¹ Zimmermann, Ekkart, Political Violence, Crises and Revolution (Cambridge, MA: Schenk-man Publishing Company, 1983)Google Scholar provides an extensive review of the literature on causes of political instability. Four perspectives of particular interest are modernization theory, psychological theories of relative deprivation, perspectives derived from Marxist political economy, and those from more conventional economics. For an explanation of modernization theory see Huntington, Samuel P., Political Order in Changing Societies (New Haven: Yale University Press, 1968)Google Scholar. Examples of the psychological approach are Feierabend, Ivo K. and Feierabend, Rosalind L., “Systemic Conditions of Political Aggression: An Application of Frustration-Aggression Theory,” in Feierabend, Ivo K., Feierabend, Rosalind L., and Gurr, Ted Robert, eds., Anger, Violence and Politics (Englewood Cliffs, NJ: Prentice-Hall, 1972), 136-83Google Scholar, and Gurr, Ted Robert, Why Men Rebel (Princeton: Princeton University Press, 1970)Google Scholar. Representative presentations of the Marxist perspective can be found in Amin, Samir, Accumulation on a World Scale (New York: Monthly Review Press, 1974)Google Scholar, Vols. I and II, and Wallerstein, Immanuel, The Modern World System (New York: Academic Press, 1974)Google Scholar. For a more conventional approach see Myrdal, Gunnar, Rich Lands and Poor: The Road to World Prosperity (New York: Harper and Row, 1958)Google Scholar, and Hirschman, Albert O., The Strategy of Economic Development (New Haven: Yale University Press, 1962)Google Scholar.

² See for example, Amin (fn. 1) and Wallerstein (fn. 1).

³ For an explanation of modernization theory see Huntington (fn. 1).

⁴ An example of a contribution from a Marxist perspective is Baran, Paul A., The Political Economy of Growth (New York: Monthly Review Press, 1957)Google Scholar. Baran's work was further developed by Frank, Andre Gunder, Latin America: Vnderdevelopment or Revolution (New York: Monthly Review Press, 1969)Google Scholar. See also Furtado, Celso, Economic Development of Latin America: A Surveyfrom Colonial Times to the Cuban Revolution (London: Cambridge University Press, 1970)Google Scholar. One of the most interesting reformulations of dependency theory is Cardoso, Fernando Henrique and Faletto, Enzo, Dependency and Development in Latin America (Berkeley: University of California Press, 1979)Google Scholar. For non-Marxist perspectives see O'Donnell, Guillermo A., Modernization and Bureaucratic-Authoritarianism, Politics of Modernization Series No. 9 (Berkeley: Institute of International Studies, University of California, 1973), 291–93Google Scholar; Collier, David, ed., The New Authoritarianism in Latin America (Princeton: Princeton University Press, 1979)Google Scholar; and Becker, David G., Frieden, Jeffry A., Schatz, Sayre P., and Sklar, Richard L., Postimperialism (Boulder, CO: Lynne Rienner Publishers, 1987)Google Scholar.

⁵ The word dependency is derived from the Spanish word dependencia. Advocates of this viewpoint see dependencia not as a theory but rather as a world view which asserts that the economic growth of peripheral economies is conditioned by the fluctuations and growth of the dominant industrial economies. The historic development of dependencia is discussed in Chilcote, Ronald H. and Edelstein, Joel C., Latin America: Capitalist and Socialist Perspectives of Development and Vnderdevelopment (Boulder, CO: Westview Press, 1986), chaps. 3 and 4Google Scholar, and Corradi, Juan Eugenio, “Cultural Dependence and the Sociology of Knowledge: The Latin American Case,” in Nash, June, Corradi, Juan, and Spalding, Hobart Jr., eds., Ideology and Social Change in Latin America (New York: Gordon and Breach Science Publishers, 1977), 7–30Google Scholar. For specific works see Baran (fn. 4) and Frank (fn. 4).

