Taming the Tiger: Voting Rights and Political Instability in Latin America

This article discusses the relationship between certain institutional regulations of voting rights and elections, different levels of electoral participation, and the degree of political instability in several Latin American political experiences. A formal model specifies the hypotheses that sudden enlargements of the electorate may provoke high levels of political instability, especially under plurality and other restrictive electoral rules, while gradual enlargements of the electorate may prevent much electoral and political innovation and help stability. Empirical data illustrate these hypotheses. A historical survey identifies different patterns of political instability and stability in different countries and periods, which can be compared with the adoption of different voting rights regulations and electoral rules either encouraging or depressing turnout. At the time of Porfirio Diaz's seventh election as president of Mexico in 1910, Francisco I. Madero called a rebellion under the motto "Effective suffrage, no rlection." As he was leaving for exile in May 1911, Diaz reportedly said: "Madero has unleashed a tiger; let's see if he can control it." Madero was elected president with broad electoral support, but less than two years later he was assassinated, and Mexico plunged into two decades of revolution and civil war. More than 1.5 million people were killed; approximately 12 percent of the Mexican population.-(Meyer and Sherman 1979, 511) This article discusses the relationship between political regime stability and certain institutional regulations of voting rights and electoral participation in several past political experiences in Latin America. Political regime stability here implies only the absence of political conflicts leading to regime change, such as rebellions, coups, revolutions, and civil wars. The analysis focuses on electoral institutions with a significant influence on electoral participation, mainly the regulation of voting rights and registration procedures and the basic characteristics of electoral systems. The main hypothesis to be discussed is that in the past, many regime crises in Latin America were fostered by the shock of sudden introductions of broad suffrage rights under exclusive electoral rules. The sudden enlargement of the incumbent electorate with new, large groups able to develop and promote new political preferences may overthrow the previously existing political equilibrium, especially under plurality and other exclusive electoral rules producing a single absolute winner-a president and his party in congress-supported by only a minority of voters and many absolute losers. In contrast to the instability effects of broad voting rights with exclusive electoral rules, some significant degree of political stability can be achieved by a nondemocratic regime holding elections with restrictive rules, able to exclude a majority of the population from the electorate or to incorporate different groups of voters gradually into the existing institutional framework. Alternatively, political regime stability can also be achieved on the basis of the opportunities for sharing power and introducing smoother policy changes usually created by proportional representation and other permissive and inclusive electoral rules that have been widely adopted in the region during more recent periods of democratization. A basic assumption in the following analysis is that citizens' support or rejection of political regimes crucially depends on the type of winners they tend to produce. In particular, a single-winner, minority outcome is likely to provoke negative reactions against the institutional framework among the many absolute losers. The second part of this article will survey a number of regime crises fostered by rejections of political outcomes produced by sudden enlargements of the electorate under such restrictive institutional frameworks. Notorious cases include Argentina in 1829, 1930, and 1955; Colombia in 1857 and 1949; Mexico in 1835 and 1911; and Peru in I860 and 1968. In all these cases, as well as in other countries and periods not closely scrutinized here, the emergence of winners manufactured by restrictive electoral institutions provoked strong rejection and was followed by reactionary or military coups, civil wars, or dictatorships. In contrast to regime crises largely provoked by sudden changes in the size and composition of the electorate, such as those just mentioned, significant stability was achieved in a number of Latin American countries during several periods of the late nineteenth and twentieth centuries by regimes that, while they held regular elections, dramatically reduced mass participation and political competition. These include, in particular, certain periods with nondemocratic but election-based political regimes in Brazil from 1824 to 1930, Chile from 1822 to 1946, and Mexico from 1876 to 1910 and 1929 to 1988, as well as less durable experiences in other countries. These nondemocratic political regimes can be ranked from softer to harder, but all of them can be distinguished, in crucial institutional features, from dictatorial regimes that did not hold elections at all. In these cases, electoral demobilization and manipulation of voters whose participation could have threatened the existing political equilibria, especially workers, indigenous people, and recent immigrants, was produced not only by the requirement of some economic or literacy conditions to receive voting rights, but also by other formal and informal institutional mechanisms that could reduce political competition. Eventually, however, also in these cases, the introduction of effective universal male suffrage provoked some dramatic regime crises; for example, in Mexico from 1911 on (as alluded to in the epigraph), in Brazil in 1904, and in Chile on several occasions during the nineteenth century and until 1973. The arguments developed in this article can be put in relation to hypotheses and postulates previously presented in some classic political studies. In particular, Samuel Huntington, who focused on the