Balancing On The Brink: Rationality, Revolution And Military Insurrection In Spain And Chile

BALANCING ON THE BRINK: RATIONALITY, RADICALISM, AND MILITARY INSURRECTION IN SPAIN AND CHILE1 Steve Snow Wagner College Journal of Political and Military Sociology Vol. 26 No. 2, Winter 1998. Note: This is a pre-publication version of the final, published article.

1.0 INTRODUCTION Why would politicians knowingly increase the probability of a military coup? At times, political incentives lead them to institute policies that redound to their harm (Kaufman and Stallings, 1991; Shepsle and Weingast, 1984; Weingast et al., 1981). To what extent such incentives can explain actions that place politicians’ own lives in danger, however, remains to be seen. I examine this issue with reference to the archetypal examples of leftist governments clearly threatened and then attacked by their domestic enemies: the Spanish Popular Front of 1936 and the Chilean Unidad Popular (U.P.) of 1970-73. Both pursued a surprisingly dangerous course after being greeted with loud military plotting. They could have reduced the threat of a coup to near zero by repressing the militant working classes, seeking full parliamentary authorization for all controversial policies, and subduing their leftist allies. Instead, they presided over massive strike waves and workplace occupations, reforms of questionable legality, and strident incitements to root-and-branch revolution by their extremist coalition partners. Why did Prime Minister Manuel Azaña in Spain and President Salvador Allende in Chile allow those conditions to develop which they knew would provoke the armed forces and their rightist supporters? Later in their rule, why did they not take those steps the military demanded as a condition for avoiding insurrection? To pursue these questions, I investigate the explanatory scope of rational-choice theory (RCT), which can help us explain the first, but not the second.

2 To account for the puzzling actions of the Spanish and Chilean politicians, it is necessary to examine the political constraints under which they ruled. Both governments faced bitterly hostile rightist opposition, a militant and well-organized working class, and depended on political support from revolutionary leftist parties. Both the Popular Front and the U.P. were the second stages of reformist processes begun by previous governments: the Spanish Republic’s “reformist biennium” (1931-33) and Chilean President Frei’s “Revolution in Liberty” (1966-70). These administrations achieved some legislative successes, but their central legacy was a mobilized, radicalized working class and peasantry disillusioned with reformism and eager for socialism. Yet the Popular Front and U.P proposed relatively modest programs. Amnesty for political prisoners, agrarian reform, and public works made up the Popular Front agenda. In Chile, the U.P called for wide expansion of the public sector of the economy and acceleration of agrarian reform.2 It was difficult to maintain these limited programs, and government leaders soon implemented more radical measures due to pressure from the left and from below. In each case, the governing coalition was split into reformist and revolutionary wings, where the reformers were politically dependent on allies to their left who denounced incrementalism and incited the workers and peasants to dismantle the capitalist system. In the face of such disruptions, which the right and the military harshly decried, the reformists knew establishing the “order” their opponents called for would redound to their political harm; meeting the demands of militant labor and the far left, on the other hand, safeguarded their political bases. Without abandoning the reformist agenda, it was impossible to allay the rightist parties’ complete opposition. To reach out for votes from centrist parties meant to water-down the government’s policies, leading to a corresponding loss of support from the extreme left. Therefore, neither government enjoyed the option of quashing their allies and the revolutionary masses, then replacing the resulting loss of votes in the legislature by those of the center and right. The choice, therefore, was between forsaking their proposed legislation, or maintaining broad leftist political support and thereby running the risk of provoking the armed forces. The Spanish and Chilean moderates’ refusal to

3 repudiate the extreme left and repress the workers and peasants was, in short, politically realistic, and therefore can be seen as rational--up to a point. In the last days and weeks, politics should have mattered less than avoiding an imminent military attack, and in that period we must view Azaña and Allende as irrational. Political incentives thus can account for the seemingly irrational decisions that brought Spain and Chile to the brink of catastrophe. RCT is less useful, however, in explaining the civilian leaders’ decisions in the unfamiliar and tense atmosphere immediately preceding military insurrection. Using these findings, I evaluate the different circumstances under which scholars have argued one can expect rational action. Such assessments are necessary to improve the rational-choice approach as a useful explanatory tool. 2.0 DETERMINING THE DOMAIN OF RATIONAL-CHOICE THEORY Researchers need a means to select possible explanatory approaches, not faith in their chosen method. The characteristics of the phenomena we analyze should determine this selection. When considering the appropriateness of RCT explanations, therefore, one must pay close attention to the conditions under which we can expect individuals to act rationally. By RCT I mean those theories that assume egoistic agents who act to satisfy their stable and ordered preferences, and recognize the alternatives and consequences. In this paper I employ a minimal definition of rational political action: actors seek first to remain alive, and second to remain in power; losing office will always be preferable to losing life. There is much current criticism of the rational-choice approach (e.g., Cook and Levi, 1990; Green and Shapiro, 1994; Smelser, 1992). Perhaps the warmest of the prevailing debates concerns the usefulness of RCT in supplying novel explanations of empirical phenomena (Green and Shapiro, 1994). Smelser (1992:400) identifies “theoretical degeneration” as a problem that exists across academic boundaries. From psychoanalysis to Marxism, theoretical formulations tend to decay into a drastic “indeterminacy. . .[i.e.,] the capacity to explain everything and hence nothing.” In the case of RCT, when one does not limit the range of assumed motivations, “it becomes possible to posit maximization of something in any conceivable situation” (Smelser, 1992:403).

