¹ Lipset, Seymour Martin, “Some Social Requisites of Democracy: Economic Development and Political Legitimacy,” American Political Science Review 53 (March 1959)CrossRefGoogle Scholar; and idem, Political Man: The Social Bases of Politics (Baltimore: Johns Hopkins University Press, 1981)Google Scholar.

² Diamond, Larry, “Economic Development and Democracy Reconsidered,” in Marks, Gary and Diamond, Larry, eds., Reexamining Democracy: Essays in Honor of Seymour Martin Lipset (Newbury Park, Calif.: Sage Publications, 1992)Google Scholar.

³ While different data sets and different estimation methods lead to somewhat divergent results, the most careful statistical study of the aggregate patterns thus far, by Burkhart and Lewis-Beck, finds that economic development Granger causes democracy. Burkhart, Ross E. and Lewis-Beck, Michael S., “Comparative Democracy: The Economic Development Thesis,” American Political Science Review 88 (December 1994), 903–10CrossRefGoogle Scholar.

⁴ A fair amount of ink has been spilled over whether the relation between development and democracy is linear. See Jackman, Robert W., “On the Relation of Economic Development to Democratic Performance,” American Journal of Political Science 17 (August 1973), 611–21CrossRefGoogle Scholar; and Arat, Zehra F., “Democracy and Economic Development: Modernization Theory Revisited,” Comparative Politics 21 (October 1988), 21–36CrossRefGoogle Scholar. We now know better. Democracy, however measured, is a qualitative or a limited variable: it assumes values of 0 or 1 under our measurement; it ranges from 2 to 14 on the Freedom House Scale created by Gastil, R. D., Freedom in the World: Political Rights and Civil Liberties, 1987–88 (New York: Freedom House, 1988)Google Scholar; from 0 to 100 on the scale of Bollen, Kenneth A., “Issues in the Comparative Measurement of Political Democracy,” American Sociological Review 45 (June 1980), 370–90CrossRefGoogle Scholar, and so on. Hence, no predicted index of democracy can become negative as the level of development tends to zero, and no predicted index of democracy can exceed whatever is the maximum value of a particular scale as the level gets very large. Only a nonlinear function, such as the normal or logistic, as suggested by Robert A. Dahl can satisfy these constraints. See Dahl, , Polyarchy (New Haven: Yale University Press, 1971)Google Scholar. This is why we use probit or logit models throughout.

⁵ This is not quite true of our data set, since different countries enter and exit the sample at different moments. For now, we consider the population of countries as fixed, but see Section IV.

⁶ Diamond (fn. 2), 45; as well as Huber, Evelyne, Rueschemeyer, Dietrich, and Stephens, John D., “The Impact of Economic Development on Democracy,” Journal of Economic Perspectives 7 (Summer 1993), 71–86CrossRefGoogle Scholar.

⁷ O'Donnell, Guillermo, Modernization and BureaucraticAuthoritarianism: Studies in South American Politics (Berkeley: Institute of International Studies, University of California, 1973), 3Google Scholar.

⁸ Goran Therborn, “The Rule of Capital and the Rise of Democracy,” New Left Review, no. 103 (May-June 1977).

⁹ Lipset (fn. 1, 1959), 56.

¹⁰ Our regime classification and the resulting list of regimes are described in Appendix 1 and in Mike Alvarex et al., “Classifying Political Regimes,” Studies in International Comparative Development (forthcoming). The reason for selecting this period and the sample is the availability of internationally comparable economic data, which we took from the Penn World Tables 5.6. The sample we describe here and use throughout does not include six countries that derive at least half of their income from oil revenues. While political data are available for 4,730 country years, data for economic growth are available for only 4, 126 country years, which is the number of observations in most analyses.

¹¹ Readers used to the UN or the World Bank GNPfiguresshould be aware that counting incomes at purchasing-power parities tends to increase significantly the levels for poor countries and to decrease slightly the numbers for rich countries. It may be useful for future reference to know what different numbers describe: by 1990, Nigeria had a per capita income of $995, Indonesia had $1,973, Czechoslovakia $4,094, Spain $9,576, and the United States $18,073.