role of modernization and economic development in political stability, notes that "the expansion of political participation," together with "the political backwardness of the country in terms of political institutionalization, make it difficult if not impossible for the demands upon the government to be expressed through legitimate channels and to be moderate and aggregated within the political system. Hence the rapid increase in participation gives rise to political instability" (Huntington 1908, $5). Taking a slightly different approach, the present article will emphasize the role of restrictive electoral institutions in the relationship between increase in participation and political instability. The potentially destabilizing effects of enlarging the electorate and the connection of this problem with an increasing interest in institutional engineering, especially in regard to electoral rules and procedures, were also grasped by Stein Rokkan in his seminal comparative studies of political life. He notes that, in Western Europe and the United States during the late nineteenth and early twentieth centuries, the bitter debates over the two issues [districting and electoral formulas] reflect fears and resentments generated through changes in the equilibrium of political power under the impact of mass democracy: the influx of new voters altered the character of the system and a great variety of stratagems were tried out to bring it back into equilibrium. (Rokkan 1970, 155). Nevertheless, Rokkan did not include Latin America among the cases under consideration, and both Huntington and he actually made the remarks quoted here in a rather speculative manner without much elaboration. More formally, Robert Dahl distinguishes two crucial political variables: the right to participate in elections and the liberalization of the political system (or "public contestation"), which can be assimilated with our attention to voting rights and the inclusiveness of institutional rules, respectively. Dahl, however, focuses mainly on the conditions for successful democratization and political stability, either by gradually introducing public contestation under limited voting rights (as in Britain), by enlarging voting rights while limiting public contestation (a model that would fit most of Mexico's twentieth-century experience), or by establishing fair institutional rules permitting pluralistic representation that can attract large social acceptance, the "polyarchy" (Dahl 1971, 4-9). To understand some other historical features in Latin American countries, by contrast, the following pages will identify alternative conditions prone to foster political instability. The analysis developed in this article draws on the distinction among three different strategies of voting rights, each of them producing different political consequences, which have been identified elsewhere in a historical and comparative perspective (Colomer 2001, chap. 2). They include the so-called Anglo model-referring mainly to the United Kingdom and the United States-which was based on a gradual allocation of voting rights to different minority groups through a slow, lengthy process of moderate reforms; and the Nordic model-corresponding to early developments in Germany and Northern European countries-which made the sudden enfranchisement of a very large electorate compatible with appreciable degrees of political stability, thanks to the introduction of proportional representation and other pluralistic institutional arrangements. In contrast to these two successful means, the so-called Latin model involved a sudden jump from a small electorate to universal men's suffrage under single-winner electoral rules provoking high electoral unpredictability and instability. Although this third framework has been applied to the analysis of a few cases in Latin Europe, the present article will expand on specific features of the problem in Latin America.1 The political and institutional implications of the analysis developed in the following pages may be diverse. From a conservative point of view, for instance, limiting the size of the electorate may be seen as an effective way to achieve some significant degree of political stability. For a revolutionary purpose, in contrast, the sudden enlargement of the electorate could be considered a favorable condition for producing radical political changes. This article pretends to illuminate the question for readers holding different normative values. It can also be held, however, that broad voting rights can be compatible with some high degree of political stability if the regime is organized with inclusive electoral institutions, such as proportional representation and absolute majority rule, able to produce encompassing winners with relatively large electoral and social support. The most recent experiences of relatively stable democracy in most countries in Latin America also seem to support this statement. The following pages first present a formal model in which the degree of political instability depends on the size and dispersion of the electorate. Second, they provide original calculations of proportions of voters in the total population for the larger Latin American countries for several periods since independence. In addition, recently available historical evidence about voting rules, which had not yet been submitted to systematic comparative analysis, is compiled and interpreted from the perspective provided by the model and discussion. MODELING ELECTORAL INSTABILITY The effects of enlarging the electorate with new groups of voters holding political preferences different from those of the incumbents can be analyzed with the help of some formal tools. Let us first assume that different political preferences, which can correspond to socioeconomic categories such as landlords, smallowners, artisans, workers, and so on, can be represented as different positions on a one-dimensional scale, such as the typical left-right axis, in such a way that different relative "distances" between voters' groups can be estimated. The assumption of a single dimension may not necessarily fit reality in some countries and periods; but it