4 To describe all behavior in terms of “maximization” is not a useful exercise, if one assumes that the purpose of theory is to advance knowledge by persuasive explanation. RCT is not out of the woods, however, when analysts assume actors possess a limited range of motivations. Applying a more circumscribed RCT “wherever the theory seems to work” is a form of confirmatory bias that does not allow for falsification, according to Green and Shapiro (1994:45). A more useful method is to specify in advance the theory’s limiting conditions and relevant domain, use them to specify possible tests, and thereby permit results counter to the theory’s predictions. This would involve abandoning the “universal” aspirations of RCT and “treat[ing] both maximization and rational calculation as variables rather than postulates” (Smelser, 1992:404; cf. Ferejohn and Satz, 1995). This approach, which Green and Shapiro (1995:264-5) term “segmented universalism,” assumes that rational-choice explanations work in some situations but not in others, and sets its sights on clearly identifying and analyzing each. In this paper I examine the usefulness and limitations of RCT in explaining the events in Spain and Chile, and employ these findings in the investigation of the circumstances in which the rational-choice approach is most applicable. Green and Shapiro (1995:267), Kelley (1995:101) and Taylor (1995:225-26) describe situations in which they believe actors are most likely to behave rationally. There are differences between these hypothetical conditions (to which I return later), but they are generally similar. As a starting point for my analysis, I use the conditions Green and Shapiro specify: “Rational choice explanations should be expected, prima facie, to perform well to the extent that the following five conditions are met:”

5 (i) the stakes are high and the players are self-conscious optimizers; (ii) preferences are well ordered and relatively fixed...; (iii) actors are presented with a clear range of options and little opportunity for strategic innovation, (iv) the strategic complexity of the situation is not overwhelmingly great for the actors...and (v) the actors have the capacity to learn from feedback in the environment and adapt. As the authors note, only the process of empirical inquiry will indicate the appropriateness of this schema (Green and Shapiro, 1995:267). In this paper I use the five conditions they offer, first, to identify an appropriate application of RCT, and second, to attempt to explain apparently irrational behavior using this approach. While these conditions characterized Azaña and Allende’s entire rule, their choices in the last months were not rational. (While I focus on these leaders’ choices, I also more broadly consider those of the governing left in both nations.) These decisions helped lead to their loss of power, and, in the case of several politicians--including Allende himself-- their deaths. This is not empirical territory that would appear conducive to RCT explanations; these cases are interesting and instructive in part because they represent situations in which political actors put their careers and lives in danger. Tsebelis (1990:120) argues that rational politicians may “commit political suicide”; this paper investigates the extent to which rationality can account for steps toward literal self-destruction. The circumstances Green and Shapiro describe depict well the situation Azaña and Allende faced. The stakes were extremely high: the actors confronted the destruction of democracy, civil war, and consequently great personal risk. It was clear to both that the military and its rightist allies posed a serious threat.3 Second, I assume that the politicians’ most important goals were staying alive and staying in office, in that order. It is reasonable, I believe, to assume that these preferences were well ordered and stable, and I will not linger on this point. Third, the politicians confronted sharply limited choices: to abide by or reject that which the armed forces and rightist groups decreed as the acceptable scope of political action. Fourth, the

6 complexity of the circumstances was not great, as the actors faced a relatively simple, if difficult, choice. Finally, they did have the opportunity to learn and adapt, albeit under severe time constraints. That is, politicians recognized that their refusal to employ repression was heightening the chances of armed conflict. These cases appear to be of the type where one can expect rational action, according to Green and Shapiro’s analysis. Yet we must still formulate a supplementary account that is consistent with RCT and that can explain the key choices made by Azaña and Allende. My argument in this regard focuses on the institutional restraints they faced. Scholars have not yet persuasively explained Azaña and Allende’s most enigmatic decisions. Existing studies of the fall of Spanish and Chilean democracy, among which one can discern four central approaches, are often highly partisan. Most grotesque is the assertion that the military stepped in to prevent a coup by the “communists” (e.g., Arrarás, 1968; Moss, 1973). From the opposite end of the political spectrum, others hold that disaster was the result of a naive, erroneous faith in a peaceful transition to socialism (e.g., Bookchin, 1980; Sweezy and Magdoff, 1974). In more balanced accounts of each government’s key choices before military insurrection, the moderate right stresses the politicians’ irresponsible partisanship (Falcoff, 1989; Payne, 1993), while the moderate left (mostly silent on this issue), emphasizes the difficult situation in which the governments found themselves (DeVylder, 1976; Preston, 1978). Finally, and unfortunately clearly in the minority, are those more analytical approaches that investigate, without apparent political biases, unexplained theoretical or empirical puzzles related to the two cases (e.g., Comín and Martín, 1984; Kaufman and Stallings, 1991). It is to this last group that this paper seeks to make a contribution; the analytical school has not yet addressed the issue of the Spanish and Chilean leaders’ possible motivations, much less from a comparative perspective. Several observers have briefly noted the similarities between the two cases (Landsberger and McDaniel, 1976; Malefakis, 1976; Morlino, 1981). Only Landsberger and Linz (1979), however, provide a comparative analysis. These authors note that a classic arms race preceded the final confrontation in each country. The left, to protect itself from the armed right, began to

