From Authoritarianism to Democracy: The Unfinished Transition in Mexico.[1]

 

 

 

Alberto Diaz-Cayeros

UCLA and Stanford University

Political Science Department

Stanford University

Stanford, CA 94304-2044

acayeros@polisci.ucla.edu

 

Beatriz Magaloni

Political Science Department

Stanford University

Bldg. 160, Room 161A

Stanford, CA 94305-2044

magaloni@leland.stanford.edu

 

ABSTRACT

 

This paper discusses the logic of institutional design in the Mexican transition to democracy. Our main argument is that electoral rules facilitate party dominance through two mechanisms: electoral rules disproportionately reward existing majorities and, at the same time, discourage potential majorities to form. More specifically, the rules reward parties that can win a majority of the vote in single-member districts; but at the same time, reward minority parties with seats from multimember districts, mitigating Duvergerian incentives to coordinate behind a single challenger. Our argument is hence about the effects of electoral rules in a dynamic setting. In the short-run, seats from multimember districts have benefited opposition parties by significantly reducing entry costs; in the long-run, however, these seats help sustain dominance by discouraging coordination among opposition parties.

 

 

How are democratic institutions crafted when there is a significant unbalance of political forces in society? Are institutions designed so as to cement the ex-ante distribution of power? If so, why do the less powerful agree to play by institutions they know are biased against them? Can democratization take place in a system where a dominant party incumbent repeatedly wins the elections? This paper seeks answer these questions by discussing the logic of institutional design in the Mexican transition to democracy. The Mexican transition is peculiar in that no overarching pact took place, but democratization proceeded in the electoral arena, where opposition parties gradually conquered political office while negotiating with the government and the incumbent dominant party, the Partido Revolucionario Institucional (PRI), incremental changes to the electoral rules. There has been a gradual institutionalization of the electoral process - namely the establishment of ever more credible controls against electoral fraud – and at the same time the creation of electoral rules that maximize the chances for the ruling party to maintain dominance. In this paper we sidetrack the question of how procedures were crafted to effectively count votes (i.e., to limit electoral fraud) and instead focus on changes in the rules of the game for the translation of votes into seats.

The conventional view unduly stresses the disproportionality of Mexico’s mixed plurality-PR electoral rules, and how they can work at “manufacturing” majorities for the PRI, as one of the key mechanisms that sustain its dominance. But the PRI has actually never needed to “manufacture” a majority for itself. It sustains its dominance simply because it is the largest electoral force. The crucial question is why an alternative majoritarian opposition coalition hasn’t been able to form in order to dislodge the ruling party. Indeed, we will show that PRI’s chances of retaining power in Mexico’s more competitive environment crucially depend on the split of opposition parties and how the existing electoral rules facilitate a non-coordinating equilibrium.

Politicians have crafted institutions in such a way that party dominance becomes a self-enforcing equilibrium. Electoral rules facilitate party dominance through two mechanisms: they disproportionately reward existing majorities and, at the same time, discourage potential majorities to form. More specifically, the rules reward parties that can win a majority of the vote in single-member districts; but at the same time, reward minority parties competing for seats from multimember districts, mitigating Duvergerian incentives for small parties to coordinate. Our argument is hence about the effects of electoral rules in a dynamic setting. In the short-term, movements toward greater proportionality in the electoral system benefited the opposition by reducing entry costs. In the long-term, however, they have helped the PRI by dividing the opposition camp.

This unique combination of rules, crafted in a thirty-year time span, did not result from an “authoritarian imposition” on the part of the ruling party. It is not obvious that in a liberalizing political environment an authoritarian incumbent can craft institutions so as to retain advantage and hold unto power, since the challengers would regard those institutions as illegitimate, and hence refuse to play the democratic game under them. We argue, instead, that the institutional changes necessary to produce rules that protect the dominance of the PRI resulted from partisan compromises. In particular, the PRI followed a “divide-and-rule” strategy, offering short-term electoral benefits to a fraction of the opposition, in exchange of rules that would eventually make the incumbent party even more difficult to dislodge. In doing so, the PRI could successfully discourage opposition parties coming together behind an overarching agenda of institutional reform.

