A comparative analysis of the social requisites of democracy Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres The failure of many newly independent nations to establish democratic rule after the Second World War challenged scholars to identify the necessary linchpins of democracies. The early efforts dating from the late 1950s and early 1960s emphasized the relationship between political systems and the socio-economic and structural conditions embodied in the processes of industrial growth and modernization, i.e. higher literacy, urbanization, Gross National Product (GNP), etc. in seeking to account for democratic development. A number of quantitative analyses in political science and sociology found economic development, a higher GNP per capita, and political democratization inextricably bound together (Lemer, 1958, pp. 43-107; Lipset, 1959, pp. 69-103; Lipset, 1960, pp. 27-64, 469-76). For democracy to become stable and entrenched, the procedures and norms of the new democratic structures, inherently low in legitimacy, must be effective, which largely meant decades of sustained economic advance. Prolonged effectiveness served to create a basis for legitimacy, a title-to-rule, which included acceptance of the rules of the democratic game, as a fixture of the political system (Lipset, 1960, pp. 64-80; Lipset, 1963, pp. 16-23, 245-47). With spectacular economic growth registered by many developing countries in the 1960s and 1970s, the expected democratic political Seymour Martin Lipset is the Hazel Professor of Public Policy at George Mason University, Fairfax, VA 22030-4444, USA, and Senior Fellow at the Hoover Institution at Stanford. He is President of the American Sociological Association for 1992-3. His most recent book is Continental Divide: The Values and Institutions of the United States and Canada. Kyoung-Ryung Seong received his PhD in Sociology from Stanford University and is an Assistant Professor at Hallym University in Chunchon, Korea. John Charles Torres is a doctoral candidate in the Department of Sociology at Stanford University. outcomes nevertheless failed to materialize for many. Some countries with outstanding economic growth records remained staunchly authoritarian, e.g. Brazil, Spain, the Republic of Korea, Taiwan, etc. A surge of literature appeared during this time which arrived at disparate conclusions about the connection between economic development and democracy. Invariably, these reflected variations in methodology, but some empirical results challenged the very relationships (Cnudde and Neubauer, 1969; Lijphart, 1972, pp. 4, 417-32; Huntington and Dominguez, 1975, pp. 59-66). The rise of dictatorships, especially in Latin America, led to arguments that economic growth is facilitated by authoritarianism, not democracy (Linz and Stepán, 1978; O'Donnell, 1973; Collier, 1979).l Analysts such as Huntington, O'Donnell, and the Latin American dependency theorists anticipated poor prospects for democracies. They claimed broadly that Western political and economic institutions had dysfunctional consequences for Third World economies. Huntington and O'Donnell postulated that the masses would be ungovernable under democracy because the social and welfare demands of the working class and the poor would impede economic advances. The dependency theorists noted the unequal relationship between industrialized core coun- 1SSJ 1361993 © UNESCO 1993 Published by Blackwell Pubtahere, 108 Cowley Read. Oxford OX4 1JF, UK and 238 Main Street. Cambridge. MA 02142. USA. 156 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres tries and the periphery; foreign investors rarely stimulated the local economies, yet extracted primary resources leaving little room for growth; peripheral nations were intentionally kept underdeveloped. As late as 1978, the distinguished sociologist Fernando Henrique Cardoso pointed to 'the resurgence of authoritarianism' in Latin America, noting that democracy is 'more of an exotic plant than a model towards which society should be said to be moving'. He asked 'what meaning can we attribute to the idea that a whole region is being modernized and at the same time becoming politically more authoritarian?'2 In any case, by the mid-1970s, a new cycle emerged, with 'young' democracies arising first in southern Europe, then in east Asia, and almost all of Latin America (Herz, 1982; O'Donnell et al., 1987; Malloy and Seligson, 1987; Diamond et al., 1989). Notably, these transitions occurred in the high growth 'bureaucratic authoritarian' nations.3 With these more recent developments, the development paradigm proposed in the 1950s deserves careful scrutiny, through a re-testing of fresh data drawn from the 1970s and 1980s. Such investigations, reported here, focus upon the crucial relationships of economic and social variables with national political values and structures. This emphasis may also shed light on whether the recent move to democracy is part of a pattern of pendulum swings between democracy and authoritarianism during different historical eras.4 The developments in eastern Europe and the former Soviet Union will not be dealt with here since to do so would require introducing factors which have not been present elsewhere. Modelling economic growth and democratization Political development and economic performance - what is the relationship?5 The issues raised depend on the causal direction posited between the two. Does economic growth advance democratic causes? Are democratic polities optimum for positive economic yields (Haggard, 1990)? Moreover, if the connection between democracy and economic wealth (or any other socio-economic condition or requisite) is not deemed important, what accounts for the correlations (Weede, 1983; Karl, 1990)? First, there is much support for the hypothesis Lerner and Lipset proposed in the late 1950s that 'the more well-to-do a nation, the greater the chances that it will sustain democracy' (Lipset, 1960, p. 31; see also Lerner, 1958, p. 63). This positive correlation between economic wealth and democratization has been empirically tested and confirmed by many researchers (Cutright, 1963; Needier, 1968; Neubauer, 1967; Olsen, 1968; Banks, 1970; Winham, 1970; Flanigan and Fogelman, 1971; Kim, 1971; Flora, 1973; Marquette, 1974; Coulter, 1975; Bollen, 1979,1980). A comprehensive effort to locate the correlates of democracy for over 100 countries, using political indicators for 1960 and 1965, by Bollen and Jackman, found 'that level of economic development has a pronounced effect on political democracy, even when other non economic factors are considered'. The latter include: an index of ethno-linguistic fractionalization, proportion of Protestants, former British colony or not, and new nationhood (former colony or not). GNP per capita remains 'the dominant explanatory variable', not only for all countries, but for the non-core, i.e. outside the industrialized world, also (Bollen and Jackman, 1985, pp. 38-9). The link between development and democratization needs to be looked at in comparative quantitative terms, as do socio-political processes in individual nation states, especially when economic and social indicators now available are more reliable and cover a wider range of nation states.6 More importantly, the relationship can be appreciated over long periods. As Howard Wiarda argues, 'The develop-mentalist approach looks far better in retrospect and from a long-term perspective than it looked in the late 1960s and 1970s' (1990, p. 80). This article does not assume that economic development alone produces democratization. As is to be noted in more detail later, national idiosyncracies, the play of historical, cultural and political forces, and the behaviour of leaders may advance or prevent democratization in any particular nation state or group of them, generally or at any one point in time. A crisis-ridden, economically declining, authoritarian system may break down as in the cases of Argentina, © UNESCO 1993. Comparative analysis of social requisites of democracy 157 Table 1. Political regime type, GNP per capita, and growth rates by economic development % semi- 1982 median 1960-82 median % authoritarian democratic % democratic GNP per capita GNP per capita regime regime regime ($) ($) Low income economies" N = 32 Low income economies6 N = 34 Lower-middle income economies N = 27 Upper-middle income economies N= 33 Industrial market economies iV = 19 75 74 37 12 0 25 24 51 48 0 11 39 100 230 260 580 1910 10930 1.1 1.1 2.1 3.4 3.1 " Excluding India and China h Including India and China Source: The evaluations of political regimes come from Raymond D. Gastil, Freedom m the World Political Rights and Civil Liberties, 1985. Westport: Greenwood Press, 1985. The economic data are from the World Bank, World Development Report, 1984. New York: Oxford University Press, 1984. Haiti, Pakistan, the Philippines, the Soviet Union and Eastern Europe generally. In one of the most comprehensive and provocative recent approaches to this subject, Huntington, after noting in 1984 that 'the correlation between wealth and democracy is thus fairly strong', surprisingly concluded that 'with a few exceptions, the limits of democratic development in the world may well have been reached' (Huntington, 1984, pp. 199,218). Yet, if there really is the correlation between wealth and democracy that he emphasized, and if a large number of states are steadily experiencing economic growth as Table 1 shows them to be doing, the expectation should have been for the growth of political liberty.7 The available data on economic levels and democratization suggest, as Table 1 indicates, that the correlation between them is more pronounced in the early 1980s than in the late 1950s. The political data come from the annual report on the state of freedom around the world by Freedom House. States are categorized as 'not free' (authoritarian), 'partly free' (semi-democratic), and 