⁶ Cardoso and Faletto (fn. 4). See also Warren, Bill, Imperialism, Pioneer of Capitalism (London: New Left Books, 1980)Google Scholar.

⁷ See, for example, O'Donnell, Guillermo A., Schmitter, Philippe C., and White-head, Laurence, Transitionsfrom Authoritarian Rule (Baltimore: Johns Hopkins University Press, 1986)Google Scholar, or Stepan, Alfred C., Rethinking Military Politics (Princeton: Princeton University Press, 1988)Google Scholar. See also Myrdal (fn. 1) and Hirschman (fn. 1).

⁸ The concept of bureaucratic authoritarianism was first presented in O'Donnell (fn. 4). See also two articles by O'Donnell, , “Reflections on Patterns of Change in the Bureaucratic-Authoritarian State,” Latin American Research Review 13, No. 1 (1984), 3–38Google Scholar, and “Tensions in the Bureaucratic-Authoritarian State and the Question of Democracy,” in Collier, David, ed., The New Authoritarianism in Latin America (Princeton: Princeton University Press, 1979), 285–318Google Scholar. For a critique see Remmer, Karen L. and Merkx, Gilbert W., “Bureaucratic-Authoritarianism Revisited,” Latin American Research Review 17, No. 2 (1982), 3–40Google Scholar.

⁹ See Becker et al. (fn. 4).

¹⁰ For the most comprehensive review of the empirical studies of political instability see Sanders, David, Patterns of Political Instability (New York: Saint Martin's Press, 1981), chaps. 6–9CrossRefGoogle Scholar; Hibbs, Douglas A., Mass Political Violence: A Cross-national Causal Analysis (New York: Wiley, 1973)Google Scholar; and Zimmermann (fn. 1).

¹¹ Sanders (fn. 10), chaps. 7–9. While Sanders finds no statistically significant relationships between dependency, modernization, and political instability on a global level, he does find some evidence for the dependency argument in Latin America at the level of bivariate correlations (pp. 159–61).

¹² Hibbs (fn. 10).

¹³ O'Donnell (fn. 4) and Cardoso and Faletto (fn. 4). For exceptions see Midlarsky, Manus I. and Tanter, Raymond, “Toward a Theory of Political Instability in Latin America,” Journal of Peace Research 4 (1967), 12–26CrossRefGoogle Scholar; Needier, Martin C., Political Development in Latin America: Instability, Violence and Evolutionary Change (New York: Random House, 1968)Google Scholar; Putnam, Robert D., “Toward Explaining Military Intervention in Latin American Politics,” World Politics 20 (October 1967), 83–110CrossRefGoogle Scholar; and Sanders (fn. 10).

¹⁴ O'Donnell' (fn. 4), 11. He specifically excludes Central America from his analysis and then focuses on the larger economies. The closest he comes to discussing regime types in the smaller Latin American states is to classify them as traditional authoritarian (p. 114). Cardoso and Faletto (fn. 4), 122–24, also provide no more than a brief analysis of how international economic forces determine class relations in Central America. They argue that the system continues to be based on dominance by traditional agricultural exporters, enclave foreign interests, and the military. They make no attempt to trace the dynamics of change in those relationships. They treat the region as a unit.