7 acquire weapons. This led to more intense collection of arms by the right. Neither side desired a more dangerous opponent, neither were irrevocably committed to violence, yet their actions led to those results. One can view this strategic interaction as a Prisoners’ Dilemma game, where leftist and rightist militias’ conduct was rational, yet led to a Pareto-inferior outcome. Using different terminology, the participants themselves shared this interpretation. Only days before the Chilean coup, a former presidential candidate wrote to the Army Chief of Staff that, As in the tragedies of Greek classical theater, all know what will happen, all do not wish it to occur, but each one does precisely what is necessary to provoke the unfortunate outcome (quoted in Valenzuela and Valenzuela, 1976:vii). The approach of civil war in Spain and Chile is thus not incompatible with rational foot soldiers on each side. To what degree we can explain in rational-choice terms the behavior of those in government has not been explored, however. The first step is to show that it is reasonable to expect rationality under the circumstances in which Spanish and Chilean politicians found themselves. To this end I apply Green and Shapiro’s five conditions for rational action to the particular social and political environments in which the elected leaders made their fateful decisions. 3.0 EXPLAINING POLITICAL PROVOCATION: THE PATH TO THE BRINK We should look for rational behavior where actors face clear choices and high stakes, in not excessively complex circumstances, and with the opportunity to learn from feedback. Elected in February 1936, the Popular Front confronted the unpleasant yet distinct options of repressing or allowing relatively free rein to the ebullient Spanish workers and the extremists who encouraged them.4 The consequences of the choice could not have been higher, as it was obvious that many on the right threatened democratic institutions. The military, supported by fascists and monarchists, repeatedly warned that they would tolerate neither democracy if accompanied by social unrest, nor reforms instituted by questionable means. The opponents of

8 the government were quite vocal about their views, and the leftist politicians would have ample opportunity to learn from responses to their actions. Before the establishment of the Spanish Republic, armed forces newspapers consistently asserted that “disorder” and revolution would inevitably accompany a transition to democracy (see, e.g., La Correspondencia Militar, 8-12 April 1931). Editorials repeatedly declared the military’s willingness to end the perceived threat. La Correspondencia Militar (19 May 1931) stated that in order to fend off the leftist menace, society must behave as “organisms invaded by microbes: eliminate them.”5 A coup was clearly among the options. Ejército y Armada (2 October 1931), for example, warned when Spain desired to “modify the regime, or change it, the Army will be at its side.” Some time later, Marte (28 November 1933) stated the military would not protect Spanish democracy. Even the more moderate conservatives refused to commit themselves to democratic institutions; their difference with the ultra right was that the latter openly declared their willingness to use force. In a prophetic warning, one financial journal referred to a nineteenth-century coup when it warned that the Republic should maintain order and discipline, “or there would come in its moment and time another General Pavía who would impose normality” (El Financiero 17 April 1931). A central fear was that leftist reforms would not follow parliamentary and legal channels. In a typical editorial, El Economista (4 July 1931) declared that the government must “carry out as its first duty the reestablishment of tranquility, and then in an ambience of calm can undertake, through legal channels, the legislative reforms.” (See also El Economista, 18 April 1931; 11 July 1931; El Financiero, 3 July 1931; La Semana Financiera, 1 May 1931.) Upon the election of the Popular Front in 1936 the government became aware of plots of rebellion in the military (Payne, 1993:279; Preston, 1978:179). It was clear that failure to control the workers and the extreme left would increase these dangers (Jackson, 1965:205-6; Payne, 1993:314-15). The army and the far right warned that only by repressing popular militancy could the Popular Front prevent military intervention (Robinson, 1970:254 and passim). The more moderate opposition made clear that any government reforms had to closely follow legal channels.