'free' (democratic) on the basis of political rights and civil liberties, whether citizens can turn out incumbent governments through free elections and whether they can organize political parties and express critical views without government interference (Gastil, 1985, p. 3). An indication of the reliability of this index may be found in Sirowy and Inkeles' report that the intercorrelations among ten measures of democracy, including Gastil's measure, were on the average of r = 0.75 and reached as high as r = 0.90. They conclude, 'This level of intercorrelation suggests that a diverse set of measurement efforts have yielded a fair amount of agreement in discriminating systems with respect to political democracy' (Sirowy and Inkeles, 1990, p. 153). The independent variables and GNP income categories are taken from the World Development Report of the World Bank.8 The overall relationship shows up in Table 1. Among the lowest per capita income nations, only India is 'free' (democratic), while the industrial market economies are all democratic.9 The proportion classified as authoritarian declines from category to category as levels of national income rise. Going from the 'lower middle' to the 'upper middle' income classifications, the © UNESCO 1993 158 Seymour Martin proportion of 'not free' (authoritarian) states declines, and that of free (democratic) states increases. These observations are supported by all the upcoming statistical analyses. The data indicate that the prospects of economic wealth and, therefore, the potential for democracy, remain bleak in the least developed nations. In most countries in the lower-income economies the national income is not increasing rapidly and they are also not 'free'. When India and China are subtracted from the statistics on 'low-income economies', the remaining nations have a growth rate of only 1.1 per cent per year from 1960 to 1982, whereas those in the three higher income categories of the World Bank classification grew much more quickly. The average annual growth in GNP for eleven of the lower-income nations decreased during this period.10 These socio-economic indicators suggest the likelihood of countries becoming democratic. For example, 'redemocratization' is exactly what one would expect of Spain, which is classified as an industrial market economy with a GNP per capita of $5430 (1982 dollars). In the early 1970s Spain was the only one of the 19 industrial market economies reported not to operate as a democracy. It had become the ninth strongest industrial power in the world, with well over half the population in the middle class and only a fifth in agriculture. As one observer noted: 'By the time Franco died in late 1975, Spain had become a modem country in every respect but the political' (Roskin, 1978, p. 630). To reiterate, the socioeconomic correlations point to probabilities. Other factors, such as the force of historical incidents in domestic politics, cultural factors, events in neighbouring countries, diffusion effects from elsewhere, leadership and movement behaviour, can affect the nature of the polity. Thus, the outcome of the Spanish Civil War, determined in part by the behaviour of other European states, placed Spain in an authoritarian mould, much as the allocation of Eastern Europe to the Soviet Union after the Second World War determined the political future of that area. Karl argues that the search for democracies' prerequisites is misguided. She accounts for democratic transitions through observations of the pact-making processes by political regimes (, Kyoung-Ryung Seong and John Charles Torres and parties. Karl contends, 'Rather than engage in a futile search for new preconditions, it is important to clarify how the mode of regime transition (itself conditioned by the breakdown of authoritarian rule) sets the context within which strategic interactions can take place because these interactions, in turn, help to determine whether political democracy will emerge and survive . . .' (emphasis in original, Karl, 1990, p. 19). Karl and others see analysis of elite behaviour in constructing pacts as mutually exclusive from the study of democratic requisites. Political pacts occur in different social contexts, and it is not possible to determine the conditions in society that make these pacts possible. Social requisites lay the foundations for successful democratic consolidation. Pacts are means toward institutionalizing democracy, whether they emerge or hold is linked to probabilities associated with the presence or absence of these requisites. It is not necessary to make an 'either-or' choice between the study of democratic conditions and pact-building as they are complementary. The statistics behind Table 1 are reflected in major regional and cultural differences. The freedom scores of African and Moslem countries are low, while the transition to democracy has been especially notable in Latin America. A democratic 'contagion' effect seems to have operated in Latin America in the late 1970s and 1980s, as free electoral institutions were adopted across the area. By 1985, President Reagan's special assistant for Latin America was able to boast that 'today twenty-six of the thirty-three countries south of our border, containing 90 percent of the population of what we call Latin America, are now democratic or in a genuine process of transition toward democracy' (Menges, 1985, p. 39). These patterns support the hypothesized relationship between economic development and democratization. The oil-rich nations apart, per capita income in Latin America ranks between the OECD nations and the rest of the Third World, far above Africa and the Muslim states. Haiti is the only Latin American state among the 'low-income economies' in Table 1; the other 33 countries are Asian or African. The Latin American countries with the lowest per capita GNP categorized as 'free' in 'lower- © UNESCO 1993 Comparative analysis of social requisites of democracy 159 middle economies' are Bolivia and Honduras, while nine out of fourteen 'free' (democratic) countries among the 'upper-middle economies' are in Latin America. In support of these illustrative analyses, our report will be composed of two sections: (1) a multivariate regression to evaluate the relationship between growth and democracy, (2) a look at the cross-sectional pattern of N-Curve relations between democracy and growth in order to identify the true pattern of the relationship. Section one: multivariate regression Using a cross-national dataset, a multivariate regression analysis was applied to panel data. This section relates economic development to political performance.11 The cases used, in this cross-sectional analysis, are limited to those nations that had experienced colonial rule so that the effect of a specific colonial heritage on subsequent regimes form could be specified.12 Dummy variables were constructed for British and French colonial rule and various control variables are also included and defined. Regression results starting from 1960 through 1985 are presented in Table 2. The panels used are for five-year intervals, excluding the 1970 panel.13 The following are a series of hypotheses that are put to the test in this analysis. (1) First, economic development should show a positive effect on the level of democracy - the original 1959 Lipset hypothesis. We next consider four other hypotheses related to the control variables in our regression: (2) in creating dummy variables for former British and French colonies, we hypothesize that British past rule is associated with democratization, while a French past has a negative effect. The details and reasons for these assumptions are discussed below; (3) regime coerciveness should have a negative influence; (4) political mobilization should also reduce the chances for democracy; (5) trade dependence has been hypothesized in contradictory ways - increasing incorporation of developing nations into the global system with economic, technological, and cultural diffusion aids in political development, but conversely some hold that increasing incorporation in the world economy creates unequal 'dependent' ties between the core and the periphery - the 'dependency' argument - and reduces democratic prospects. The measures used for this regression are summarized at the end of Table 2, yet we will now outline them briefly. Economic development (ECON) represents a measure of GNP per capita.14 As noted, dummy variables (BR and FR) were created to code former British and French colonies. Regime coerciveness (MILEXP) was measured by amount of military expenditure as a percentage of GNP. However, for the 1960 panel, regime coerciveness was measured by (INTSEC), the percentage of the working population that served as internal security force members. The inconsistency in this particular measurement arises from data limitations. Trade dependence, (WSTRAD), represents total trade, the sum of exports and imports as a percentage of GNP. Political mobilization (RESIST) accounts for the yearly sum of protests, riots and strikes. Our dependent variable involves two different measures. For the 1960 and 1965 panels, the Bollen Index of Democracy was used. This index is a composite of six indicators. The first three indicators gauge popular sovereignty: (1) fairness of elections; (2) method of executive selection and (3) method of legislative selection. The last three indicators measure political liberties; (4) press freedom; (5) freedom of group opposition and (6) government sanctions. A 100-point scale combines these scores where one is the least democratic and 100 is the most. For all the later panels, the thirteen-point Freedom House scale, not available for years before 1973, measuring a composite score of political and civil liberties is utilized. Since the dependent variables are different, cross-time comparisons involving panels with different dependent variables are not appropriate, yet by using