is used here as an analytical simplification, which is legitimate and useful because it compares only relative distances between groups of voters' preferences. Alternatively, it would have been possible to develop a two-dimensional model, which is also common in certain spatial analyses of political competition; but that would have produced analogous results and a more baroque and difficultto-read presentation. Let us also assume that for each composition of the electorate, different winning coalitions of voters can be formed. For instance, if the existing electoral rules require an absolute majority of voters to make a winner, the corresponding winning coalitions will always include the position of the median voterthat is, the voter having less than a majority of voters on each side of the political space. If, however, the winner can be made by plurality rule, the winning coalition may not include the median voter among its supporters. The set of all those positions in the political space that can form a winning coalition is called the win-set. The size of the win-set is a proxy for the unpredictability of the electoral outcome: the larger the set of potentially winning positions, the more unpredictable the electoral outcome will be. If the outcome of each single election is highly unpredictable, the series of outcomes at successive elections will be likely to create high political instability. We thus develop a simple formal analysis from the perspective that the larger the size of the win-set of a given electorate, the higher the instability of the outcomes that can be expected. In this approach, the argument about the effects of changes in the electorate on the degrees of political instability can be synthesized the following way. * First, the more "distant" the new voters from the initial winner, which will be catted status quo, the larger the win-set. This means, in other words, that the more the new voters' political preferences differ from those of the incumbent voters, the higher the political instability of the corresponding electoral outcomes that can be expected. It can be assumed, for instance, that industrial workers have preferences more different from landlords than from artisans, artisans have more different preferences from landlords than from smallowners, and so on. The formal model will allow us to make the more precise statement that the size of the win-set and therefore the degree of outcome instability tends to increase more than proportionally with the distances to the status quo introduced by new voters. With some degree of dispersion of voters' preferences, new winning positions can be located even at significantly more extreme positions on the political spectrum than the preference of the most extreme voter. This formal finding may give a well-founded analytical grounding to traditional fears that new suffrage rights could be a slippery slope to the unknown. These fears were expressed in the past as risks that permissive regulations covering voting rights granted to workers, immigrants, or alternative ethnic groups could become occasions of "hazardous," "threatening," "extremist," or "chaotic" electoral results. * Second, the more exclusive the electoral rule, the larger the size of the win-set and the higher the corresponding degree of outcome instability. This will be illustrated in particular by comparing sizes of win-sets in elections by the highly exclusive plurality rule with those in elections by the less exclusive simple majority rule. As analysis will show, the more exclusive the rule, the larger the set of potentially winning positions and the more unpredictable the electoral outcome will be. * Third, the effects of political instability can be reduced by a gradual enlargement of the electorate with new groups of voters producing successive new winners at different stages in the process. Also, the introduction of new voters with political preferences located on intermediate positions among the incumbent voters may reduce the size of the win-set and the corresponding degree of potential outcome instability. Even more, if different groups of new voters are located on opposite sides of the political space regarding the status quo, they could neutralize each other and confirm the size of the previous win-set or even reinforce the stability of the status quo. These formal findings may give well-founded analytical grounds to certain political strategies that favor a gradual introduction of new voting rights for different social groups in order to preserve some degree of political stability. They can also explain certain historical arguments about the "innocuousness" or even the potential "counterweight" effect of conceding new voting rights to some groups, especially certain fractions of peasants and bourgeois women, compared to the likely destabilizing consequences of enfranchising workers or recent immigrants. Figure 1. Enlarging the electorate --------------------------------------------------------------------------------------------------1.1. Three voters: A, B, C A B C !xxxxxxxx!xxxxxxxx!___!______ B* Note: Status-quo A. Pivotal median voter B. Distance AB = distance BB*. Win-set: A-B*: xxx 1.2. Five more dispersed voters: A, B, C, D, E A B C D,E !xxxxxxxx!xxxxxxxxxxx!xxxxxxxxxxx!xxx!xxxxx!_________ """""""""""""""""""""""""""C** C* Note: Status-quo A. Pivotal median voter C. Distance AC = CC*. Win-set from A: A-C*: xxx Status-quo B. Pivotal median voter C. Distance BC = CC**. Win-set from B: B-C**: """" 1.3. Five more dispersed voters with plurality rule (2/5) A B C D,E !xxxxxxxx!xxxxxxxxxxx!xxxxxxxxxxxxxxx!xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx! D* Note: Status-quo A. Distance AD = DD*. Win-set from A: A-D*: xxx 1.4. Five less dispersed voters: A, B, C, F, G A F,G B C !xxxxx!xx!xx! !