9 The Chilean government also confronted distinct, uncomplicated options, high stakes, and had the opportunity to adapt to the loud, threatening feedback of its opponents. Allende could either limit its reforms to those passed by the legislature, or utilize workers’ direct action to advance the U.P. program over the objections of its opponents. Taking the second option would increase the hostility of the right and the armed forces, which Allende knew was a considerable hazard. After the Army Chief of Staff was assassinated following the presidential election of 1970 in an attempt to prevent Allende from coming to power, he recognized that the Chilean extreme right and elements of the armed forces comprised a clear threat to democracy (Allende, 1973:58). When asked about military uprisings, Allende said “confrontations will occur,” and declared he expected “reactionary violence...we know that they are going to break the rules” (Debray, 1971:97-100). Public revulsion at the assassination allowed Allende to gain Christian Democratic (PDC) support for his election in Congress (required of those presidential candidates lacking a majority of the popular vote). In return for their votes, the PDC insisted that Allende agree to a series of constitutional amendments pledging him to uphold democratic norms. The statutes provided the explicit rules that Allende would have to follow if the armed forces were to remain out of politics, insisting, for example, on “the full maintenance of the rule of Law” (quoted in Alexander, 1978:127). On the other hand, if the far left were to tempt Allende to violate the guarantees, the military would have a clear rationale for intervention (Sigmund, 1977:120). One author has termed the statutes as the “Damoclean sword (whose embodiment was the armed forces), [which] was to remain suspended” over the heads of the governing left (Roxborough, 1976:198). Thus the U.P. knew the conditions under which armed opponents would allow it to operate. The extreme right (rather hypocritically) warned that it would not tolerate illegality by the U.P. The fascist Patria y Libertad, for example, declared “if the government moves away from the Constitution by a millimeter, we will use force” (Marshall, 1970:5). During the entire 1970-73 period, much of the rightist propaganda was a denunciation of the government’s allegedly illegal deeds (Hinkelamert, 1976; Knudson, 1984). The question of legality was tied to that of public order, and members of the administration acknowledged that

10 losing control of the masses meant sacrificing the allegiance of the army--which in turn implied civil war (Garcés, 1972, 33). In this overview of the political situation at the outset of the rule of the left in Spain and Chile, we can see that both governments operated under the conditions specified by Green and Shapiro. Both knew well what was at stake: the survival of democratic institutions. Both knew what their options were, i.e., what the right and the military considered to be the tolerable limits to political action. The strategic complexity of the situation was not great: comply with the guidelines their opponents specified, or go beyond them. Further, the politicians in each country were able to learn from--although they did not adapt to--the actions and pronouncements of their enemies, as I detail below. As Green and Shapiro would predict, one can employ a rationalchoice perspective to explain the initial decisions Spanish and Chilean politicians took, which can be seen as rational if we examine each political system’s institutional constraints and inducements. 3.1 Breaking the “rules”: political incentives. Despite the clarity of the dangers they faced, each government allowed to develop precisely those circumstances the military and the right explicitly condemned: reforms of questionable legality, social disorder, and incitements to revolution. In both cases, the institutional inducement formed by a political dependence on the far left led the moderates to countenance situations that brought them to the brink of conflict with the military. The Spanish Popular Front presided over extreme social unrest and politicians’ loud calls for revolution, in spite of the clear warnings. This can be seen as a coherent if very dangerous strategy chosen in response to the political realities of the period. The crux of the matter was the reliance of the moderate left on their extremist brethren. Azaña led the Republican forces of the Front, who were dependent in the Cortes (Parliament) on the votes of the Socialists, Communists and other “bolshevized” parties who refused to serve in the government. This meant that the revolutionary left often drove governmental decisionmaking, which the administration itself publicly recognized (Robinson, 1970:264; Bolloten, 1979:35). As one Spanish historian phrased it, “the

11 Republicans were required to pay for the victory which the Popular Front had given them” (quoted in Payne, 1993:287). Conservative forces saw the danger of such dependence shortly after the election. In the words of La Semana Financiera (28 February 1936), “the proletarian parties will try to push the left-republicans [to extremist policies]. . . the question that is on everyone’s mind is whether the left-republicans will proceed down that road, from concession to concession.” Events soon confirmed such fears, as the appeasing tendencies of the Popular Front were apparent from the start. In response to pressure from the streets and its allies, on several occasions the government took actions that served to galvanize the already hostile opposition. Soon after the Popular Front election, for example, the Ministry of Labor, responding to street demonstrations and appeals from the far left, quickly promulgated a radicalized version of the compensation promised for those convicted of political crimes. The Ministry termed its own measure “a policy of pacification” (quoted in El Economista, 7 March 1936), and business groups declared that it stemmed from a desire to appease the working classes, and thus signified “a suppression of all juridic meaning in Spain” (El Sol, 1 March; quote in El Sol, 10 March). The response of the Spanish Employers’ Federation was typical when it declared that the government “pays more attention to the extreme elements than to the conscientious citizens of good sense” (Labor, 28 March 1936). The many worker demonstrations after the election were nothing, however, compared to the wave of strikes and farm occupations of the spring and summer. From May to mid-July 1936, there were 719 industrial strikes, more than the total for any previous year (Malefakis, 1970:371 n. 31). The extreme left always encouraged and often directed the widespread labor militance. “Forward and seize!...The Azaña government will back us up” (New York Times, 16 March)--this quotation distills the advice the radicals gave to the Spanish masses. The government did indeed “back up” the workers. On March 3 1936, for example, a mass occupation of farms began that came to involve tens of thousands of peasants. The government responded by eliminating virtually all restrictions on agrarian reform in a desperate effort at appeasement. Unwilling to