Bollen's index for 1960 and 1965, we may make comparisons and relate our results against Bollen and Jackman's findings for the 1960s (1985, pp. 27-48). Our findings test the assumptions from the 1970s and through the mid-1980s. The goodness-of-fit is very high, i.e. see the R square and the adjusted R square for the panels in Table 2.15 The most significant finding of Bollen and Jackman for the 1960s and ours for the 1970s and 1980s is that economic development is the single most important predictor of © UNESCO 1993 160 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres Table 2. Unstandardized regression coefficients for political development on selected independent variables: cross-national data in five-year panels. Independent Dependent variables BOLLEN 60 BOLLEN 65 FREE 75 FREE 80 FREE 85 INTSEC -0.353° (0.216) — — — — RESIST -0.253 (0.253) -0.267 (-0.525) -0.005 (0.059) 0.015 (0.012) 0.008 (0.011) ECON 11.240» (3.887) 15.097» (2.256) 1.794» (0.271) 1.205» (0.292) 1.872» (0.279) BR 20.915" (8.869) 11.694" (5.498) 1.315" (0.698) 0.816 (0.796) 0.441 (0.773) FR -5.558 (15.426) -10.122" (5.489) 1.303" (0.784) -2.175" (0.910) -2.585» (0.841) WSTRAD 9.445 (15.105) 3.830 (8.363) 0.757° (0.594) -0.281 (0.600) -0.978" (0.590) MILEXP — -2.948» (1.004) -0.180» (0.070) -0.190" (0.098) -0.163" (0.091) CONSTANT -5.149 (24.405) -30.558" (14.603) -4.117" (1.795) -0.297 (2.051) -4.048" (1.937) Adjusted Rz 0.423 0.538 0.422 0.291 0.462 N 22 66 86 85 93 ° p < 0.05 * p < 0.001 » p < 0.001 Numbers in parentheses are standard errors. ECON represents log (GNP per capita), BR and FR are former British and French colonies dummy constructs, MILEXP = military expenditure as percentage of GNP, INTSEC = the percentage of the working population that served as internal security force members, WSTRAD = sum of exports and imports as percentage of GNP, RESIST = yearly sum of protests, riots, and strikes, BOLLEN = Bollen index of democracy on a 100-point scale, and FREE = Gastil and Freedom House 13-point scale of political and civil liberties. Independent and dependent variables are coded from the same year with two exceptions. The 1960 panel includes independent variables that are lagged. MILEXP was lagged in the last two panels by two years due to data availability. political democracy when controlling for other variables - strong support for the generalizaitons drawn from the 1950s data. We also found, reiterating Bollen and Jackman, that former British colonies are more likely to have political democracy through the 1970s than countries that have been ruled by other colonial powers. The relationships, however, are not significant for 1980 and 1985. This fall-off does not appear to reflect less democracy on average in these states, so much as growth of political liberty in nations which had been ruled by other powers. Our analysis indicates, across all panels, that former French colonies are more likely to be non-democratic than the British. Others which are largely authoritarian include the smaller number of nations once ruled by the Portuguese, Dutch and Belgians. French rule has had a negative effect on political democracy, after controlling for the other variables. The negative correlation between French background and democracy is, in fact, more substantial than the positive one between British colonial rule and democracy. The regression results also point to regime coerciveness, as reflected in military expenditures, as inversely related to political democracy. Stepán argues that in most Third World countries, the military are concerned with the maintenance of internal order (control of domestic conflict and instability). He calls this the 'new professionalism' as distinct from the 'old © UNESCO 1993 Comparative analysis of social requisites of democracy 161 Democracy Economic development Figure A. Positive linear relationship, as suggested by Lipset, in a single time point using cross-sectional data. professionalism' - the maintaining of national security and defence where the goal has been to deter external threats, not to defend civil order (Stepán, 1973). Ironically, therefore, in Third World countries, spending heavily on defence is associated with political instability, not civil order. Political mobilization does not produce significant consequences for political democracy; however, trade dependence shows slight negative effects on democratization in the 1975 and 1985 panels - the p-value registers at 0.103 which makes it barely significant at the 0.10 level and, for 1985, the p-value is significant at 0.051. Overall, the effects of trade dependence on democratization produced inconsistent results. Section two: cross-sectional pattern of /V-curve relations Lipset, in 1959, described a linear relationship between economic growth and political democracy, as shown in Figure A (Lipset, 1959, pp. 75-85; Lipset, 1960, p. 31). Jackman's empirical findings qualified the Lipset hypothesis by indicating that the shape of the relationship is curvilinear (1963, pp. 611-21). However, if the curve is log-transformed, the relationship between economic growth and political democracy becomes linear. The pre-log transformed Jackman model appears in Figure B. Although Jackman's model works well in general, he does not consider, and ignores, the wide potential variations in segment (b) of Figure B. If an upper threshold is incorporated © UNESCO 1993. Democracy ] (a) >f(b) / ! (c) Economic development Figure B. Jackman argued a curvilinear relationship between democracy and economic growth. The approach is also used with cross-sectional data using a single time point. into the modelling, a variation in (b) can be clearly observed. O'Donnell, having a more pessimistic forecast for democratization, particularly in nations with 'bureaucratic authoritarian' or (BA) regimes, represents one of the possible variations in segment (b) of Jackman's model. O'Donnell suggests that late industrializers benefit from bureaucratic authoritarian rule. This form of rule deepens import substitution from consumer goods, intermediate goods to capital ones. B A regimes, typically an alliance of the military and technocrats, impose economic austerity programmes, constrict consumption, repress labour, and provide the political stability needed to attract foreign and long-run investments. O'Donnell contends that BA regimes give developing countries the needed advantage for quick industrialization.16 O'Donnell argues succinctly, 'More socioeconomic development equals more political Democracy Economic development Figure C. O'Donnell also argues for a curvilinear relationship, yet after an initial positive correlation between democracy and economic development, a long-run negative relationship ensues. This argument is cross-temporal for individual countries. 162 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres pluralization does not equal more likelihood of political democracy . . . political authoritarianism - not political democracy - is the more likely concomitant of the highest levels of modernization' (1973, p. 8). The curve in Figure C maps O'Donnell's argument. According to O'Donnell's statement, the curve should fail to climb after the capital goods phase has been entered. Despite these differences, the level of analysis utilized by O'Donnell is clearly different; Figure C applies to case studies or single countries over an extended period. The argument is not cross-sectional, but cross-temporal. Kurth suggests an N-lype relationship between phases of industrialization and form of political regimes. Using historical examples from Europe and Latin America, he identifies three distinct phases: the simple non-durable consumer goods one involving the production, for example, of textiles, shoes, and household items. Next is the capital goods phase, for example, steel products such as rails, locomotives, merchant vessels, and machinery. The last period, or the durable consumer goods phase, is characterized by the production of products such as automobiles and appliances. Pre-industrial societies, Kurth notes, were typically ruled by absolute monarchies in Europe and oligarchic landed elites in Latin America. The simple consumer non-durables period ushered in the European liberal democracies and populist politics in Latin America. As nations moved into capital goods industrialization, authoritarian regimes assumed power -fascist-corporatist governments in Europe and bureaucratic authoritarian ones in Latin America. After the Second World War, as European nations redemocratized, they also entered the durable consumer phase. Latin America may be following this pattern as it increases production of hard consumer durables. Kurth's arguments are summarized by Figure D (1979, pp. 319-62). We suggest a modification of Kurth to indicate that an JV-type relationship between economic growth and political democracy exists cross-sectionally (see Figure E). In our model, democracy is not strictly linked to phases of industrialization. There need only be an upper threshold of economic development, beyond which an increase does not produce higher prob- Democracy ] ŕ ^v BA regime Agriculture Capital goods Phases of industrialization Non-durable Durable consumer goods consumer goods Figure D. Kurth contended that an Af-type relationship exists between political regime type and the phases of industrialization. Note that this argument is also about a 'genetic developmental course' of a single country using cross-temporal data. Democracy Economic development Figure E. This article proposes the relationship shown in this figure. We find an JV-Curve relationship cross-sectionally at single time points or panels. Moreover, there is an upper threshold of economic development where an increasing level of economic development does not affect a higher level of political democracy. Beyond the threshold, political democracy is stabilized. ability for political democracy. In other words, above