            The paper unfolds as follows. In the first section, we discuss the logic of institutional design focusing on the establishment of institutional constraints for opposition coordination. The next two sections after that then provide evidence of the arguments, demonstrating the effects of different sets of rules on opposition coordination and the maintenance of party dominance. We proceed by discussing first, that coordination of challengers is possible, as witnessed in state and municipal executive contests. We then turn to showing how challenges to the PRI fail at the national level because within a system of mixed electoral rules comprising single member and multimember districts, the later dominate both voters and parties’ strategic calculations. The last section concludes with some thoughts on the future.

 

1. Institutional Constraints for Opposition Coordination

 

As summarized by Cox (1997), the literature on dominant party systems argues that coordination failure plays a key role in sustaining dominance. This argument can be traced back to Riker’s (1982) explanation for the failure of Duverger’s law in the case India. In his view, the law failed because the central location of the Congress party kept its challengers divided at opposite ends of the political spectrum. Such explanation assumes that competition occurs on one dimension, and that the dominant party successfully occupies the median location, such that other parties cannot credibly challenge it there. Sartori (1976), Laver and Schofield (1991) and Pempel (1990) all provide a similar argument, which is based on the dominant party’s central location, to explain dominance in Japan, Italy, Israel and Sweden. Presumably, the same argument could be used to explain the dominance of the PRI: the ruling party, that is, faces competition form two major opposition parties, which are positioned to the left and to the right in the issue space, the PRI located in the center.

This argument is not persuasive for the Mexican case however, because empirical evidence suggests that party competition is multidimensional. On the left-right ideological dimension, the right-wing Partido Acción Nacional (PAN) and the left-wing Partido de la Revolución Democrática (PRD) stand at opposite sides of the spectrum, but both parties stand next to each other on the political dimension, sharing an interest in the democratization of the rules of the game and the PRI’s defeat.[2] Employing survey data, Magaloni (1994) shows that these dimensions are also reflected in the structure of voter’s preferences. By constructing complete preference orders from respondent’s binary comparisons among the three major parties, she found that some are ideological voters, who rank the PRI second, and whom would not be willing to set their ideological differences aside in order to defeat the PRI. However, there is a high percentage of voters, whom she calls radical or potentially strategic opposition voters, who rank the PRI third and therefore would be willing to cast a strategic vote in order to defeat the incumbent. Multidimensionality implies that, while ideological differences among opposition forces complicate coordination, ideology alone cannot account for the split of the opposition vote. There are potential gains of forming electoral alliances between the PAN and the PRD. These parties and their supporters share a common interest, namely defeating the PRI, and this political dimension is highly relevant in Mexican society (see Dominguez and McCann 1996; Magaloni 1994 and 1997; Moreno 1999 for analyses of how the political dimension determines the vote choice).

True, generally speaking it is easier to coordinate parties that share similar ideologies or that are “connected” in some ideological space (Axelrod, 1970). Successful coordination among opposition parties in Mexico has, in fact, usually included ideologically connected leftist parties. In the 1988 presidential elections the left-wing opposition parties came together in the Frente Democrático Nacional to support the candidacy of Cuauhtémoc Cárdenas, which until today has represented the most serious electoral challenge to the PRI in presidential elections. After the 1988 election, the Cardenistas created the new left-wing party, the PRD. Unlike other candidate-centered organizations of the past, the PRD was able to survive as a political party in subsequent elections and, moreover, to attract virtually all the leftist vote, leaving the other left-wing alternatives as tiny electoral parties. Hence, coordination has succeeded in fighting the excessive fragmentation of the left that characterized Mexico’s party system in the past, but if has failed to generate a credible majoritarian opposition coalition against the PRI.