12 use force to dislodge the militants, it instead legalized the seizures (Malefakis, 1970:368). Indeed, the Minister of Agriculture declared the pace of agrarian reforms “will not be detained, on the contrary, it will pick up speed” (El Sol, 14 April 1936). Politically, he could do little else. The entire premise of the Popular Front was a link between the moderate and extreme left. In other circumstances, given the danger from the right and the army, one would expect the government to crush the extremists. There was no repression because Azaña needed the votes of the far left. His “policy had become predicated on alliance with proto-revolutionary activity” (Payne, 1993:363; see also Malefakis, 1970:375). By the end of the summer, the policies of the Popular Front were, in the words of El Financiero (1 July 1936), nothing more than “the translation of the republicans’ notorious concessions to the marxists into legal formulas.” This was precisely the dynamic that the armed opponents of the regime had vociferously warned against. The institutional inducements the Chilean government faced can also account for many of the otherwise apparently irrational decisions it took. The centrist forces in Unidad Popular (i.e., the Communists and the reformist faction of the Socialists that Allende led) encountered within its own coalition parties, factions within parties and extra-parliamentary groups that rejected a democratic transition to socialism. Among the radicals were many members of the Socialist Party, the Movement of Unity Popular Action (MAPU), and the Movement of the Revolutionary Left (MIR). Members of the MIR and MAPU were not exceedingly numerous or powerful. Their loud and sometimes violent actions, however, often expressed many of the sentiments of the more important and influential revolutionary elements in, for example, Allende’s own Socialist party. These extremists mobilized the masses for occupation of farms and factories, greatly reduced the credibility of the government’s commitment to legalism, and put constant pressure on the moderates to abandon incrementalism. The left frontier of the U.P. did not have a clear border, and no one was ever certain which group of revolutionaries reflected government policy and which were outside the pale. Indeed, Allende was in a sense hostage to the far left. In an electoral agreement of 1970, he had pledged to consult continually all the parties and

13 movements in the coalition. Once the U.P. was in power, this had the effect of giving the far left a veto on, for example, a moderation of government policies in order to obtain centrist parties’ political support (Sigmund, 1977:90; 277; Davis, 1984:56; 404). Allende was not as enthusiastic as some of his supporters about takeovers of farms and factories. Yet even the moderate components of Unidad Popular recognized that their central aims of nationalization of industry and agrarian reform would likely fail unless the workers and peasants took the initiative (see quotes in Allende, 1973:30-31; Stallings, 1978:128; and Zammit, 1973:261). This political dependence helped lead to government acquiescence in direct action from below. In the first 18 months of the UP regime alone, for example, 1,700 land seizures took place, often with the encouragement of government officials (Loveman, 1976:266). Facing a recalcitrant legislature, to effect agrarian reform and nationalizations the government used laws dating from the 1930s that allowed takeovers of firms during labor disputes. Use of such legislation gave radicals incentives to encourage labor militance in order to provoke government intervention and control of firms (Valenzuela, 1978:62-3). Indeed, such popular pressures in the countryside and factories were often a more important cause of state intervention than government intent (DeVylder, 1976:138; Nove, 1976:56). The administration allowed the masses to engage in direct action because it faced an unpleasant choice: abandon its program and admit the failure of a gradualist transition to socialism--and thereby lose electoral support to the far left--or go forward with its help and that of the workers. The U.P. responded to political inducements and took a course of action that it realized would substantially antagonize its opponents. Despite the right’s repeated warnings not to step outside constitutional channels, the U.P. circumvented the resistant Congress. Yet to this point we can view moderate government leaders as trying to avoid both a showdown with their rightist enemies and a loss of support from their leftist allies. Reliance on the political support of the revolutionaries led Azaña and Allende to cross the line in the sand their opponents had drawn. It was a perilous course of action, but one explicable in terms of rational political choices. There is a great difference between the Spanish and