¹⁵ There are many interesting studies of the problems of social discontent, violence, and repression in Central American countries. For Costa Rica, see, for example, Bell, John Patrick, Guerra Civil en Costa Rica (San Jose: EDUCA, 1985)Google Scholar; for Salvador, El, Vejar, Rafael Guidos, El Ascenso del Militarismo en El Salvador (San Jose: EDUCA, 1982)Google Scholar, Menjivar, Rafael, Formation y Lucha del Proletariado Industrial Salvadoreno (San Salvador: UCA Editores, 1979)Google Scholar, and Anderson, Thomas R., El Salvador: Los Sucesos Politicos de 1932 (San Jose: ED-UCA, 1976)Google Scholar; for Guatemala, Peralta, Gabriel Edgardo Aguilera and Imery, Jorge Romero, Dialectica del Terror en Guatemala (San Jose: EDUCA, 1981)Google Scholar, and Poitevin, Rene, El Proceso de Industrialization en Guatemala (San Jose: EDUCA, 1977)Google Scholar; for Honduras, Rosenberg, Mark B. and Shepherd, Phillip L., Honduras Confronts Its Future: Contending Perspectives on Critical Issues (Boulder, CO: Lynne Rienner Publishers, 1986)Google Scholar, and Meza, Victor, Historia del Movimiento Obrero Hondureno (Tegucigalpa: Editorial Guaymuras, 1980)Google Scholar; for Nicaragua, see Julio, López C, Orlando, Núnez S., Barrios, Carlos Fernando Chamorro, and Serres, Pascual, La Caida del Somotismo y la Lucha Sandinista en Nicaragua (San Jose: EDUCA, 1979)Google Scholar, or Millet, Richard, Guardians of the Dynasty (Maryknoll, NY: Orbis Books, 1977)Google Scholar.

¹⁶ See for example, Frank (fn. 4) and Furtado (fn. 4). A limited number of recent works on Central America provide some basis for further empirical treatment of the role of international forces in Central American political instability. See Camacho, Daniel and Menjivar, Rafael, Movimientos Populares en Centra America (San Jose: EDUCA, 1985)Google Scholar; Barry, Tom, Roots of Rebellion: Land and Hunger in Central America (Boston: South End Press, 1987)Google Scholar; and Williams, Robert G., Export Agriculture and the Crisis in Central America (Chapel Hill: University of North Carolina Press, 1986)CrossRefGoogle Scholar.

¹⁷ This part of the thesis most closely parallels the global discussions of economic forces in Frank (fn. 4) and Cardoso and Faletto (fn. 4), the regional discussions of Camacho and Men-jivar (fn. 16) and Williams (fn. 16), and the country analysis, for example, of Menjivar (fn. 15) or Poitevin (fn. 15).

¹⁸ External economic crisis can be thought of here as the element which stimulates the frustration which in turn sets off the dynamics of aggression and social discontent described by Feierabend and Feierabend (fn. 1) and Gurr (fn. 1).

¹⁹ This argument runs counter to Huntington's modernization theory (fn. 1). It fits O'Donnell's general idea (fn. 4) that economic stagnation sets off a cycle of discontent which results in repression. However, in O'Donnell's argument for the large Latin American countries, stagnation occurs when the easy state of industrial import substitution ends. In small, highly open export economies like those found in Central America the source is more clearly international economic instability.

²⁰ This part of the thesis builds upon Cardoso and Faletto's (fn. 4) contention that in Central America in the twentieth century there has been no regime change but only leadership or government change as the coalition of military, traditional agricultural, and enclave forces reestablish themselves. It supports O'Donnell's (fn. 4) general contention that at least until the 1980s some five of the Central American countries might be classified generally as traditional authoritarian regimes.

²¹ Cardoso and Faletto (fn. 4); O'Donnell (fn. 4).

²² One interesting hypothesis which will have to be explored elsewhere is that the period from 1986 to 1989 can be viewed as a unique example of regime change from traditional authoritarian to democratic as opposed to simply the continued unprogrammed rotation of military leaders and governments. This may be due to the demands of the larger urban middle and working class which developed as a result of import substitution stimulated through Central America's common market between 1960 and 1978. Certainly levels of urbanization increased from 1965 (37.5%) to 1983 (45%).

²³ See, for example, Daniel Rapport, “American Foreign Policy in Central America: A Method of Analysis and Results” (Unpublished paper, John F. Kennedy School of Government, Harvard University, June 1988).