Table 1 presents the results of presidential elections in Mexico from 1964 until 1994. The table shows the percentages of votes received by the PRI in each election and votes received by the first and second loser. There is a consistent tendency for the PRI to lose votes. The 1988 elections, however, mark a departure from the pattern of secular decline by increasing the rate of decay. These elections were marked by popular discontent due to the economic recession and by the Cardenista split from the PRI. From 1964 until those elections, the PAN was the only relevant opposition party, placing second in all the elections but the one in 1976, when it was unable to agree on a candidate to run for the presidential race. The fractionalization of the left is evident if one looks at the percentage of votes received by the second loser in those elections. On average, the largest leftist party won less than 2% of the vote and there were often four or five left-wing parties competing.[3] After successful coordination in 1988, when the left obtained 30% of the vote by cross-endorsing Cardenas, almost all these parties unified into a single one, the PRD. Today, the PRI faces serious competition from only two opposition parties, the PAN and the PRD, which have been unsuccessful at putting together a majoritarian alliance.

 

[Table 1 about here]

 

Incentives to coordinate are also shaped by election specific dynamics and institutions (Cox, 1997). As made explicit by Duverger’s Law and a more recent restatement, the smaller the magnitude of the district, the larger the incentives to coordinate. With the smallest district magnitude, when M=1, as in plurality single member one-round races, the incentives to form alliances are biggest, unleashing tendencies towards the consolidation of an M+1 candidate race, that is, a two party system.[4]

Coordination may fail, however, due to election specific dynamics or voter’s probabilistic assessments about the parties’ chances of winning. Cox (1997) argues that voters might fail to coordinate when they calculate that 1) only one party has real chances of winning or 2) the second and first loser are so tied for second that they do not know which one to discount. In either of these cases, probabilistic assessments lead voters to cast a sincere vote, rather than a sophisticated one, where all votes are coordinated behind the first loser and the winner(s).

The last column in table 1 shows the measured proposed by Cox to test for the second possibility, namely failure to coordinate due to virtual ties. The second-to-first loser vote ratio (SF) is used to test whether the parties vote shares are such that strategic voting, in equilibrium, will in fact reduce the number of parties to M + 1. The parties vote shares fall into two categories: Duvergerian equilibria (in which strategic voting undercuts completely the support of the trailing party) and non-Duvergerian ones. According to Cox, under Duvergerian equilibria, the SF ratio will be near 0 (when second losers are far behind first losers so that strategic voting reduces their support almost completely) and under non-Duvergerian equilibria, near 1 (when second losers are so nearly tied first losers that they do not lose their support due to strategic voting). An SF ratio near 1, that is, reflects an exception to Duverger's law explained by various kinds of ties such that no serious challenger becomes the unique victim of strategic desertion and more than M + 1 viable parties therefore survive[5].

As shown in table 1, while the PAN was virtually the only challenger in 1964 and 1970, by the 1980s the failure to coordinate among opposition parties did not come from virtual ties between the PAN and the left. In the 1988 elections, the coalition of the left was the strongest challenge to the PRI, with the PAN trailing far behind the first loser. In the 1994 elections, the order was reversed, PAN jumping to a clear second place after the presidential debate that took place in the middle of the campaigns. But in both cases the SF ratio is distinctively different from the Duvergerian (SF close to 0) or the non-Duvergerian (SF close to 1) equilibria.

What then accounts for the split of the opposition? Institutions play a crucial role at hindering coordination. Table 2 presents a summary of the transformation of the electoral rules in Mexico between 1960 and 1996. The first column shows the rules for each of the elective posts in 1960; the second column provides some of the details of the reforms these rules underwent; and the last column shows the current rules. The next section provides a more thorough examination of the effects of these rules. Here we want to highlight that the general trend in Mexico’s electoral reforms has been to keep the single-member districts in place while adding multimember district races.

Thus, until 1964 the Lower Chamber was elected through a variable number of single-member districts increased according to population estimates with each census. Two senators from each of the states were elected under a binomial “formula” in statewide races where voters could cast one vote.  Governors, local legislators and municipal presidents were all elected in single-member plurality races. Thus, elections everywhere were single-member district races[6]. In the years of party hegemony, the electoral system created an insurmountable threshold for minority opposition parties. Virtually all the elective offices in the country were controlled by the PRI. The consecutive electoral reforms have gradually added multimember district races without eliminating the majoritarian nature of the system. The resulting set of rules creates strong incentives for opposition parties to compete against each other as long as they can’t seriously attempt to win in single-member districts. As the last column in the table shows, under the current system single member districts are retained for two fifths of the Chamber of Deputies, half of the Senators, Governors and the Majors.