¹² Note that we do not distinguish successive dictatorships. If President Viola succeeds President Videla or even if ayatollahs succeed a shah, we treat it as one continuous spell of dictatorship.

¹³ Huntington, Samuel P., Political Order in Changing Societies (New Haven: Yale University Press, 1968), 43Google Scholar.

¹⁴ An analogy may be useful. Suppose that someone runs the risk of 0.01 of dying from accidental causes during each year of her life and that at the age of seventy-eight she gets hit by a falling brick. To attribute this death to development is to conclude that she died of old age.

¹⁵ The claim about the prewar period is based on rather heroic backward extrapolation of 1950 incomes, but the levels at which democracies fell in Europe were an order of magnitude lower: we guess it to have been $1,825 in Austria in 1934, $1,974 in Finland in 1930, $1,474 in Germany in 1933, and $1,814 in Italy in 1922.

¹⁶ Londregan, John B. and Poole, Keith T., “Poverty, the Coup Trap, and the Seizure of Executive Power,” World Politics 42 (January 1990)CrossRefGoogle Scholar. Londregan and Poole found a similar pattern with regard to coups. In their sample of 121 countries between 1950 and 1982 coups were twenty-one times more likely to occur among the poorest than among the wealthiest countries.

¹⁷ Expected life in any state is the inverse of the probability of transition away from this state.

¹⁸ Adam Przeworski, “Why Democracies Survive in Affluent Countries?” (Paper presented at the annual meeting of the American Political Science Association, San Francisco, August 28-September 1, 1996).

¹⁹ In the long run the proportion of democracies equals p_AD/ (p_AD * P_DA), where p stands for transition probabilities, A for dictatorship (“authoritarianism”), and D for democracy. See Appendix 2. The numbers in the text are derived from Table 1.

²⁰ O'Donnell (fn. 7), 4.

²¹ Huntington, Samuel P. and Nelson, Joan M., No Easy Choice: Political Participation in Developing Countries (Cambridge: Harvard University Press, 1976) 19CrossRefGoogle Scholar.

²² Olson, Mancur Jr., “Rapid Growth as a Destabilizing Force,” Journal ofEconomic History 23 (December 1963)Google Scholar.

²³ Huntington (fn. 13).

²⁴ Lipset (fn. 1, 1981), 54.

²⁵ This finding parallels again the results of Londregan and Poole (fn. 16) with regard to coups, which they found to be less likely when the economy grows.

²⁶ Diamond, Larry and Linz, Juan J.. “Introduction: Politics, Society, and Democracy in Latin America,” in Diamond, L., Linz, J. J., and Lipset, S. M., eds., Democracy in Developing Countries: Latin America (Boulder, Colo.: Lynne Rienner, 1989), 17Google Scholar.

²⁷ Huntington (fn. 13), 1.

²⁸ Ibid., 7.

²⁹ Ibid., 35–36.

³⁰ O'Donnell, , Modernization and Bureaucratic Authoritarianism: Studies in South American Politics, 2d ed. (Berkeley: Institute of International Studies, University of California, 1979), 204Google Scholar.

³¹ O'Donnell was careful about not making general claims: his purpose was to explain the downfall of democracies in the Southern Cone. But his theory of “bureaucratic authoritarianism” captured the imagination of scholars around the world, who treated it as applicable almost everywhere.

³² Huntington, Samuel P., The Third Wave: Democratization in the Late Twentieth Century (Norman: University of Oklahoma Press, 1991), 16Google Scholar. Huntington considered 74 countries while our sample covers 135 countries; hence, the data are not exactly comparable.

³³ Moore, Barrington Jr., Social Origins ofDictatorship andDemocracy (Boston: Beacon Press, 1965)Google Scholar.

³⁴ The exception is Czechoslovakia, but note that no alternation in office between parties occurred during this period. Indeed, the first alternation resulting from elections in the history of Eastern Europe occurred in Poland in 1991.

³⁵ World Bank, World Bank Development Report 1991 (Washington, D.C.: World Bank, 1991)Google Scholar, Table 1.1.