a certain level of economic development (say GNP per capita of $5000), the chances for political democracy are highly stabilized. This relation, if confirmed empirically, clarifies the linearity assumption of the original Lerner-Lipset model (see Figure A) showing that it occurs over extended time. This new hypothesized relation between economic development and political democracy indicates a long run positive correlation, despite the possibility of negative relationships at intermediate ranges (i.e., the range of GNP per capita between $1500 and $3500). The model equation hypothesized is: Y = aX3 - bX2 + cX + d where X = GNP per capita and Y = the political freedom score, the 13-point scale from Gastil © UNESCO 1993. Comparative analysis of social requisites of democracy 163 Table 3. Unstandardized regression coefficients for political development on quadratic and cubic independent variables - the N-cur/e relationship: cross-national panel data. Independent variables (1) FREE 73 Dependent variables (2) (3) FREE 75 FREE 80 (4) FREE 85 ECON 0.009- 0.006' 0.009- 0.008- (0.003) (0.001) (0.002) (0.002) ECON2 -4.44E-06* -1.75E-06- -3.55E-06- -2.97E-06» (2.25E-06) (6.16E-07) (1.25E-06) (1.24E-06) ECON3 7.502E-10" 1.79E-10- 4.39E-10* 3.72E-10" (4.156E-10) (7.05E-11) (2.01E-10) (1.91E-10) CONSTANT 3.405- 3.675- 1.871" 2.256* (0.905) (0.665) (0.837) (0.859) Adjusted R2 N 0.266 100 0.331 104 0.363 89 0.336 91 • p < 0.05 * p < 0.01 - p < 0.001 Numbers in parentheses are standard errors. ECON represents log (GNP per capita), FREE liberties. Gastil and Freedom House 13-point scale of political and civil Independent and dependent variables are coded from the same year. For Model (1), $4000 represents the upper threshold. For Model (2), the upper threshold equals $7000. Model (3) and (4) have upper thresholds of $5000. These are illustrated in Figures F, G, H, and I. and the Freedom House. Testing it, the following nations are excluded from the sample (a) rich oil-exporting countries, i.e. Oman, Libya, Saudi Arabia, Kuwait, and the United Arab Emirates, (b) communist countries and (c) advanced industrial market economies because they are above the upper economic threshold.17 The results for all panels, presented in Table 3 and illustrated in Figures F-I show that a statistically significant TV-Curve relationship exists. For the 1973 panel (Table 3, column 1), an N-Curve relation occurs for countries with a GNP per capita under $4,000. This is shown graphically in Figure F. Variables A3, A2, and GNPPC70 are statistically significant and the FREE 73 mean: 13.73 (AT =15) GNPPC75 $2730 $3775 $7000 Figure G. hypothesized nature and direction of the relationship (positive or negative) are confirmed. The N-Curve patterns for the 1980 and 1985 panels (Table 3, columns 3 and 4) are significant under the threshold of $5000 as illus- FREE73 $1780 $2168 $4000 mean: 13.75 (N= 16) GNPPC70 FREE 80 $2034 $3350 $5000 mean: 12.46 (N=26) GNPPC75 Figure F. Figure H. © UNESCO 1993 164 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres trated by the corresponding Figures H and I. The goodnesses-of-fit for all the N-Curve models are impressive. Caveats The N-Curve model obviously does not suggest that crossing any economic threshold guarantees democracy. Our analysis merely attempts to trace and model the relationship between wealth and democracy through the 1980s. Clearly, the latest wave of democratization in Latin America is not secure. Quite the contrary. The pattern of democratization there has been cyclical: falling in the mid-1950s, moving up in the 1960s, then declining until the last years of the 1970s, improving again in the 1980s. Given continued poverty and extreme inequality, the democratic structures in the region remain vulnerable. The outlook for Haiti, Bolivia, Honduras, and El Salvador, all low GNP states, is not good. The first was authoritarian until 1986. The second has changed its chief executive, on average, once a year for over a century. The weakness and fragility of their economies would appear to be the prime reasons for their instability. Similarly, El Salvador and Honduras, partially 'free' states, need external support to avoid regression into authoritarianism. At the other extreme, however, relatively well-to-do countries like Argentina and Colombia also appear unstable, the first because of continued economic crises, the second because of tensions related to the role of drugs in the economy. Economic and political failures may also undermine authoritarian regimes. The level of income dropped precipitately during each of the three years before the Uruguayan military turned power back to civilians in 1985.18 In Argentina, a military defeat and economic malaise led to widespread revulsion against the FREE 85 $2346 $2979 $5000 mean: 12.92 (AT =26) GNPPC75 Figure I. Mean of freedom score reflect cases above the upper thresholds. military regime and to re-establishment of democratic norms in a country whose experiences in the 1960s and 1970s gave rise to the theory of 'bureaucratic authoritarianism'.19 Economic failures have also encouraged shifts in political structures of several other nations, such as the Philippines and the Soviet Union. The health and economic welfare of the Filipino population declined in the last years of the Marcos dictatorship. Glasnost in the Soviet Union is a reaction to the social crises and alienation produced by a long period of economic 'stagnation'. The examples of authoritarian breakdown following economic failures do not necessarily negate our emphasis on the positive relationship of growth and democratization. Though an immediate impetus to democratization may be economic decline, a nation's overall economic and social level establishes the pattern of expectations, interrelationships and pressures that makes democracy appear more attractive than authoritarian rule. As in Brazil and Argentina, the long-term strengthening of relevant economic and social forces under military rule makes democracy feasible while, at the same time, economic downturns or political failures may undermine the entrenched dictatorships.20 The economic approach to political systems is, of course, not limited to the correlations with development. In recent years, a number of comparative analysts and world system theorists have suggested that economic dependency has a negative effect on less developed countries' chances for political democracy, in particular by increasing inequality (Kaufman et al., 1975; Higgott, 1983). The available evidence is ambiguous, but in the main support for the hypothesis is weak. Bollen, using a variety of dependency measures, found negative correlations with estimates of democracy for 1965. However he is still uncertain about the relationship because he was unable to estimate the size of the indirect effects of dependency on political democracy through economic development. He reports, 'These indirect influences [between world system position or dependency and democracy] may be just as (or more) important than the direct effects analyzed in this analysis' (Bollen, 1983, p. 478). Our regression analysis also tested the hypothesis for more recent years. Our measure of dependency (amount of export ©UNESCO 1993 Comparative analysis of social requisites of democracy 165 and import as percentage of world trade), a crude proxy for the concept, does not show consistent statistically significant relationships.21 The emphasis on the politically dysfunctional effects of dependency is also confounded by the finding that lesser population size seems to be associated with free political institutions in Third World countries. The relationship is analytically important, since, as Simon Kuznets and Robert Triffin have noted, small states are inherently much more economically dependent on those they trade with than large ones (Kuznets, 1963, p. 16; Triffin, 1963, p. 248). A large proportion of the nations with three million or less population are democratic, e.g. Belize, Botswana, Costa Rica, the Bahamas, Mauritius, Gambia, Trinidad, Jamaica, Fiji, Papua New Guinea.22 The first post-colonial western state, Iceland, which has a tiny population, was easily able to institutionalize representative government long before Europe (Tomasson, 1980). Limited population apparently reduces the chances for potentially repressive conflict. The empirical relationship between smallness and democracy however, may be spurious as Myron Weiner argues, noting that the small states are disproportionately former British colonies (Weiner, 1987, pp. 18-22). The thesis that economic development is crucial for democracy has been challenged by Dankwart Rustow, who points to historical evidence that democracies existed in now-wealthy countries when they were at relatively low levels of economic development, e.g. the United States in 1820, France in 1870, and Sweden in 1890 (1970, p. 352). These and other early democracies, however, had the historical advantage of having formed their political institutions prior to the emergence of a worldwide communications system which might make it apparent that other countries were much wealthier than they (and few were), and before the appearance of electorally significant popular movements that demanded more equal distribution of worldly goods. A condition for a stable polity is a level of popular expectations appropriate to the economic level of the society, a pattern that characterized these early democracies. The less developed countries of today, however, draw their expectations from more affluent nations, and are culturally dependent on them (Huntington, 1968, p. 46; Lipset, 1963, pp. 15-60). Nor did the nineteenth century democratic polities face the overlapping political crises of contemporary less developed polities, which need to gain legitimacy: in the context of simultaneous mass demands for participation and distribution that has led to the collapse of representative institutions in