 

[Table 2 about here]

 

One could argue, however, that rules for the composition of the assemblies should not significantly discourage coordination in the presidential race. Shugart and Carey (1992) show that in presidential systems, the timing of presidential elections with respect to congressional elections is essential for determining the number of parties and ultimately the incentives to coordinate. All Latin American countries have a combination of plurality-win presidential races with different forms of PR congressional elections. "When presidential and congressional elections are held at the same time, the presidential election imposes a single-seat nationwide district over the assembly districts" and two major parties tend to emerge. (Shugart and Carey, 1992:207).[7] The authors argue that concurrency works at reducing the number of competitors toward two, regardless of the assemblies’ rules, because in presidential systems parties tend to gradually merge into larger coalitions capable of winning the most important election, the single-seat nationwide district presidential race. While such logic holds in gubernatorial races in Mexico, which are often concurrent with local legislatures, it does not prevail at the national level. The reason for this anomaly is related to the peculiarities of a dominant party system.

As long as opposition parties do not perceive that they can seriously attempt to challenge the dominant party in the executive, they focus their electoral resources in the assembly races, where they can attempt to establish themselves as relevant players in the system by winning some of the multimember seats. Rather than imposing a single-seat nationwide district over the assemblies, concurrency under a dominant party system implies that assembly multimember districts contaminate the presidential race.

Although the incentives to coordinate are slim, a further restriction came from the rule for cross-endorsement established in the electoral reform of 1993, as revealed in the last row of table 2. After the 1988 experience, the PRI changed the rule that allowed parties to nominate common presidential candidates while keeping a separate identity for the congressional and senatorial races taking place concurrently. The new rule implies that if two parties want to nominate a common presidential candidate, they must also share candidates in all the congressional and senatorial races - all 628 other races at stake. If parties intend to coordinate, they must do so much before the start of the campaigns. If opposition parties wanted to nominate a common presidential candidate, they must first craft alliances for each of the races taking place simultaneously and sacrifice their internal dynamics for the distribution of seats within the party. This clearly involves great transaction costs, although the regional distribution of support for opposition parties makes this a controversial issue probably in only half of the single member districts. The current electoral legislation establishes no public campaign finance to the coalition as a whole, but only to the largest partner (which means foregoing around 40% of the public money, with strict limits for private finance). The coalition must also produce a joint electoral platform, notwithstanding ideological differences among the partners.

This rule obviously protects the PRI from risky coordination among opposition forces in the single most important race, the presidency. But the PRI did not impose it. The PAN voted for it in a congressional coalition with the PRI. The PAN supported a vote against cross-endorsement in presidential races because in the past presidential alliances had benefited the PRD, not the PAN. The 1993 electoral reform also offered important side-payments to the PAN. They eliminated the "self-certification" by the electoral colleges, granting the Federal Electoral Institute (IFE) with the faculty to certify the electoral results. The 1993 reform also changed the rules for the composition of the Senate. The 1993 rule implied that three senators where going to be elected by plurality and a fourth one given to the second largest party in the state. As it will become apparent in the last section, the new rule for selecting the Senate largely benefited the PAN over the PRD, since the right-wing opposition party was the second largest party in most states.

Heading for the 2000 presidential elections, the opposition proposed eliminating the existing rule for cross-endorsement so as to open up the possibility for an overarching opposition alliance, one which presumably would include PAN and PRD. The Senate, where the PRI enjoys a comfortable majority, blocked the reform. Back in 1993, the PAN might have failed to calculate that drafting a rule that makes presidential alliances so costly would turn out to be so useful for the PRI. The PAN approved the existing rule because it was shortsighted. In the longer-run, the PRI won because the likelihood of opposition coordination is now even slimmer. This underscores the logic of institutional design in Mexico: the PRI has been able to play the interests of PAN against PRD by offering short-term payoffs to either party[8].