14 Chilean governments’ first stages and the last months of their rule, however. The circumstances in both periods were similarly conducive to rational choices--according to Green and Shapiro’s schema--and the political incentives remained the same. Yet in their final days, Azaña and Allende faced the choice of either violently repressing their followers and allies or hastening civil war. One cannot define as rational their selection of the latter option. 3.2 At the brink. As the end of democracy approached, the choices remained obvious and uncomplicated, if difficult, and it was clearer than ever that the governments were confronting the highest possible stakes. Even more than in the initial period of their rule, however, the Popular Front and the U.P. received clear feedback from which they learned that their policies were unacceptable to the right; I will concentrate below on this dynamic. An important aspect of those situations conducive to rationality is the ability of decisionmakers to learn from their previous choices. If this opportunity does not exist, actors have little to guide them, and observers should not expect rational action. In Spain and Chile, the moderates had the opportunity to learn and adapt based on their opponents’ responses to governmental decisions. It became abundantly apparent that only by acquiescing in the militaries’ demands could the administrations avoid insurrection. Yet while politicians learned what they needed to know about the consequences of their actions, they did not use this information to try to save themselves or democratic institutions. In the final year of its rule, the U.P. was not working in the dark: Allende was aware of which policies would improve his relations with the army and thus lessen the odds of a coup (Alexander, 1978:309). The government received clear responses from the armed forces, which repeatedly asked Allende to change his political course and control the far left (Davis, 1985:16566). The moderate leaders of the U.P. did not choose to heed these warnings, and took measures its opponents declared to be intransigent and revolutionary. This dynamic began after the inconclusive elections of March 1973, when the armed forces presented Allende with 14 conditions for further support. They included a clarification of firms to be placed under state

15 control, the release of some businesses recently expropriated, and the disarming of leftist paramilitary groups (Davis, 1985:146). Allende refused, and the military representatives left the cabinet, whereupon Allende requisitioned 43 additional businesses, defying the comptrollergeneral who had classified this action as illegal (Falcoff, 1989:280). The warnings to the government were not all private: in May 1973 began open declarations in the press of the danger of civil war (Alexander, 1978:312). As the country slid towards armed conflict, in July the Christian Democrats called for regulation of expropriation policies, the return of many factories under state control, and increased efforts to disarm civilians. The government declined this course; in August, the military tried again to get Allende to come to an accord with the PDC, making clear that “such an agreement, including a settlement of the strikes, could forestall a coup” (Davis, 1985:201). There would be war if the U.P. did not drastically alter its policies and violently repress its followers. The government had the information it needed in order to make the choice to avoid military insurrection. I have argued that one can explain Allende’s initial policies in terms of rational responses to political incentives. Yet within the context of RCT we cannot account for his unwillingness to take all steps possible to avoid a coup in the last months of the government. The U.P.’s desire to maintain the votes of its political allies cannot explain this refusal. While the support of the far left was still contingent upon the government’s uncompromising line, a rational politician would have broken with key allies to avoid civil war. Uppermost in Allende’s mind seems to have been his historical legacy, not the dangers he and his government faced,. According to one of his interviewers, “he was determined not to tarnish the image which he wanted to leave to history” (quoted in Sigmund, 1977:230). Allende would not “be a Gonzalez Videla” (Alexander, 1978:143). This referred to a Chilean Radical president elected in 1946 with Communist support, who later outlawed the party, instituted severely repressive measures and relied on centrist and rightist forces to continue governing. No Chilean leftist had forgotten Gonzalez’ betrayal (Davis, 1984:51).

16 Such motivations also seemed to play a crucial role in Spain, where the government faced quite similar conditions for the avoidance of civil war. Beginning in April 1936, there were talks aimed at reducing tensions by forming a broadly based government that would end the power of the leftist militias and the “anarchy” in the countryside. Indalecio Prieto, a Socialist moderate, was the most likely candidate to undertake the very difficult task of marshaling such a coalition, but he refused. Two factors seemed to have influenced him in this decision. First, Prieto recognized that the question of support for a bipartisan administration would split the Socialist party. In addition, he was concerned about possible historical judgments, and shied from playing the role of a “Spanish Briand” (Payne, 1993:324; Jackson, 1965:208-9). Briand, despite making a career as a labor lawyer, crushed a strike in 1910 by arresting and drafting strikers into the army. This reference to the French parliamentarian closely parallels Allende’s preoccupation with Gonzalez Videla. This response is puzzling given the context. As in Chile, the government operated in those circumstances Green and Shapiro argue are conducive to rational choices. As in the outset of their rule, there were clear, uncomplicated options and high stakes. To these circumstances we can add consistent and constant feedback from their opponents and others regarding the government’s previous decisions. In the late spring of 1936, the chaotic situation in the Spanish countryside and many towns grew worse and the dangers of war increased. Moderate political figures repeatedly warned of the inevitable violent outcome if the government did not quash social unrest and disarm extremists, but the Popular Front proceeded on its course. In May, for example, the National Republican Party proposed formation of a new government dedicated to repression of revolutionary violence, disarmament, and the prohibition of paramilitary groups. The National Republicans further stated that if the proposed government did not receive parliamentary support, it should suspend the Cortes in conformity with the constitution. These measures were unacceptable to the Socialists, and Azaña’s government, unwilling to split the Popular Front, rejected them (Jackson, 1965:216-17; Payne, 1993:345-6). Later that month, after Azaña’s election to the presidency, the new Prime Minister Casares Quiroga attacked