²⁴ For discussions of the impact of population density and ethnicity on political discontent and repression see Richard Adams and Michael Stone, “Memorandum on Relations between Native Americans and the State in Central America” (Paper prepared for the International Commission for Central American Recovery and Development, April 1988). Specific country examples are Armando Brown, “Los Pueblos Indigenas Ante el Mundo,” Revista de la Es-cuela Nationalde Antropologia e Historia, Ano 1, 1 (June 1980), 2–5, and CIDCA Development Study Unit, “Ethnic Groups and the Nation State: The Case of the Atlantic Coast in Nicaragua” (Department of Anthropology, University of Stockholm, 1987). For more general sources, see Oit, Cepal Fao, Tenencia de la Tierra y Desarrollo Rural en Centroamerica (San Jose: EDUCA, 1980)Google Scholar, and Barry (fn. 16). For excellent treatment of the development of the class structures in Guatemala and Costa Rica and the different treatment of indigenous groups, see Peláez, Severo Martínez, La Patria del Criollo: Ensayo de Interpretation de la Real-idad Colonial Guatemalteca (San Jose: EDUCA, 1985)Google Scholar, and Stone, Samuel Z., La Dinastia de los Conquistadores: La Crisis del Poder en la Costa Rica Contemporanea (San Jose: EDUCA, 1975)Google Scholar.

²⁵ The accentuated effects of economic crisis on countries with dense population and indigenous groups whose traditional relationship to the land has been broken is noted by Men-jivar (fn. 15), and Colindres, Eduardo, Fundamentos Econotnicos de la Burgesia Salvadorena (San Salvador: UCA Editores, 1977)Google Scholar. The opposite effect is noted in less dense Costa Rica in Carcanholo, Reinaldo, Desarrollo del Capitalismo en Costa Rica (San Jose: EDUCA, 1981)Google Scholar.

²⁶ See fn. 16.

²⁷ While these two factors are extremely important, it was difficult to get accurate region-wide data. The best recent source in this matter is the International Commission for Central American Reconstruction and Development at Duke University. Background papers compiled for the commission are being published by Duke University Press.

²⁸ The trade time series at constant prices was constructed from each country's customs data because no official GDP data for Central America between 1900 and 1950 exist. The portion of the series used in this study extends from 1930 to 1988 and correlates highly with the GDP projections recently completed by Thomas, Victor Bulmer, “World Recession and Central American Depression: Lessons from the 1930s for the 1980s,” in Duran, Esperanza, ed., Latin America and the World Recession (Cambridge: Cambridge University Press, 1985), 130-51CrossRefGoogle Scholar. After 1950 GDP growth data come from the time series in the Economist's World Business Cycles. World GDP growth rate data from 1930 to 1985 are the unweighted averages of U.S. and U.K. growth between 1930 and 1950 as presented in World Business Cycles. These two countries were selected because they were Central America's primary trading partners during that period. From 1950 to the present, World Business Cycles computes overall world GDP growth rate data.

²⁹ Crisis periods included both contraction and recovery because there was a general sense that an economic crisis was not over until there was a long enough recovery period to demonstrate that a trend reversal had taken place.

³⁰ Leadership instability is distinguished here from government, regime, and political instability. Helpful definitions of terms can be found in Sanders (fn. 10), 49–69; Blondel, Jean, An Introduction to Comparative Government (New York: Weidenfeld and Nicholson, 1968)Google Scholar; and Easton, David, The Political System: An Inquiry into the State of Political Science (New York: Knopf, 1953)Google Scholar. Using Sanders's definitions, leadership instability refers to higher than normal levels of nonprogrammed changes of the chief executive; government instability to high levels of nonprogrammed change of the chief executive and cabinet. Regime instability refers to high levels of nonprogrammed changes in goals, norms, and authority structures, for example, moving from oligarchic rule to democratic participation in elections or decisions. Political instability might be defined as the higher than normal level of peaceful or violent social discontent which challenges or replaces presidents, governments, or regimes.