The following sections provide evidence to these arguments. We will first show that coordination is possible, and that when it takes place, it generally works against the PRI. We then explore the national electoral arenas, showing how the rules summarized in table 2 discourage coordination and determine electoral payoffs.

 

 

 

 

2. Opposition coordination: state and municipal races

 

            The election of state and municipal executives in Mexico is carried out in direct elections under an “at large” single member district, namely, the state or the municipality.[9] If coordination occurs, one should find two candidate contests in these races. In Table 3 we present the electoral results for all gubernatorial elections taking place between 1993 and 1999 in Mexico. The table also reports which party won, the effective number of parties in each race, whether a coalition was formed, which party previously held office in the state, and the second-to-first (SF) loser's vote total ratio proposed by Cox (1997).

                       

[Table 3 about here]

 

            Gubernatorial elections in Mexico have become extremely competitive. Between 1993 and 1999, the PRI lost 11 out of 32 gubernatorial races. It should be noted that in the elections taking place prior to 1993, the PAN had won three gubernatorial races (Baja California, Guanajuato, and Chihuhua) and only in Chihuahua it was not reelected. Only in eight races did the PRI obtained an absolute majority of the vote. Most significantly, the table shows that the ruling party tends to lose where the opposition coordinates. In all the states where the PAN won (Baja California, Guanajuato, Jalisco, Nuevo Leon, Queretaro, and Aguascalientes), the PRD virtually doesn’t exist (average SF ratio of 0.14). In these states, that is, a bipartisan pattern of competition prevails.[10]

            Generally speaking, where the PRD wins, the PAN does exist as a significant competitor but it is nonetheless subject to strategic defections. The PRD won Zacatecas and Baja California Sur by nominating candidates that split from the PRI after losing the party’s nomination. Thus, many of the votes to the winning governors came directly from the PRI and its traditional bases of support. In these states PAN’s support was weak, and although there was no full coordination, opposition voters probably tended to vote strategically for the PRD’s nominee. The evidence of strategic voting in the Mexico City 1997 election is quite persuasive: the PAN started in first place and the PRD in third; by the end of the race the order was reversed, presumably due to strategic voting.

            Among the 11 states that the PRI lost, two of them have been cases of opposition coalitions. In Tlaxcala, the coalition was connected, involving only left-wing parties. In Nayarit, all opposition parties, including PAN and PRD, crafted a winning coalition. Hence some form of coordination among opposition parties seems to be a necessary condition for defeating the PRI. Forming a coalition is no guarantee of winning, as the partial coalitions of the left in Guerrero, Chihuahua and Nuevo Leon reveal.[11] But as the SF ratios demonstrate, the PRI loses where the trailing opposition party is left with virtually no support, either due to local bipartism, to strategic defections, or to coalitions formed ad-hoc.[12]   

            Conversely, the PRI tends to win were the opposition splits. This is particularly so in those cases where the PRI wins despite being a minority party (Yucatan, Michoacan, Campeche, Colima, San Luis Potosi, Sonora, Durango, Veracruz, Oaxaca, Sinaloa, Quinta Roo, Guerrero, and Estado de Mexico). Except for Yucatan, San Luis Potosi and Guerrero, the PRI was a minority because the opposition split (the average SF ratio was 0.60). Of course that the ruling party might also win because there is little opposition presence (Morelos, for example). However, as can be appreciated in the table, there are very few “hegemonic” states left. The data hence suggest that the PRI’s chances of retaining power in Mexico’s more competitive environment crucially depend on the split of opposition parties.

            Table 4 summarizes the results of gubernatorial races. It presents the average SF ratio and number of parties depending on the identity of the winning party and on whether the PRI won as a hegemonic party, defining hegemony as winning with a margin of more than 25% of the vote. As mentioned earlier, it is clear from the data that, at least in gubernatorial races, PRI hegemony is a thing of the past. The larger the SF ratio, the biggest the coordination dilemmas are. SF ratios are systematically larger were the PRI wins. In these cases, coordination appears to fail because both opposition parties are equally strong such that voters do not know which one to discount. SF ratios are systematically smaller were the opposition wins. It should be noted, however, that SF ratios are smaller were PAN or a coalition wins, than were PRD wins. SF ratios appear to always be smaller where the PAN wins because the PRD does not exist in all states.