17 conservative politicians and declared that in order to placate the revolutionaries he would implement the Popular Front program more quickly (Rivas, 1976:233). Far from heeding the warnings and learning from its mistakes, the government declared it would move even farther to the left. By June, the situation had deteriorated substantially. One financial journal (Vida Económica, 30 June 1936) summed up the atmosphere thus: “neither the Government, nor the Parliament, nor the Popular Front mean absolutely anything. They don't rule. They who rule are those who inspire inconceivable strikes [and violence].” In response to this atmosphere, there were open calls in the Cortes for military insurrection. The leader of the monarchists, for example, denounced those in the Army who “would not be willing to rise up for Spain against anarchy” (Diario de las Sesiones de las Cortes, 16 June 1936). There could have been no clearer signal to the government that the end was approaching. In a final gesture, hoping to avoid the looming war, one centrist politician called for a “national republican dictatorship,” with support from both the left and the right (Bolloten, 1979:35-36; Payne, 1993:350). Again, however, the goal of party unity made such a course impossible (Jackson, 1965:216-17; Robinson, 1970:263). Maura (quoted in Bolloten, 1979:36) summarized the stakes, of which all were aware. Either the politicians “decide to subordinate their party interests to the supreme interests of the regime...or they resign themselves to watching it die in the agonies of a bloody civil war.” Azaña and other leftist moderates had learned which were the policies that would satisfy their opponents, yet they were unwilling to take the drastic, violent measures that might have prevented military rebellion. During their final days in power, Spanish and Chilean politicians seemed irrationally motivated by party loyalty and their place in history, even though they operated under those conditions argued to be conducive to rational action. First, the stakes were exceedingly high: the government leaders knew that they risked the end of democratic institutions, massive violence, and consequently great personal danger. Second, it is reasonable, I believe, to assume that the ruling politicians possessed the ordered and fixed preferences of remaining alive and remaining in office. Third, they confronted clear and limited choices: tolerate “direct action” by the

18 working masses and the far left, or violently repress these groups. Fourth, the circumstances were not terribly complex; actors faced a simple if difficult choice. Finally, they did have the opportunity to learn from previous decisions and adapt: there was considerable warning that the governments’ actions were heightening the military’s hostility, and the consequent danger of civil war. The preceding has focused on the politicians’ failures to avert disastrous military attack. Yet perhaps the Spanish and Chilean militaries would have rebelled regardless of government policy; perhaps their hostility made conflict unavoidable. It is indeed tempting to place blame exclusively on those who later established abhorrent dictatorships. We must recognize, however, that although segments of each military reacted to the election of the left by plotting its overthrow, the eventual institutional consensus on this issue was not inevitable. The Spanish and Chilean militaries’ crucial decisions were influenced by the social and political conditions each government presided over and helped shape. In this context it is useful to examine the official explanations for the coups. The Spanish and Chilean dictatorships’ accounts of their insurrections are often fabricated, yet they are regardless quite informative: in them we see that the warnings issued upon the election of the U.P. and Popular Front became the eventual justifications for revolt. The militaries and their apologists argued that their rebellions were pre-emptive strikes designed to prevent an imminent coup by the communists, which would have established, in conjunction with foreign powers, a totalitarian system (Chile. Secretaria General de Gobierno, 1974; Cierva, 1967:299). These fictions emphasize the armed forces’ hatred and suspicion of the extreme left. As we have seen, they had consistently warned of the dangers posed by the revolutionaries, and their official justifications for the destruction of democracy indicate an intense preoccupation with this perceived threat. The official apologias also assert that the military was forced to overthrow an “illegal” government in order to save the nation from “chaos.” These are echoes of the original rightist warnings to the newly elected leftist governments to maintain the letter of the law and control their followers and allies. In the Manifesto of Francisco Franco (reprinted in Cierva,

19 1967:660), for example, Franco declares “anarchy reigns in the majority of the countryside and towns; government authorities preside over, when they do not foment, the revolts. . . Revolutionary strikes of all types paralyze the life of the country, ruining and destroying the sources of wealth.” The Francoists complained of an “illegal” government that had suspended the constitution--which was in “total eclipse”--and accused the government of tolerating “revolutionary anarchy” (quoted in Cierva, 1967:660, 695; see also Jackson, 1965:233). Likewise in Chile, General Pinochet denounced “the illegitimate and anarchic [U.P.] government...The economic life of the country was in total disorder and prostration; the productive, labor and commercial activities were suffering the most complete anarchy” (Pinochet, 1983:19). The Chilean junta stated that the Armed Forces had overthrown “an unlawful government,” and emphasized that the dictatorship would not tolerate more “disorder or anarchy” (Chile. Junta del Gobierno, 1974:33, 36; see also Chile. Secretaria General de Gobierno, 1974:79-97 and passim). In the official justifications for the Spanish and Chilean military dictatorships, we see the same themes as in their warnings to the Popular Front and U.P: the extreme left, social “disorder” and strict legality. While one cannot know if the Popular Front and the U.P. could have avoided insurrection, the militaries’ often fanciful defense of their decisions indicate that they acted because their warnings were not adequately heeded. 4.0 ASSESSMENTS. Can we explain in rational-choice terms the Spanish and Chilean leaders’ refusal in the last stages of their rule to abandon ideology in order to avoid civil war? One could, perhaps, argue that they wished to “maximize” a type of historical legacy, and thus acted “rationally.” This is tautological, however: we can account for all behavior in terms of maximization of something, and if rationality is simply maximization, all behavior is rational. A more promising line of reasoning would center on the institutional constraints the politicians faced; I used such an account to explain in rational-choice terms the initial decisions of each government. Yet we have seen that this explanation can account for only a limited number of the key political decisions. Even with a reasonable chance of winning a civil war--such as existed in Spain-- a