_________ G* Note: Status-quo A. Pivotal median voter G. Distance AG = GG*. Win-set from A: A-G*: xxx -----------------------------------------------------------------------------------------------------------Let us start the formal analysis with the simplest possible example of three voters, which can be called A, B, and C, respectively, as presented in figure 1.1. The preference of A is assumed to be the initial status quo, as may have resulted, perhaps, from an extremely small previous electorate or just from the lack of elections. The reader can figure out, for instance, that A is a group of large landowners, B represents smallowners and shopkeepers, and C includes artisans and other modest taxpayers. Each group can be incorporated into the electorate as a consequence of different voting rights regulations, requiring, for instance, some amount of property, income, or taxes to be given suffrage rights-as typically happened in countries holding regular political elections during the nineteenth century. Figure 1.1 suggests that the successive groups hold preferences at increasing relative distances from each other, so that A is relatively closer to B than B is to C. This might reflect relatively high levels of social inequality among the groups. Different distances among groups, which can reflect social structures, values, and ideological opinions, can, of course, modify the scope of the results. But in qualitative terms, the present model can be generally valid to illustrate the type of relations developed between the variables under consideration. For the sake of simplicity, let us assume that each group represents a similar number of voters and that a winner can be made by absolute majority; in this case, by any coalition of two of the three groups. The three possible winning coalitions are AB, AC, and BC. The two coalitions including the initial winner, A, whose preference defines the status quo, will probably be unable to introduce any change, because the members of A will resist any move from the status quo to any other position away from their preference. Only the winning coalition BC can create new winners. But in this innovative coalition, the median voter, B, is a pivotal voter, able to limit the scope of changes and define the win-set around its preference. The new winning position can be located anywhere between the status quo and the point B*, which is found around B at symmetrical distance from it to the status quo A (distance AB = distance BB*), as shown in the figure. This means that B can prefer, instead of the status quo A, any point between the status quo A and B, and also any point between B and B*, because the latter will be located at a smaller distance from B's preference than the status quo. In the particular example in figure 1.1, B* is located somewhat closer to the status quo than to C, thereby denying C the opportunity to make its preference the winner. Thus the win-set encompasses all political positions between the status quo A and B*; in other words, the interval A-B*, as shown by the X's in the figure. Let us now compare this with the result corresponding to an enlargement of the electorate with two new voters, D and E, located an even greater distance from the incumbent voters, as presented in figure 1.2. They can represent, for instance, a group of previously disenfranchised industrial workers. (Two groups with the same preference are added now in order to keep an odd number of voters able to produce a winner by simple majority without a tie). This new configuration of the electorate with five voters comprises a number of winning coalitions including A (ABC, ABD, ABE, ACD, ACE, ADE), which give A the opportunity to resist any move from the status quo; and some winning coalitions including B (BCD, BCE, BDE), which will confirm the win-set A-B* corresponding to the previous electorate with three voters. Only the coalition CDE will be able to introduce changes. Now the median voter C becomes the pivotal voter, able to accept any position between the status quo A and C* (AC = CC*). Observe that the new win-set A-C* may include potential winners located beyond the most extreme voters' positions, in this case D and E. If, for instance, by enlarging the electorate with new voters, the distance between the most distant voters is multiplied by two, the size of the new win-set may be multiplied by three or four. The intuition behind this is that because C, as well as the new voters D and E, holds very distant preferences from the initial status quo A, these voters can be ready to accept even highly distant winners on the other side of the spectrum because these will be located at a relatively smaller distance from them than the very distant initial status quo. The consequence is that the new winset, A-C*, is huge. Some of the older voters, A and B, can really fear that the introduction of new voters, such as D and E, may produce new outcomes at enormous distance from their preferences, thereby creating a feeling of extreme potential instability. Now assume that, in this enlarged electorate with five relatively dispersed voters, the election can be won by simple plurality rule, requiring not an absolute majority but only more votes than any other alternative. Let us operationalize the winner by plurality at 40 percent of the vote, which means that, in the example, any coalition of two out of five voters can make a winner. (A 40 percent winner is a fairly realistic estimate for elections by plurality rule in the real world; it is consistent, for instance, with the presumption that two other candidates split the remaining 60 percent of the vote.) The analysis can be followed in figure 1.3. Analogous to the previous examples, now all winning coalitions including A will confirm the status quo, all remaining winning coalitions including B will confirm the win-set A-B*, and all remaining winning coalitions including C will confirm the win-set A-C*. But the innovative winning coalition DE will produce an extremely large win-set A-D*, encompassing potential winners located at extremely distant positions from both the status quo and the most extreme voter, thereby creating overwhelming unpredictability of the outcomes and potential instability at successive elections. Almost any position can win by plurality rule, including a large variety of minority-supported alternatives, which may provoke widespread dissatisfaction among the numerous losers. We see, thus, that with a more exclusive electoral rule, such as plurality, the size of the win-set and