³⁶ A third question has also been posed: when D. A. Rustow, pointed out that the levels of development at which different countries permanently established democratic institutions vary widely, Lipset's (fn. 1, 1981) rejoinder was that the thresholds at which democracy was established were lower for the early democracies; see Rustow, , “Transitions to Democracy,” Comparative Politics 2 (April 1970)Google Scholar. A rough guess at comparing the levels at which democracies emerged before and after the war indicates that levels at which democracy was established before the war must have been on the average lower. But the distribution of incomes during the two periods was not the same: it is doubtful that many countries enjoyed incomes above $4,000 before the war. Hence, we do not know how long the countries that were poor at the time would have waited before becoming democracies. At most, we can compare the distribution of levels at which democratization occurred before the war with the distribution in the postwar period truncated at $4,000. If in addition to the guesses presented in Table 6 we also assume that incomes were lower in Eastern Europe and most of Latin America, the two distributions will be highly similar. But that is too many guesses to take seriously.

³⁷ Suppose that the function which relates regimes to level is Pr[REGIME(t)=DEMOCRACY]=REG(t)=F[α+βLEVEL(t)], where F stands for a normal or logistic distribution. Now subtract and add βLEVEL(0) within the square brackets, to get REG(t)=F(α+βLEVEL(0)+ β[LEVEL(t)-LEVEL(0)]}. Defining LEVEL(0) as INI and LEVEL(t)-LEVEL(0) as DEV(t), and allowing the (cross-sectional) effect of the initial level to differ from the (dynamic) effect of development yields REG(t)= F[α+β_c INI+β_DDEV(t)].This is the model we estimated, by dynamic probit.

³⁸ Huber, Rueschemeyer, and Stevens (fn. 6) go back just a few decades but the question remains: why would conditions found in the 1920s cause events in the 1960s, not earlier or later?

³⁹ Przeworski, Adam, Democracy and the Market: Political and Economic Reforms in Eastern Europe and Latin America (New York: Cambridge University Press, 1991)CrossRefGoogle Scholar.

⁴⁰ O'Donnell, Guillermo and Schmitter, Philippe C., Transitionsfrom Authoritarian Rule (Baltimore: Johns Hopkins University Press, 1986)Google Scholar.

⁴¹ Galenson, Walter, “Introduction” to Galenson, ed., Labor and Economic Development (New York: Wiley, 1959), 3Google Scholar.

⁴² Karl de Schweinitz Jr., Industrialization, Labor Controls and Democracy,” Economic Development and Cultural Change 7 (July 1959).

⁴³ Huntington, Samuel P. and Dominguez, Jorge I., “Political Development,” in Greenstein, F. I. and Polsby, N. W., eds., Handbook of Political Science, vol. 3 (Reading, Mass.: Addison-Wesley, 1975), 60Google Scholar.

⁴⁴ Huntington and Nelson (fn. 21), 23.

⁴⁵ For a full explanation and historical details, see Alvarez et al. (fn. 10).

⁴⁶ Inkeless, Alex, “Introduction,” Studies in Comparative International Development 25 (Spring 1990), 3–6Google Scholar. Note, however, that different measures appear to be biased in somewhat different directions. See Bollen, Kenneth A., “Liberal Democracy: Validity and Method Factors in Cross-National Measures,” American Journal of Political Science 37 (November 1993)CrossRefGoogle Scholar.

⁴⁷ Coppedge, Michael and Reinicke, Wolfgang H., “Measuring Polyarchy,” Studies in Comparative International Development 25 (Spring 1990), 51–72CrossRefGoogle Scholar.

⁴⁸ Bollen, Kenneth A., “Political Democracy and the Timing of Development,” American Sociological Review 44 (August 1979), 572–87CrossRefGoogle Scholar.

⁴⁹ Gurr, Ted Robert, Jaggers, Keith, and Moore, Will H., “The Transformation of the Western State: The Growth of Democracy, Autocracy, and State Power since 1800,” Studies in Comparative International Development 25 (Spring 1990)CrossRefGoogle Scholar.

⁵⁰ Gastil (fn. 4).

⁵¹ Convergence is monotonic if p_AB + p_DA < 1; otherwise, the proportions of regimes will oscillate around the equilibrium.

⁵² Amemyia, Takeshi, Advanced Econometrics (Cambridge: Harvard University Press, 1985), chap. 11Google Scholar.