the second phase of heroic and tragic struggle. Sansom (fn. 14), chaps. 2 and 3. On die J-curve dieory, see Davies, James C., “Toward a Theory of Revolution,” American Sociological Review, XXVII (February 1962), 5–18.CrossRefGoogle Scholar

34 Taylor, Charles L. and Hudson, Michael C., World Handbook of Political and Social Indicators, 2nd ed. (New Haven 1972).Google Scholar With two exceptions, the World Hand book data used in this study were taken from tapes supplied by the Inter-University Consortium for Political Research through the Social Science Data Center of the University of Pennsylvania. The exceptions are the totals of Deaths and Riots for 1948–67, which were punched directly from the Handbook. I am grateful to Barry Cohen of the Data Center for his assistance and to Dennis Paranzino, who supplied cards for several of the variables used.

35 Feierabend, , Feierabend, , and Gurr, (fn. 3), 213–15.Google Scholar The Gurr indices are available for only 51 of the countries for which the World Handbook provides Gini scores.

36 The latter four variables were computed from the World Handbook or its tapes (see fn. 34). 1960 population totals were employed.

37 These time periods were selected to avoid missing cases. The data were from the World Handbook tapes.

38 Gurr, Ted Robert, Why Men Rebel (Princeton 1970), 335Google Scholar; emphasis added.

39 For both Internal War and Turmoil, the best prediction is afforded by GNP growth rate alone; the respective correlations are —.37 and —.27. In words, economic growth is associated with reduced Internal War and reduced Turmoil. This accords with common sense, but not with the speculations of Olson, Mancur Jr, “Economic Growth as a Destabilizing Force,” Journal of Economic History, XXIII (December 1963), 529–52.CrossRefGoogle Scholar

40 Improvement in R2 is to be expected whenever one deletes a term whose coefficient has t < I. See Haitovsky, Yoel, “A Note on the Maximization of R2,” American Statistician, XXIII (February 1969), 20–21.Google Scholar

41 A country was classified as agricultural if the percentage of its male labor force in agriculture equalled or exceeded 50 in either 1960 or 1965, according to the World Handbook data. Two years were employed because many cases were missing on each, but only three were missing on both. For those diree—Mali, Kenya, and South Vietnam—I made educated guesses and included all in the agricultural group.

42 The single year was used in these tests because fewer cases were missing for 1965 than for 1960 and because substantial changes in this variable often occurred between the two dates. Missing cases reduced N to 40 for World Handbook dependent variables, and to 38 for Gurr's indices.

43 Russett (fn. 1), reports that inclusion of the percentage of the labor force in agriculture raised the predictive power of his regressions; but he evidendy incorporated the variable additively rather than multiplicatively. I see no theoretical rationale for the former course.

44 Simple correlations of the growth rate with instability variables were as follows: Internal War, —.53; Deaths 1948–67, —.52; Deaths 1961–65, –46.

45 Gurr (fn. 38), passim, esp. chap. 10.

46 One set of national RD scores is available—the data gathered between 1957 and 1962 by Hadley Cantril using his Self-Anchoring Scale. Unfortunately, he studied only 14 nations, and we have Gini scores for just 11 of these. I regressed this small sample against the Gini, its square, and GNP per capita, but obtained significant results only for GNP per capita, which correlates –.57 with average deprivation. (The Cantril data were taken from ibid., 65.)

47 Russett (fn. 1).

48 See, e.g., Brown, Lester R., “The Agricultural Revolution in Asia,” Foreign Affairs, XLVI (July 1968), 688–98CrossRefGoogle Scholar, or Wharton, Clifton R. Jr, “The Green Revolution: Cornucopia or Pandora's Box?” Foreign Affairs, XLVII (April 1969), 464–76.CrossRefGoogle Scholar

49 The seven countries are from a table in Brown, Lester, Seeds of Change: The Green Revolution and Development in the 1970's (New York 1970), 40.Google Scholar Ranked in order of increasing Gini indices (in parentheses), the countries are Taiwan (.46), India (.52), the Philippines (.53), Turkey (.59), Iran (.62), Pakistan (.65), and Mexico (.69). Brown also lists four nations for which I have no Gini indices: Ceylon, Afghanistan, Indonesia, and Morocco. Except for two pioneers, Taiwan and Mexico, all experienced (or were expected to experience) their “yield takeoffs” between 1965 and 1970. Increasing rebelliousness in India has been reported in the press and in Sharma, Hari P., “The Green Revolution in India: Prelude to a Red One?” in Gough, Kathleen and Sharma, Hari P., eds., Imperialism and Revolution in South Asia (New York 1973), 77–102.Google Scholar The Moslem and Communist insurgencies in the Philippines have been described in frequent New York Times dispatches during 1972–73.

50 Serano, Joseph A., S.J., , un tided paper (University of Pennsylvania, December 1972).Google Scholar

51 Davis, James A., “A Formal Interpretation of the Theory of Relative Deprivation,” Sociometry, XXII (December 1959), 280–96.CrossRefGoogle Scholar The seminal reference on status inconsistency is Lenski, Gerhard, “Status Crystallization: A Non-Vertical Dimension of Social Status,” American Sociological Review, XIX (August 1954), 405–13.CrossRefGoogle Scholar The literature on both concepts is mammoth.

52 The correlation of inequality and per capita GNP, however, is only —.08.

53 Sansom, (fn. 14), 234–35.Google Scholar