 

[Table 4 about here]

 

            A similar pattern is found within the 2,400 municipalities. These elections take place every three years and are often not concurrent with gubernatorial ones. Municipal races tend to be less competitive than gubernatorial races. As table 5 shows, the PRI won more than 60% of municipal races between 1993 and 1995. The table overstates, however, the PRI’s absolute level of support simply because most of these municipalities are small rural localities. The opposition has overwhelmingly won in densely populated urban municipalities, including many of the state’s capitals. The table also reports the average SF ratio, the effective number of parties, and the margin of victory. Among the PRI victories -and in sharp contrast to gubernatorial elections-, the distinction between hegemonic and non-hegemonic races is quite relevant at the municipal level. Only the indicators for non-hegemonic races are really meaningful, since in the hegemonic races the opposition barely exists, with the PRI winning with more than 90 percent.

           

[Table 5 about here]

 

            In terms of the SF ratios at municipal races, there is no difference statistically speaking between cases where the PAN and the PRD won. But the SF ratios for PRI victories are higher, suggesting, again, that there is a greater tendency for a failure at coordination where the PRI wins. However, in contrast to gubernatorial races, the SF ratios are low even where the PRI wins, suggesting a high degree of coordination into two party races in municipal elections. The SF ratios are virtually unimodal, which means that only two relevant parties tend to survive in each municipality, the PRI and either of the major opposition parties.[13]

            The findings of subnational elections discussed in this section suggest that voters in Mexico are sophisticated enough to cast their vote for the candidate with the greater chances of winning. Such coordination around only one challenger candidate leads to the prevalence of Duvergerian equilibria where the PRI can be defeated. The presence of bipartyism in local and state elections is not reflected in national elections though. As the next section discusses, this is not because patterns of bipartyism at the local level get aggregated into multipartyism at the national level, as in the case of Canada. In sharp contrast with the findings of this section, single-member district congressional elections at the national level are not characterized by strategic voting or coalitions. Two-way races, where the PRI can be defeated are seldom observed in congressional elections. In senatorial elections, as we will see, the rules create strong incentives for the major opposition parties to compete against each other in an attempt to be the largest minority in the state.

 

 

 

3. Failure to coordinate: the national party system

           

            Having discussed electoral results in the local and state executive races, this section now discusses the effects of electoral rules at the national level. In particular, we examine institutional change in the Lower Chamber of Deputies and the Senate. As previously mentioned, coordination failure is more prevalent at the national level.  We argue that electoral rules for the translation of votes into seats in the legislature play a crucial role in dividing the opposition vote for the national executive. The logic of institutional design has been threefold: 1. Preserve the majoritarian nature of the original single-member district races such that the system tends to disproportionately reward the existing majority. 2. Reduce entry costs to opposition parties by adding multimember seats. 3. Minimize the risk of losing the PRI’s majority control of both chambers by establishing “safety vote thresholds” above which a majority of seats was artificially manufactured. This safety threshold, as it will become apparent below, was never used in practice because it was not until the 1997 midterm election that the PRI officially lost the absolute majority of the vote.  

Overall, electoral reforms have tended to preserve the existing balance of forces emerging from previous elections: protect the PRI’s dominance while accommodating a slowly growing opposition divided camp. Reforms for the translation of votes into seats continuously took place from 1960 until 1996.[14] Throughout most of that period, the PRI was particularly successful at crafting institutions to its advantage.

 

3.1 The Chamber of Deputies

 

The composition of the Chamber of Deputies has been the focus of the most hotly debated electoral reforms taking place in Mexico during the last 25 years. From an assembly in which deputies were elected only through single member districts, it has been transformed into a two- tiered body with 300 deputies resulting from single member district races and 200 members elected through closed lists in five regional districts. Strictly speaking, those two hundred seats are not compensatory seats, since a party winning the absolute majority of the single member districts still receives a fraction of those seats and those districts do not ensure proportionality of the chamber as a whole as in the German system.