the corresponding degree of outcome instability increase compared to the less exclusive majority rule. Using Diaz's metaphor, we might say that a feeling can spread that "the tiger" is loose, he can attack anywhere, and his claws can really hurt. Now let us consider some alternative hypotheses that could produce smaller win-sets and, therefore, higher political stability. Let us first assume that the enlargement of the electorate with new groups of voters is enforced gradually. As a first step, a relatively complex but still reduced electorate formed by A, B, and C is established. Let us assume that this electorate endures a period sufficiently long as to create a new winner; say, for instance, the median and pivotal voter B, whose position becomes the new status quo. From this new status quo, an enlargement of the electorate, for instance with voters D and E, will have lessdestabilizing consequences than the same enlargement would have if it were adopted suddenly while the initial status quo were still located at A, as assumed in our previous discussion. Now, with the new electorate of five dispersed voters, A, B, C, D, and E, as represented in figure 1.2, the median voter C becomes again the pivotal voter, able to define the win-set around his preference. But the win-set now will be defined in comparison with the status quo B (not the previous status quo A), so it will be the interval B-C** (BC = CC**), which is smaller than the win-set A-C* previously found. Likewise, as, at successive stages of a gradual enlargement of the electorate, new pivotal voters emerge, able to produce new status quo points away from the initial status quo A, the corresponding win-sets will be relatively reduced, and therefore the electoral results will be more predictable and less unstable than those produced by a sudden enlargement of the electorate. Now consider the case in which the initial electorate ABC is not enlarged with new voters located at more dispersed positions, as previously assumed, but with other new voters, which can be called F and G, whose preferences are located somewhere in between the incumbent voters; for instance, between A and B, as shown in figure 1.4. Applying the same kind of analysis as in the previous cases, the reader can see that F and G are now the pivotal voters, because it is not possible to form a majority winning coalition without including their positions, and the new win-set is defined as the interval A-G*. As can be noted, this new win-set is even smaller than the one created by the initial electorate ABC, as shown by figures 1.1 and 1.4. So, in this case, the introduction of new voters actually reduces the size of the win-set and, therefore, can be expected to increase the degree of outcome stability. The extent of this kind of moderating effect will depend on the exact position of the new voters, but in general it can be stated that the introduction of new voters not increasing the dispersion of the electorate preferences will not increase and may reduce the size of the winset; that is, the dispersion of the potential winners and the corresponding political instability. This kind of argument in favor of moderating effects of new intermediate voters was used during the nineteenth century, for instance, by some strategists who held that if voting rights were conceded to them, some peasants would vote as their patrons did, and most women as did their fathers and husbands. Particularly moderating effects can also be produced by enlarging the electorate with new voters located at opposite sides of the previous win-set, as would be the case, for instance, if the electorate ABC were enlarged with a voter D at a more extreme position on one side, as in figures 1.2 and 1.3, and also with another new voter H located on the other side; that is, at the left of the status quo A (not included in the figure). The new "centrist" and pivotal voter would now be B, so that there would be no possible winning coalition without B, and the corresponding win-set would be the small interval A-B, as can be easily checked out. In this case, it could be said that the introduction of the new voter H would have a "counterweighting" effect to the introduction of the new voter D, resulting from the significant distance and opposite directions of their respective preferences to the previous winners. By all the means mentioned in the last few paragraphs, it may seem possible to political strategists to "tame the tiger"; to reduce his moves within a cage, so to speak. VOTING RIGHTS IN HISTORY The model just presented can be illustrated with a certain amount of empirical observation. The following is only an exploratory analysis, which should be further developed and refined. Although it takes great benefit from a number of single-country stories-the typical product of the existing historiography-it proposes an explicitly comparative, theorydriven analysis, which is more characteristic of social science literature. The basic features of regulation of voting rights and electoral procedures, including voting rights requisites, registration of electors, voluntary or compulsory vote, and secret or open ballot, for the six largest countries of Latin America are presented in table 1. Note that data are presented for election years, not including years in which legal provisions were approved but not enforced. Also, estimated percentages of voters in the total population are given at the far right column. The table looks not only at legal regulations giving voting rights to certain categories of citizens, but also at their practical consequences in promoting electoral mobilization or demobilization, thus encouraging or depressing turnout. Table 1. Size of the Electorate in Latin America ----------------------------------------------------------------------------------------------------------Country/ Voting rights requirements Electoral regulations Voters as % of Election years total population ------------------------------------------------------------------------------------------------------------Argentina 1826 Men, age 20 or married, citizens Voluntary registration, voluntary vote, open ballot, indirect elections 5-1