20 rational politician would not risk life and limb when the option of fleeing the country was available.6 Finally, the available evidence indicates that other factors--party loyalty, concern for their followers, and an eye for their historical legacy--more convincingly explain the actions of Spanish and Chilean politicians. There are two plausible explanations why Azaña and Allende were unlikely to make rational decisions under the situations they confronted. First, the politicians faced the threat of military attack, which added a high degree of tension not present at the outset of their rule. As Holsti (1979) found in regard to international relations, such stress impairs the rationality of decisionmakers. Second, the decisions in the later stages of their administrations were not of the same type as those Azaña and Allende had made all their political lives. Instead of the accustomed political logrolling, these leaders confronted credible threats by armed opponents. Trying to avoid being pushed into civil war is not like trading votes, and the moderates of both governments were entirely unfamiliar with the former circumstance. That is, during these final stages of the Popular Front and Unidad Popular, there was a complete absence of the “repetition” and “choices that repeat themselves” that Kelley (1995:101) and Taylor (1995:226) argue are conducive to rational choices. Had the Spanish and Chilean politicians more experience with the sort of situation in which they found themselves, they might have behaved in a more rational manner. In any case, the conditions that Green and Shapiro propose are clearly insufficient to explain the Spanish and Chilean governments’ puzzling decisions. One of the findings of this paper, therefore, is that Green and Shapiro’s schema could usefully be supplemented with the condition of repetition Kelley and Taylor identify. Further research is necessary, of course, to determine whether “choices that repeat themselves” are indeed a necessary component of the circumstances under which RCT can be an effective explanatory tool. 5.0 CONCLUSION The aim of theory should be explanation that advances knowledge; RCT is currently under considerable criticism in this regard. Critics have charged that RCT practitioners either

21 arbitrarily restrict its explanatory domain (by applying their models “wherever they seem to work”), or resort to defining all behavior as maximization of some sort. More convincing application of rational-choice explanations necessarily implies the possibility of their falsification, and to this end, we must define rational behavior and state where RCT should hold. Several scholars have begun to map the theoretical circumstances of a “segmented universalism,” rejecting the universalist claim that actors make choices the same way in all circumstances, but not abandoning use of RCT as an explanatory tool. Evaluations of the different accounts within this approach have been lacking, however. In this paper I have provided a minimalist definition of rationality, delineated those areas in the cases under consideration where one can explain behavior in terms of rational choices given institutional constraints, and suggested reasons that help account for those situations where this course is untenable. In that regard, the assessments this paper offers concerning the boundaries of a “segmented universalism” should be useful in future attempts to employ RCT in explaining empirical phenomena.

ENDNOTES 1

An earlier draft of this article was delivered at the 1996 Western Political Science

Association meeting, San Francisco, March 14-16. For their comments and criticisms, I thank anonymous reviewers, John Keeler, Margaret Levi, Richard Sherman and Cheryl Wheeler. 2

To placate the revolutionary members of the U.P., the electoral program asserted that

these measures would “initiate” the movement to socialism. This apparently radical language should be seen in its political context. The centrist Christian-Democratic Party, for example, also called for the replacement of capitalism; and even the distinctly rightist National Party was unwilling to defend the capitalist system as such (Sigmund, 1977:95; Roxborough, 1976:207). 3

There were differences, of course, between and within the Spanish and Chilean

militaries. Both were divided on different issues, and the Spanish armed forces were generally

22

more hostile to democracy than their Chilean counterparts. For stylistic purposes I will speak of both institutions as unified wholes. 4

In my analysis of conservative and rightist views during the Spanish Popular Front, I

employ, among other sources, the quite under-explored Second Republic financial press and military periodicals (cf. Alpert, 1989; Bahamonde and Toro, 1981). 5

Translations from the Spanish are by the author.

6

Note that my minimal definition of rational action is not so restrictive as to exclude all

decisions to wage war. If the political support were very high, odds of success overwhelming and politicians faced no likely physical danger, war could be a rational choice. In the Spanish and Chilean cases, however, the politicians knew that the odds of victory were even at best (in Spain), and that war would place them at significant risk.