one developing nation after another and the increasing conviction among their political elites that what power exists in the system must be tightly concentrated if the multiple crises it faces are to be resolved. Developing nations of the post-WWII era are thus disadvantaged, because they are denied the luxury of the slow, deliberate, sequential unfolding of these crises that many European countries experienced. (Diamond, 1980, p. 122; see also LaPalombara and Weiner, 1966, pp. 14-18; Huntington, 1968, p. 46) As noted by Lipset and reiterated by Diamond, There is an increasingly inescapable link in the modern age between socioeconomic development and belief in the legitimacy and effectiveness of a regime. With every decade, international demonstration effects from the wealthier nations become more compelling and pervasive and popular expectations of government performance further escalate. Only the most autarchic of nations could build, even in principle, a democracy on a foundation of persistent poverty, and in a rapidly shrinking world such nations have all but disappeared. It is thus difficult to imagine how any democracy in today's world can survive without at least some visible progress toward modernization. (Diamond, 1980, p. 106; see also Lipset, 1960, pp. 27-86) Patterns and variables behind the relationship Why the apparent relationship between levels of income and the likelihood of a nation adopting democratic structures? Broadly speaking, behind the variation in income levels reported in Table 1 lie corresponding national differences in class structures and degrees of inequality as reflected in the provision of health care and general quality of life. In other regression analyses using the same data bank, a 'powerful predictor of political and civil liberties' is the physical quality of life, as measured by infant mortality, life expectancy at age one, and adult literacy. "This finding, particularly striking among the less developed countries of the world, implies that economic development improves the prospects for democracy as it yields broad improvements in popular well-being' (Diamond et al., © UNESCO 1993 166 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres Jf^* "*T"fiS5P V' *i ^J&i Ä ^,'5 ' '~**?~ Colonial rule ceremony, Conakry, 1921. There are correlations between type of colonial rule and the functioning of post-independence democratic institutions. Sygma 1987, p. 10). With a significant rise in national income, consumption becomes more equal, the middle class grows, more people have access to health care, the level of illiteracy drops, and more teenagers stay in secondary school. This occurs whether the society is Islamic or atheist, Marxist or market-oriented, African, Asian, or European. (Again, the one major exception is the oil-rich states, where a sharp increase in per capita income does not necessarily lead to these consequences). People with more income, in complex and widely interdependent work situations, with more education, and more access to health and other services are more likely to ask for increased political freedom. The data in Table 1 sugest that, as the level of per capita income grows in a nation, economic producers increasingly desire and are able to support a system in which they have more influence on political leaders. Where incomes are low, economic interconnections among citizens are not complex. Many survive on their own, growing their own food, tilling the land of others, producing for a local market. Rising per capita income and the presence of technical innovations point up the emergence of more complicated economic relationships among citizens, and the appearance of new norms and values. As incomes rise, citizen demands become more visible, as does a pattern of politics which can accommodate them. In Eastern Europe, the democratic revolutions were preceded by state efforts to deal with economic stagnation by opening the system to more personal initiat- © UNESCO 1993. Comparative analysis of social requisites of democracy 167 ives, clearly the introduction of political liberalization. The decentralization of economic authority may demonstrate over time that societal decisions are often better when the input of citizens plays a genuine role. In a review of the research findings in many cross-national studies of attitudes and values, Inkeles and Diamond present considerable evidence to sustain the hypothesis that the level of a country's economic development independently affects the orientations conducive to democracy of its citizens. They report consistent relationships between the per capita GNP of nations and such characteristics as personal satisfaction, personal efficacy, anti-authoritarianism, and trust. The median rank-order correlations (within socio-economic groups) between per capita GNP and these traits are: 0.76 for anit-authoritarianism, 0.85 for trust; 0.55 for efficacy; and 0.60 for satisfaction (Inkeles and Diamond, 1980, pp. 73-109). Thus, they state, 'over a wide range of studies, with diverse samples and all manner of composition of the sets of countries represented, there is a clear, consistent tendency for authoritarianism and its cognate manifestations to be inversely related to level of national development' (1980, p. 83). The state of the economy exerts a substantial independent influence in shaping the attitudes and values of its citizens, and this effect is generally consistent at all levels of the standard domestic socioeconomic hierarchies such as those based on education or occupation. The direction of the effect is generally what most observers would rate as "positive". Living in a country that is more highly developed seems to be ego-enhancing it gives individuals a greater sense of personal worth, satisfaction and competence beyond what would be predicted from knowing only their education and occupation. In addition, the more economically advanced the country, the more individuals otherwise alike in status seem to develop qualities that contribute to stable politics and effective economic behavior because such individuals are more trustful and tolerant of others yet are more confident of their capacities. (Inkeles and Diamond, 1980, pp. 103-4) One of the greatest impacts of higher income is the possibility of further education. There is a striking relationship between income levels and the proportion of the relevant age group in secondary school: the median secondary school attendance was only 18 per cent of the age group in 1985 in the low-income Table 4. Changes in school enrolments, 1965-85 Median enrolment Median enrolment in secondary school as % of in the % of tertiary as age group age group 1965 1985 1965 1985 Low income 3.5 18 9 1 economies Lower-middle 8.5 29 1 7 income economies Upper-middle 25 55 4 16 income economies Industrial 62 96 13 30 market economies Source: The World Bank, World Bank Development Report, 1990 (New York: Oxford University Press, 1990). economies (including China and India), while median attendance was 96 per cent in the rich industrial market economies. The correlation between proportion enrolled in higher education and political freedom scores reported is unquestionably significant. The increase in schooling, plus rising national income, may have a dramatic effect on the attitudes and the political demands of the newly educated generation. As Inkeles and Smith conclude on the basis of interviews with samples of 6000 men in six developing countries, 'In large-scale complex societies no attribute of the person predicts his attitudes, values, and behaviour more consistently or more powerfully than the amount of schooling he has received' (1974, p. 133). Inkeles and Smith note that advanced schooling makes people's attitudes and values more 'modem' in a specific series of ways and these same 'modern' attitudes make citizens feel that they can and should have some say in the actions of their governments.23 Table 4 makes clear how much the availability of secondary and university education varies with the levels of national income. Among the low-income countries, some nations provide far more secondary education than the average for their class, such as India, China, or Sri Lanka, and some provide less, such as Uganda or Haiti. An extensive level of secondary education can exist with different political systems; it may reinforce a 'free' society (India), a 'par- © UNESCO 1993 168 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres tially free' society (Sri Lanka), or undermine a 'not free' society (China). But, where the distribution of education is as low as in Uganda or Haiti, the provision of modern, participant norms for citizens remains extremely difficult. Table 4 also demonstrates that the proportion of school-age people in secondary school rose dramatically in many nations between 1965 and 1985: from 35 to 63 per cent in Sri Lanka (low-income economy), from 41 to 65 per cent in the Philippines (lower-middle), from 28 to 70 per cent in Argentina (upper-middle), from 35 to 94 per cent in the Republic of Korea (upper middle), and from 38 to 91 per cent in Spain (industrial market). The sharp rise in the proportion of secondary school students around the world in the last decades of the twentieth century may have had a special impact on citizens' demands and so indirectly on political structures as well. Cultural factors It should be reiterated that economic factors explain only part of the causal process. Relevant non-economic influences include 'cultural factors', e.g. religion and values, and particular historical experiences. As noted earlier, post-Second World War new nations, that is, post-colonial states, which experienced Anglo-American, mainly British, rule are more likely to be democratic than those which were once colonized by France, Portugal, the Netherlands, or Belgium (Blondel, 1972, pp. 169-70; Smith, 1978, pp. 70-102; Huntington, 1984, pp. 205-6). This pattern holds for