The Chamber is characterized by a set of rules that, we argue, have retained majoritarian districts in order to sustain the advantage of the incumbent PRI, while at the same time opening spaces to opposition parties unable to win in the single member districts. The opening of such spaces, however, has prevented the coordination of opposition parties around single challengers in single member districts, so that the PRI keeps on winning most of the majoritarian races, but even more importantly, the tendency for opposition not to coordinate in the legislative races contaminates the concurrent presidential race.

Changes to the rules for the composition of the Chamber of Deputies were publicly justified on the grounds of enhancing political plurality. While it is clear that new parties were able to enter the Chamber thanks to those rules, they also weakened in the long run the prospects for established opposition parties to win single member districts. Rules for the translation of votes into seats were first changed in 1963, when so-called “party deputies” were first introduced (see table 2). The mixed system was established with the 1977 electoral reform, which greatly reduced the barriers to entry of new parties into the system. The reform drew 300 single-member districts plus 100 compensatory seats, distributed in a rather proportional manner only to minority parties, namely those winning less than 60 single member districts.

 The 1986 reform increased the number of seats coming from multimember districts to 200, that would be assigned to any party obtaining between 1.5% and less than 50% of the vote. In addition, the 1986 established a so-called “governability clause”, which automatically gave the largest party the majority of seats if its vote was above 35% but below 51%.  The rule also established an upper bound of 350 seats to the largest party. But the most long lasting effect supporting PRI dominance of the 1986 reform was to establish that voters could cast only one vote in the single-member district races that would automatically be counted for the allocation of seats coming from the regional multimember districts. Thus, split-ticketing voting was no longer possible, which has momentous implications for the coordination of the opposition.

In 1989 the rule for the translation of votes into seats was changed again such that a perfectly proportional system was established if no party obtained more than 35% of the vote. If, however, the largest party obtained between 35% and 60% of the vote, a rather disproportional system was established that assured an absolute majority to the largest party. If the vote share of the largest party fell between 60% and 70%, the rule gave to this party an equal seat vote ratio. Finally, in 1993, a new reform established a limit of 8% overrepresentation to any party, which in practice meant that if the PRI failed to reach that threshold, the existing rule could no longer manufacture a majority for the incumbent. The inelasticities in the translation of votes into seats of the previous rule were kept. It should be noted that the PAN voted, together with the PRI, to approve these last electoral reforms. [15]

While there are several accounts of the electoral reforms and their effects, particularly focusing on the rules to manufacture majorities in case a party fell short from them,[16] in this section we seek to highlight three issues that have received less attention. First, we point out that the rules to manufacture majorities were actually not used, because the PRI had the necessary majority in both votes and single member district victories to retain control of the Chamber of Deputies. Second, although the overall degree of disproportionality in the system is not significantly different to that of democratic systems, what is unique about the Mexican case is how the effects of disproportionality were distributed among parties, benefiting the party more capable of winning in single-member races, which happens to be the PRI, and punishing the second largest party in those races, namely the PAN[17]. Third, unlike other analyzes, we highlight the degree to which the system created strong incentives for opposition parties not to coordinate to credibly challenge the PRI in the single-member races. In this respect, the prohibition to divide votes among the single-member and multimember races is particularly important. Through the prohibition of split-ticketing, the rules significantly increased coordination costs among the opposition since electors can’t strategically cast a vote for the strongest contender to the PRI in single-member district races without, at the same time, sacrificing a vote for their most preferred alternative in the multimember races. The result is an utter lack of opposition coordination in the single-member district races.