³¹ Governments engage in official repression through formal decrees of censorship, suspension of the right to strike, suspension of constitutional guarantees, denial of rights of assembly, states of emergency, martial law, and war. In addition to studying formal decrees we could have looked at the number of political prisoners, deaths by torture or arrests, and injuries in confrontations with government security forces. The latter measures are so difficult to document that we confined this study to formal measures of government repression—decrees which appear in the official government records called Gazettas which have been published since the early 1900s. The index is grouped into five categories: censorship, interventions and nationalization, militarization, limits on organizational meetings and assembly, and others. The governments' stated reasons for each repressive measure leave a fascinating historical picture of the formal justification for their actions and the individuals, groups, and social classes against which the measures were directed. The data on repression which appear in this paper are based on a 100 percent review of the Gazettas for all Central American countries and Panama since 1930.

³² The intensity of such discontent might be measured by the number of people involved, or the frequency or seriousness of their actions. It is equally useful to try to establish the source of such discontent, which may be rural versus urban, and upper, middle, or lower class in its origins. It may come as well from the mainstream or from ethnic minorities, from men or from women. The data on social discontent are among the hardest to amass for Central America for the study period. There are no accurate records of the number of strikes, confrontations, combats, deaths, etc., and no systematic evidence about the origins of social action. In addition to using traditional historical sources this study developed a 28 percent random sample of Central American newspapers between 1930 and 1986. This sample selected two papers per week per country for every week in the study period and then performed a content analysis to identify numbers of stories, editorials, and opinion articles about strikes, demonstrations, sabotage, armed confrontations, and combats. Each action was classified according to its sector of origin (i.e., urban versus rural) and by the kind of group involved (political party, interest group, paramilitary group, etc.).

The newspaper material on social discontent proved to be the weakest of all the data sources developed. They required the most time and effort to collect and netted the least results because of the problems of censorship and changes in the formats of the newspapers themselves during the more than fifty-year study period. Precisely at the moments when historical sources indicate that there was the most discontent, evidence of the volume of that discontent disappears in countries whose newspapers were censored. While data on discontent are more plentiful in crisis periods than in periods of military rule, they appear to understate the volume and intensity of that discontent. The newspaper data are a better source for establishing who was protesting or fighting and what their social origin was.

³³ Nonprogrammed presidential turnover is a building block of Sanders's definitions of government instability and regime instability (fn. 10), chap. 3 and p. 62.

³⁴ One potential problem in using the population density data over a fifty-year period is that country ranks may change. For example, the population of Honduras could be less dense than Costa Rica in 1940 but more dense in 1980. An analysis of population data since 1950 shows that rankings for Central American countries did not change. El Salvador and Guatemala retain the highest densities, Costa Rica and Honduras medium, and Nicaragua and Panama low.

³⁵ Central American ethnic populations are of two types: on the one hand, Indians of Mayan and other stocks populating northern Guatemala, parts of El Salvador and Honduras, and Panama including its islands, and on the other, Negro groups which settled the eastern coastal areas of Central America.

³⁶ See World Bank, World Development Report 1985 (London: Oxford University Press, 1986), 216Google Scholar.

³⁷ The two periods of discontinuity require special explanation. From 1935 to 1943 the world economy grew due to the buildup related to World War II, but the Central American economies lost their European coffee markets due to the war and were subjected to stringent price controls in their remaining market, the United States. This is reflected in declining terms of trade, market diversification, and export prices compared with the previous period as shown in Lindenberg, Marc, “Central America: Crisis and Economic Strategy 1930–85. Lessons from History,” Journal of Developing Areas 22, No. 2 (1988), 155-78Google Scholar. However, Central American data for 1935 to 1943 can really be grouped in two periods: 1935 to 1937, when prices, growth, and terms of trade begin to recover along with the world economy, and 1938 to 1943 when markets close, prices drop, and growth plummets. In the second countercyclic period, 1944 to 1949, the world economy contracts during the post—World War II recession, while the Central American economy grows spectacularly but temporarily, due to a reopening to European markets and the liberation of price controls. However, by 1948 and 1949 its decline mirrors that of the world economy.