South-Southeast Asia, Oceania, the Caribbean and Africa. As reported earlier, the impact of British rule shows up strongly in the comparative statistical analysis of Bollen and Jackman (1985, pp. 34—41) while both former French and British colonization produce significant relationships in our results. Although a number of former British colonies, including Bangladesh, Burma, Ghana, India (briefly), Kenya, Nigeria, Pakistan, Tanzania, Uganda, and most recently Singapore, moved from democracy to an authoritarian system, Myron Werner has pointed up a striking phenomenon which underlies the statistics: 'Every country with a population of at least 1 million (and almost all the smaller countries as well) that has emerged from colonial rule since World War II and has had a continuous democratic experience is a former British colony' (emphasis in original. Weiner, 1987, p. 20). The British, American (The Philippines), and Australians (Papua New Guinea) helped to socialize many of their colonial peoples in democratic politics by providing systems of electoral representation and local governance prior to independence, native participation in the civil service, and primary and sometimes secondary education in the vernacular languages. These patterns were much less prevalent in other European colonial systems. The transfer of power was also more gradual and orderly in the Anglo-American colonies (von Albertini, 1982). Table 5, a cross-tabulation of last colonial power by freedom scores, clearly reflects these relationships. Earlier developments in the Americas point up the importance of pre-independence processes and the nature of the transition to independence. The predominantly British white settler societies of the United States and Canada (and Australia and New Zealand also) had considerable experience with elections, self-government, and strong judicial institutions prior to independence. In Spanish America, by contrast, the move to independence, with an electoral system for choosing rulers, was sudden, and as in much of non-British Africa over a century later, failed to take hold. The history of Brazil is more complicated because of the special role of monarchy, and will not be discussed here. Latin America differs in other significant ways. The area has a common cultural and institutional background. Its roots are largely in the Iberian peninsula with its prolonged history of armed struggles during the many centuries of the Reconquista, which fostered military values. To this must be added the impact of the ideals of the Enlightenment, which took hold among the elites in the late eighteenth and early nineteenth century. From the liberators of the early nineteenth century to dedicated democrats like Raul Alfonsin and Jose Napoleon Duarte in recent years, Latin America enjoys a history of public commitment to democratic values, particularly among the educated elites. But the culture contains major contradictory elements. Although its literature, political © UNESCO 1993 Comparative analysis of social requisites of democracy 169 Table 5. Countries of the developing world by colonial heritage and democratic status Predominant colonial heritage Current status: Democratic (FH Semi-democratic score = 2-5) (™ = 6-9) 1989 Freedom Authoritarian (FH = 10-14) Score Row total British 25 (48.1) N = 52 (65.8) 10 (19.2) (45.5) 17 (32.7 (34.7) 52 (48.0) French 0 JV= 25 4° (16.0) (18.2) 21 (84.0) (42.9) 25 (23.0) Spanish 9 (52.9) N = 17 (23.7) 6 (35.3) (27.3) 2(11.8) (04.1) 17 (16.0) Portuguese 1* (16.7) N = 6 (02.6) 0 5 (83.3) (10.2) 6 (06.0) Belgian 0 N = 3 0 3(100) (06.1) 3 (03.0 Dutch 0 N = 2 lc (50.0) (04.5) 1 (50.0) (02.0) 2 (02.0) U.S. 1 (100 ) N = 1 (02.6) 0 0 1 (01.0) Australian 1 (100 ) N = 1 (02.6) 0 0 1 (01.0) Japanese V (50.0) N = 2 (02.6) V (50.0) (04.5) 0 2 (02.0) Column total 38 (35.0) 22 (20.0) 49 (45.0) 109 (100.0) ' Madagascar, Senegal, Tunisia, French Guiana ' Brazil Surinam ' Republic of Korea Taiwan values, and pantheons of national heroes are replete with democratic emphases, it also retains the normative stereotypes of caudillaje (strong leaders) and machismo. It is a Catholic region, a background which Pierre Trudeau and others have argued is relevant. In seeking to account for the relative weakness of democracy before 1960 in the Canadian Catholic province of Quebec, as compared with the more institutionalized competitive party systems of anglophone Protestant Canada, Trudeau argued 'Catholic nations have not always been ardent supporters of democracy. They are authoritarian in spiritual matters; and since the dividing line between the spiritual and the temporal may be very fine or even confused, they are often disinclined to seek solutions in temporal affairs through the mere counting of heads' (Trudeau, 1968, p. 108). Lipset also noted that Catholicism appeared antithetical to democracy in pre-Second World War Europe, as well as in Latin America (1960, pp. 72-3). He pointed out that Catholicism earlier 'meant being allied with rightist or conservative groups in polities', while in Catholic countries, the left forces, including moderates, were anticlerical. This often helped turn partisan politics into 'a deep-rooted conflict between God and Satan. . . . And as long as religious ties reinforce secular political alignments, the chances for compromise and democratic give-and-take are weak'.24 The geography of Latin America places it close to the United States, and North American pressures have affected it substantially for good and ill. The colossus to the north has helped to destabilize a number of both democratically elected and authoritarian countries. Most recently, the emphasis on civil liberties and human rights in the Carter administration and the stress on democratic structures during Reagan's second term have had a positive effect © UNESCO 1993 170 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres on the domestic politics of nations below the Rio Grande. Some scholars claim that the Islamic faith makes political democracy in a Western sense difficult, since there is no separation of the secular and the religious worlds under Islam. Gastil's freedom scores fail to show any Islamic nation in 1989 to be free (democratic). In his analysis of Islamic orientations to the state, P. J. Vatikiotis concludes that the Muslim and Western worlds have 'sharply diverse cultural traditions and intellectual attitudes'. These are reflected particularly in the fact that 'notions of political freedom are not held in common . . . they are alien to Islam' (1988, pp. 114, 118).25 It may also be noted that the geopolitical influence of the Soviet Union had, in the past, prevented the Eastern European states from deviating from practices acceptable to the Soviets, particularly one-party rule, although their income and educational levels indicated the potential for a pluralist system. The citizens of the German Democratic Republic, Hungary, Poland, Czechoslovakia, Bulgaria, Romania and Albania vigorously demonstrated their desire for greater freedom in uprisings and massive protest waves in 1953, 1955, 1956, 1968, and most significantly in 1989 and 1990. But it was only after the breakdown of totalitarianism in the Soviet Union itself that they could erect democratic regimes. Efforts at liberalization largely supported by the growing intelligentsia, the educated managerial, professional and intellectual groups, have been aided by educational growth. Conclusions Transitions to democracy in the late 1970s and the 1980s raise the familiar issue of connections between economic development and democratization. Comparisons of political structures with levels of per capita income suggest that the relationship in the late 1980s is even more striking than in the 1950s, when the emphasis on this correlation first became part of the growing literature on development. This is logical, since the decades from 1960 to the early 1980s encompassed dramatic increases in various social indicators as well as in indices relating to national economic capability and citizens' expectations and demands. For nations with rising national incomes during this period, far higher proportions of the youth cohorts attended secondary and tertiary schools, and service and technical occupations grew significantly. Before the wave of transitions to democracy beginning in the mid 1970s, some argued that the relationships hypothesized in the 1950s did not hold.26 But the subsequent record argues for interconnections between democratic structures and rising levels of income, although economic growth does not determine political democracy by itself. Given the cyclical record of the past four decades, the proposition that economic development is a necessary requisite for political democracy cannot be made unequivocally. The results of new inquiries using sophisticated statistical techniques will not settle the argument. Although economic growth is only one element in democratization, clearly it is an important aspect. If we cannot assume the impact of growth to be mechanistic in encouraging pluralism, we should also recognize that those nations able to raise their standards of living and educational levels establish the bases for democratic structures, increasing the probabilities that democratic efforts may become institutionalized and legitimate. Finally, it may be noted that the emergence of multi-party electoral systems in Africa and the ex-Communist states of Eastern Europe and the Soviet Union in the late 1980s and early 1990s will sharply reduce the relationships reported here when research based on data from the 1990s is completed. Many extremely poor countries are now much freer than before. The diffusion of democracy to many African states is in large part a consequence of the end of a bi-polar world. The conclusion of the Cold War enabled the international system to be used to foster human rights and multiparty systems. Third World dictators could no longer play off the Soviet Union against the West. The World Bank and the International Monetary Fund have begun to set political conditions on aid. The ex-Soviet nations are also under international pressure to become or remain democratic. But if the correlations may be much lower using 1990s data, the prior research may enable us to predict the probability of breakdown - which © UNESCO 1993. Comparative analysis of social requisites of democracy 171 newly democratic regimes are most likely to fail. Thus, Haiti, many of the poorest African countries, and the ex-Soviet republics in Central Asia are unlikely candidates for stable democracy. There is need for considerable caution about the long-term prospects for multi-party structures in many of the newer systems. Their regimes are both poor and low in legitimacy. As this article demonstrates, what they need above all is efficacy, particularly in the economic arena, but also in the polity. If they can take the high road to economic development, they can keep their political house in order. But the opposite is true as well. Notes An earlier version of this text was presented on September 1990 at the 86th Annual Meeting of the American Political Science Association, San Francisco. We are indebted to Leo Goodman, Alex Inkeles, Bruce Bueno de Mesquita and Fred Turner for advice. 1. More popularly, see Revel (1985). 2. Cardoso wrote that we should reject 'nostalgia for the logic and coherence of former explanations that ignore the unexpected and contradictory aspects of real political life', that we should try to understand 'the disparity between what was postulated as a political consequence of economic growth (that is, democratization) and the actual course of a political history marked by military coups and the flourishing of authoritarian regimes' (1978, pp. 1, 24, 32). Cardoso's thesis, along with the similar assumptions of O'Donnell, were based on evidence drawn from the 1960s and 1970s. They wrote before the subsequent experiences of redemocratization in Latin America, southern Europe, and some Asian countries. 3. Bureaucratic authoritarianism grew out of the failure of progressive populist regimes, in Latin America, to control economic stagnation, balance-of-payments problems, and increasing popular demands of the 1930s and 1940s. The industrial bourgeoisie, civilian and military technocratic bureaucrats and elites coalesced to regulate the economy and social structure in hopes of mobilizing capital by suppressing the popular sectors and fostering links with foreign capital and technology (O'Donnell, 1973). 4. Seligson (1987) argues that the present cycle toward democracy in Latin America is more robust and more durable than the previous ones. 5. We specifically refer to the level of economic growth, not the rate of growth 6. As Dogan (1979, p. 747) has suggested, sociology and political science m the post-behavioural era need to build cumulative knowledge through systematic comparison. 7. For the short term, there may be a reversal of the relationship in the form of the emergence of authoritarian regimes despite increasing levels of development. However, there is a clear long-term trend in the relationship between economic development and political democracy. All the consequences of economic development produce a 'correlate change' in a political regime in a way that makes it more pluralized and inclusive. Hence, there is a 'structural compatibility', 'internal attraction', 'elective affinity', or 'inevitable tendency' between the two variables (Almond and Powell, 1978; Hoogvelt, 1982). In a sense, political pluralization or democratization is a structural imperative or a matter of efficiency. Thus, Gorbachev suggests that communist regimes must be liberalized if their societies are to develop successfully (Dahl, 1971; White et al., 1987). 8. The World Bank data do not reflect the cost of living in each nation, so that its comparisons of Gross National Product (GNP) may be somewhat misleading in the comparisons of particular nations. The Organization for Economic Co-operation and Development (OECD) has computed living costs in its member nations through the elaborately constructed index of Purchasing Power Parities (PPP), and here the buying power of someone with a median per capita income in the United States turns out to be higher than for people in several European nations where, in the World Bank comparisons, per capita income appears to be higher (OECD, 1980, pp. 91-97; OECD, 1982, pp. 31-2). 9. The World Development Report of the World Bank categorizes separately the five 'high-income oil exporters' (Oman, Libya, Saudi Arabia, Kuwait, and the United Arab Emirates) and the eight 'East European nonmarket economies' (Hungary, Romania, © UNESCO 1993. 172 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres Albania, Bulgaria, Czechoslovakia, the German Democratic Republic, Poland and the USSR). These are all authoritarian states; in terms of the Freedom House classifications, Poland, Hungary, Kuwait, and the United Arab Emirates are 'partly free' (semi-democratic) and the other states are authoritarian. The 'high-income oil exporters' are all Islamic states of North Africa or the Arabian peninsula, and the 'East European nonmarket economies' are all proximate to the Soviet Union and are heavily influenced by it. The World Bank classifies these nations separately, however, because they have something distinctive in common -high income from oil, or dependence on the USSR (World Bank, 1984, pp. 218-19). 10. Average Annual GNP growth declined for the following eleven nations during the period 1960-82: Chad, Ethiopia, Nepal, Zaire, Uganda, Niger, Madagascar, Sri Lanka, Ghana, Sudan, and Zambia. 11. All data, for every analysis in this article, were supplied by a large-scale project centred at the Hoover Institution of Stanford University. This project includes a comprehensive research file of aggregate data on every country and has commissioned case studies of 27 Third World nations (Diamond et al., 1989, p. xix). 12. If a nation has experienced colonial rule under different powers, only the last colonial power was considered. 13. The 1970 panel was excluded because a dependent variable was not available. Gastil's freedom scores first appeared in 1973. 14. GNP per capita (ECON) and political mobilization (RESIST) was log-transformed because they were skewed. 15. Note that careful scrutiny should be placed on the 1960 panel because the number of effective cases (N = 22) is too small, thus the results may not be robust. 16. (See O'Donnell, 1973). Note that, technically, O'Donnell's relationship is not just a variation of segment (b) in Jackman's model; there is a permanent downturn through segment (c). (See Sirowy and Inkeles 1990, pp. 126-34). 17. The countries in (a) and (b) were excluded because their political regimes are largely influenced by religion and ideology. 18. Uruguayan national income, compared with the previous year, fell 9.4 per cent in 1982, 4.7 per cent in 1983, and 1.8 per cent in 1984 (Sanders, 1985, p. 6). 19. For detailed analysis of the survey data on Argentine public reactions to the Falklands/Malvinas War and the electoral campaign of Raul Alfonsin (see Smith and Turner, 1984, pp. 809-13). 20. The USSR urbanized, industrialized and created a large educated intelligentsia (middle class) after the Second World War, though conditions of life worsened under Brezhnev. 21. In two separate regression analyses that also used the Hoover Institution data bank, measures of dependency involving an index of the degree of commodity concentration and trade partner concentration, were not found to be statistically significant in influencing political democracy. 22. This factor also does not produce an invariant relationship, as recent developments in Singapore have indicated. For a general discussion see Dahl and Tufte (1973). 23. They write, 'Our data show unambiguously that the schools in each of our six developing countries [Argentina, Chile, India, and Israel, Nigeria, and East Pakistan], flawed as they undoubtedly are, clearly had a substantial effect on the pupils exposed to their influence. . . . They had a different sense of time, and a stronger sense of personal and social efficacy; participated more actively in communal affairs; were more open to new ideas, new experiences, and new people; interacted differently with others, and showed more concern for subordinates and minorities. They valued science more, accepted change more readily, and were more prepared to limit the number of children they would have. In short, by virtue of having had more formal schooling, their personal character was decidedly more modern' (1974, p. 143). 24. Bollen and Jackman (1985, pp. 38-41) did not enter Catholic influence in their equations, but report no statistically significant relationship to Protestantism. 25. See also the 1988 'Islam & Politics' April issue of Third World Quarterly. 26. Kenneth Prewitt (1982, p. 10) writes, for example, "The Lipset hypothesis . . . maintained that democratic political stability and economic development were mutually reinforcing. Yet economic development has covaried as often or more often with authoritarian regimes (Taiwan, Republic of Korea, Argentina, South Africa, pre-Khomeni Iran . . .)'. © UNESCO 1993. Comparative analysis of social requisites of democracy 173 References Almond, G. A.; Powell, G. B. Jr., 1978. Comparative Politics: System, Process, and Policy. 2nd ed. Boston: Little, Brown. Banks, A. S., 1970. 'Modernization and Political Change: The Latin American and Amer-European Nations'. Comparative Political Studies, No. 2, pp. 405-18. Blondel, J., 1972. Comparing Political Systems. New York: Praeger. Bollen, K. A., 1979. 'Political Democracy and the Timing of Development'. American Sociological Review, No. 44, pp. 572-87. Bollen, K. A., 1980. 'Issues in the Comparative Measurement of Political Democracy'. American Sociological Review, No. 45, pp. 370-90. Bollen, K. A., 1983. 