³⁸ Rapport (fn. 23) generates some interesting hypotheses about why Central American economic cycles appeared to be more weakly related to political instability (see Table 3) than world economic cycles (Table 2). Rapport's content analysis of Central American cable traffic, which appears in U.S. Department of State, Foreign Relations of the United States, ig^-1^40, Vols. 1 and 2 (Washington, DC: U.S. Department of State, 1980), indicates that shifts in U.S. policy toward Central American governments might have been sufficient either to override the political instability which normally accompanied economic crisis (1934–1943) or to exacerbate political instability in spite of economic growth (1944–1949). For example, between 1934 and 1943, a period of economic collapse in Central America, U.S. diplomatic recognition of Central American military governments was more ready than in the previous or subsequent periods. In fact it recognized all new governments without hesitation, while from 1930 to 1934 it recognized only 18% without hesitation and from 1944 to 1949, 50%. The United States might have been more concerned with reliable Central American allies against the Axis powers in World War II than with their tendencies toward dictatorship. In late 1944 the United States altered its criteria for recognition of new governments in Latin America. Cable traffic reflected active support for democratic openings and elections, a new sympathy in U.S. policy circles for the end of colonialism, and the enunciation of a new policy of U.S. recognition of newly independent, democratic governments.

³⁹ While it would have been useful to look at accurate data on the volume of overt expressions of social discontent in the 28% sample of Central American newspapers, the data proved to be unreliable because of changes in newspaper format and reporting style (see fn. 32).

⁴⁰ Space limitations prohibit the presentation of the data for the other six periods.

⁴¹ Two new military rulers who emerged through coups in the late 1940s in El Salvador (Osorio) and Honduras (Galvez) joined Nicaragua's General Anastacio Somoza Garcia, who had survived the instability of the previous period. Another military leader, General Jacobo Arbenz of Guatemala (elected in the previous period), was joined by civilians Anulfo Arias of Panama (installed after the incumbent president died in office) and Utillio Ulate of Costa Rica (elected after the revolutionary junta sponsored a return to civilian rule after the revolution of 1948).

⁴² Honduras, Gazetta Offcial, February 3, 1956.

⁴³ Honduras, Gazetta Oficial, January 5, 1960. In El Salvador the military school was founded (Gazetta Oficial, June 24, 1955).

⁴⁴ Costa Rica, Gazetta Oficial, January 6, 1960.

⁴⁵ Guatemala, Gazetta Oficial, all of 1954–1955.

⁴⁶ Guatemala, Gazetta Oficial, July 16, 1954.

⁴⁷ Guatemala, Gazetta Oficial, August 20, 1954.

⁴⁸ Guatemala, Gazetta Oficial, September 11 and 30, 1954.

⁴⁹ Guatemala, Gazetta Oficial, June 6, 1955.

⁵⁰ The only notable exception is the coup by Omar Torrijos in Panama in 1968, which established his rule for fifteen years.

⁵¹ Costa Rica, Gazetta Oficial, October 29, 1962.

⁵² Costa Rica, Gazetta Oficial, October 26, 1962.

⁵³ Guatemala, Gazetta Oficial, November 1963.

⁵⁴ Rank order correlations between % indigenous population, population density, % urban population and political instability, repression, and social discontent were all statistically insignificant with one exception. There was a positive (.80) rank order correlation between % indigenous population and social discontent during economic crisis. The lack of consistent income and land distribution data prohibited their inclusion in the analysis.

⁵⁵ See the sources cited in fns. 15 and 16 for more detail.

⁵⁶ Costa Rica's PQLI score gained 21 points between 1950 (67) and 1980 (88). This was the largest gain of any Central American country for the period.