'World System Positions of Dependency and Democracy: The Cross-National Evidence'. American Sociological Review, No. 48, pp. 468-79. Bollen, K. A.; Jackman, R. W., 1985. 'Economic and Noneconomic Determinants of Political Democracy in the 1960s'. Research in Political Sociology, No. 1, pp. 27-48. Greenwich: J AI Press. Cardoso, F. H., 1978. 'On the Characterization of Authoritarian Regimes in Latin America'. Working papers No. 30. Cambridge: Cambridge University Paper Series No. 30, Centre for Latin American Studies. Cnudde, C. F.; Neubauer, D. E. (eds.), 1969. Empirical Democratic Theory. Chicago: Markham. Collier, D. (ed.), 1979. The New Authoritarianism in Latin America. Princeton: Princeton University Press. © UNESCO 1993. Coulter, P. B., 1975. Social Mobilization and Liberal Democracy: A Macroquantitative Analysis of Global and Regional Models. Lexington: Lexington Press. Cutright, P., 1963. 'National Political Development: Measurement and Analysis'. American Sociological Review, No. 28, pp. 253-64. Dahl, R. A., 1971. Polyarchy: Participation and Opposition. New Haven: Yale University Press. Dahl, R. A.; Tufte, E., 1973. Size and Democracy. Stanford: Stanford University Press. Diamond, L. J., 1980. "The Social Foundations of Democracy: The Case of Nigeria'. PhD dissertation, Department of Sociology, Stanford University. Diamond, L. J.; Linz, J. J.; Lipset, S. M., 1987. 'Building and Sustaining Democratic Government in Developing Countries'. World Affairs, No. 150, pp. 5-18. Diamond, L. J.; Linz, J. J.; Upset, S. M. (eds.), 1989. Democracy in Developing Countries. 3 vols. Boulder: Lynne Rienner. Dogan, M., 1979. 'In Memoriam: Stein Rokkan, Sociologue politique comparatiste, 1921-1979', Revue Francoise de Sociologie, No. 20, pp. 747-8. Flanigan, W. H.; Fogelman, E., 1971. 'Patterns of Political Development and Democratization: A Quantitative Analysis'. In John V. Gillespie and Betty A. Nesvold (eds.), Macro-Quantitative Analysis: Conflict, Development, and Democratization. Beverly Hills: Sage. Flora, P., 1973. 'Historical Processes of Social Mobilization: Urbanization and Literacy, 1850-1965'. In S. N. Eisenstadt and Stein Rokkan (eds.), Building States and Nations: Models and Data Resources, vol. I. Beverly Hills: Sage. GaSTIl, R. D., 1985. Freedom in the World: Political Rights and Civil Liberties, 1984-85. Westport: Greenwood Press. Haggard, S., 1990. 'Democracy and Economic Growth'. Paper prepared for the US Agency for International Development, Democratic Pluralism Initiative. Herz, J. H. (ed), 1982. From Dictatorship to Democracy. Westport: Greenwood Press. Higgott, R. A., 1983. Political Development Theory. Beckham, UK: Croom Helm. Hoogvelt, A. M. M., 1982. The Third World in Global Development. Hampshire: Macmillan. Huntington, S. P. 1968. Political Order in Changing Societies. New Haven, Conn.: Yale University Press. Huntington, S. P., 1984. 'Will More Countries Become Democratic?' Political Science Quarterly, No. 99, pp. 199-202. Huntington, S. P.; Dominguez, J. I., 1975. 'Political Development'. In Fred Greenstein and Nelson Polsby (eds.), Handbook of Political Science, vol. 3. Reading, Mass.: Addison-Wesley. Inkeles, A.; Diamond, L. J., 1980. 'Personal Development and National Development: A Cross-National Perspective'. In Alexander Szalai and Frank M. Andrews (eds.), The Quality of Life: Comparative Studies. London: Sage. Inkeles, A.; Smith, D. H., 1974. Becoming Modern: Individual 174 Seymour Martin Lipset, Kyoung-Ryung Seong and John Charles Torres Change in Six Developing Countries. Cambridge, Mass.: Harvard University Press. Karl, T., 1990. 'Dilemmas of Democratization in Latin America". Comparative Politics, No. 23, pp. 1-21. Kaufman, R. R.; Chernetsky, H. I.; Gellner, D., 1975. 'A Preliminary Test of the Theory of Dependency'. Comparative Politics, No. 7, pp. 303-30. Kim, C. L., 1971. 'Socio-economic Development and Political Democracy in Japanese Prefectures'. The American Political Science Review, No. 65, pp. 184-6. Kurth, J. R., 1979. 'Industrial Change and Political Change: A European Perspective'. In David Collier (ed.), The New Authoritarianism in Latin America. Princeton, NJ.: Princeton University Press. Kuznets, S., 1963. 'Economic Growth of Small Nations'. In E. A. G. Robinson (ed.), Economic Consequences of the Size of Nations. London: Macmillan. LaPalombara, J.; Weiner, M., 1966. 'The Origin and Development of Political Parties'. In LaPalombara and Weiner (eds.), Political Parties and Political Development. Princeton, NJ.: Princeton University Press. Lerner, D., 1958. The Passing of Traditional Society. Glencoe, 111.: Free Press. Lijphart, A., 1972. 'Toward Empirical Democratic Theory: Research Strategies and Tactics'. Comparative Politics, No. 4, pp. 417-32. Linz, J. J.; Stepán, A. (eds.), 1978. The Breakdown of Democratic Regimes. Baltimore: Johns Hopkins University Press. Lipset, S. M., 1959. 'Some Social Requisites of Democracy: Economic Development and Political Legitimacy'. American Political Science Review, No. 53, pp. 69-105. Lipset, S. M., 1960. Political Man: The Social Bases of Politics. New York: Doubleday. Expanded Edition, Baltimore: Johns Hopkins University Press, 1981. Lipset, S. M., 1963. The First New Nation: The United States in Historical and Comparative Perspective. New York: Basic Books. Expanded Edition. New York: Norton Library. Malloy, J. M.; Seligson, M. A. (eds.), 1987. Authoritarians and Democrats: Transition in Latin America. Pittsburgh: University of Pittsburgh Press. Marquette, J., 1974. 'Social Change and Political Mobilization in the United States: 1870-1960'. The American Political Science Review, No. 68, pp. 1058-74. Menges, C. C, 1985. 'Commentaries'. AEI Foreign Policy and Defense Review, No. 5, pp. 39-41. Needler, M. C, 1968. 'Political Development and Socioeconomic Development: The Case of Latin America'. The American Political Science Review, No. 62, pp. 889-97. Neubauer, D. E., 1967. 'Some Conditions of Democracy'. 77ie American Political Science Review, No. 61. pp. 1002-9. O'Donnell, G., 1973. Modernization and Bureaucratic Authoritarianism: Studies in South American Politics. Politics of Modernization Series, no. 9. Berkeley: Institute of International Studies, University of California. O'Donnell, G.; Schmitter, P. C; Whitehead, L. (eds.), 1987. Transitions From Authoritarian Rule. 4 vols. Baltimore: Johns Hopkins University Press. OECD, 1980. 'Main Aggregates, 1951-1980'. National Accounts, Vol. 1, part 4. Paris: Organization for Economic Cooperation and Development. OECD, 1982. 'Comparing Real Standards of Living'. Observer, No. 115, pp. 31-32. Paris: Organization for Economic Cooperation and Development. Olsen, M. E., 1968. 'Multivariate Analysis of National Political Development'. American Sociological Review, No. 33, pp. 699-712. Prewitt, K., 1982. The Impact of the Developing World on U.S. Social Science Theory and Methodology'. In Laurence D. Stifel, Ralph K. Davidson and James S. Coleman (eds.), Social Sciences and Public Policy in the Developing World. Lexington: D. C. Heath. Revel, J.-F., 1985. How Democracies Perish. Translated from the French by William Byron. New York: Harper & Row. Roskin, M., 1978. 'Spain Tries Democracy Again'. Political Science Quarterly, No. 93, pp. 62ÍM7. Rustow, D., 1970. 'Transitions to Democracy'. Comparative Politics, No. 2, pp. 337-63. Sanders, T. G., 1985. Uruguay's Return to Democracy. Indianapolis: Universities Field Staff International. Seligson, M. A., 1987. 'Development, Democratization, and Decay: Central America at the Crossroads'. In James H. Malloy and Martin A. Seligson (eds.), Authoritarians and Democrats. Pittsburgh: University of Pittsburgh Press. Sirowy, L.; Inkeles, A.. 1990. "The Effects of Democracy on Economic Growth and Inequality: A Review'. Studies in Comparative International Development, No. 25 , pp. 126-57. © UNESCO 1993 Comparative analysis of social requisites of democracy 175 Smith, B. H.; Turner, F. C, 1984. 'Survey Research in Authoritarian Regimes: Brazil and the Southern Cone of Latin America'. In James W. Wilkie and Adam Perkal (eds.), Statistical Abstract of Latin America, vol. 23, Los Angeles: UCLA Latin American Center Publications, University of California. Smith, T., 1978. 'A Comparative Study of French and British Decolonization'. Comparative Studies in Society and History, No. 20, pp. 70-102. Stepán, A., 1973. 'The New Professionalism of Internal Warfare and Military Role Expansion'. In Alfred Stepán (ed.). Authoritarian Brazil: Origins, Policies, and Future. New Haven: Yale University Press. Tomasson, R., 1980. Iceland: The First New Society. Minneapolis: University of Minnesota Press. Triffin, R., 1963. 'The Size of the Nation and its Vulnerability to Economic Nationalism'. In E. A. G. Robinson (ed.). Economic Consequences of the Size of Nations. London: Macmillan. Trudeau, P. E., 1968. Federalism and the French Canadians. New York: St. Martin's Press. Vatikiotis, P. J., 1988. Islam and the State. Beckham, UK: Croom Helm. von Albertini, R., 1982. Decolonization. New York: Afričana Publishing Co. Weede, E., 1983. The Impact of Democracy on Economic Growth: Some Evidence from Cross-National Analysis'. Kyklos, No. 36, pp. 21-39. Weiner, M., 1987. 'Empirical Democratic Theory'. In Myron Weiner and Ergun Ozbudun (eds.). Competitive Elections in Developing Countries. Durham: Duke University Press. White, S.; Gardner, J.; ShopAin, G., 1987. Communist Political Systems. New York: St. Martin's Press. Wiarda, H. J., 1990. Rethinking Political Development: A look Backward Over Thirty Years, and a Look Ahead'. Studies in Comparative International Development. No. 24, pp. 65-82. Winham, G. R., 1970. 'Political Development and Lemer's Theory: Further Test of a Causal Model'. The American Political Science Review, No. 64, pp. 810-18. World Bank. 1984. World Development Report, 1984. New York: Oxford University Press. © UNESCO 1993.