Table 6 presents several indicators to analyze the effects of the rules of the lower chamber on its composition. The third column presents a measure of the magnitude of overrepresentation obtained by the PRI in the Lower Chamber of Deputies, namely the difference between seat share and vote share. Generally speaking, the rules have overrepresented the PRI. In 1960, its overrepresentation was 6%, a rather low number considering that the system was purely majoritarian.[18] This number reflects the degree to which the system was hegemonic: only the PRI got votes, thus it was not utterly overrepresented despite controlling virtually all the Chamber’s seats. Through the introduction of “party deputies”, the PRI was actually “giving away” seats to some parties that, at least with the official vote tally, could not surpass the vote thresholds required by the legislation, let alone win single-member races. In fact, between 1964 and 1976, the PRI was slightly underrepresented in three out of the five elections despite controlling virtually all the seats in the Chamber. After the mixed system was introduced in 1977, the tendency has been to continue to overrepresent the PRI, though not by a large margin. Of noticeable importance are the figures for the 1994 and 1997 elections. Overrepresentation got closer to the two-digit range, reflecting the inelasticities of the most recent electoral formulas. It is important to notice that as elections became more competitive, the electoral rules have tended to overrepresent this party more. Another important point is that, as other analyzes have revealed, the second largest party, the PAN, is generally underrepresented. However, as rules for multimember seat allocation changed and elections became more competitive, underrepresentation of the PAN has been reduced or even reversed. This implies that between 1960 and 1985, electoral rules originally benefited the PRI and small parties at the expense of the second largest party, increasing incentives for opposition fragmentation. Since 1988, when the left consolidated and the elections became competitive, the PAN fares much better in the system and it is not systematically punished to the benefit of the tiny parties. Between 1988 and 1997, that is, the electoral rules benefit the PRI at the expense of all the opposition parties.

 

[Table 6 around here]

 

Despite the PRI’s overrepresentation, the existing mixed system has been better for the opposition than a purely majoritarian system. The fourth column in the table reports the expected deviation from proportional representation under a single-member plurality system. It tells the magnitude of overrepresentation the ruling party would obtain had a mixed system not been introduced, namely the difference between percentage of seats from single-member races won by the ruling party minus its total vote share. Clearly, all of the reforms significantly reduced entry cost to opposition parties. The reason is that, as column two shows, the opposition has been extremely unsuccessful at winning single-member races. Throughout this period, the PRI has won the overwhelming majority of plurality races, the only exceptions being 1988 and 1997, when it won only 75% and 68% of the seats coming from single-member districts. Thus, in the short-term, the mixed electoral system has significantly reduced entry costs to opposition parties. But the existing system, we argue, creates perverse incentives in the longer-term, since it discourages coordination among these parties, which as shown in the previous section, is essential to dislodge the ruling party.

The reason as to why the existing system discourages coordination can further be explored by analyzing which effect dominates in the mixed electoral system, the majoritarian or the PR one. Presumably, if the majoritarian effect clearly dominanted, opposition parties would face stronger incentives to coordinate. We follow Taagapera and Shugart (1989), who argue that when the number of compensatory seats is smaller than the expected deviation from proportionality, the majoritarian effect dominates over the PR one. The table reports, in column 6, the effective number of compensatory seats occupied by the opposition, namely its total number of seats not coming from single-member races over the total number of seats. This number must be subtracted from the number in column 2 to assess which effect dominates in the system. A positive sign means the majoritarian effect overshadows the PR one. A striking result, as reported in column 4, is that Mexico’s mixed electoral rules are quite balanced: in some years the majoritarian effect dominates over the PR one or the other way around but in no year did one effect overshadow the other one by a large margin. Interestingly, in the most competitive elections – 1988 and 1997 – the PR effect tended to dominate, which implies that the payoffs not to coordinate have been highest precisely when the vote for the PRI fell the most.

To further explore coordination failure, we followed Cox (1997) in calculating SF ratios for the elections since 1988.[19] The graph below only reports results for PRI victories in the 1997 elections.[20] The graph shows an utter lack of coordination among opposition parties (the PAN and the PRD mostly, although in some districts a fourth party effectively competes) in those races where the PRI won. There is no bimodality, which would be observed with Duvergerian (SF=0) and non-Duvergerian equilibria (SF=1). Such pattern was exhibited in an even more overwhelming manner in previous elections, where the PRI won almost all the single member districts.

 

[Insert figure 1 around here]

 

The persistence of this pattern of non-coordination suggests that this is not just an instance of an out of equilibrium outcome, but that the current mixed system provides incentives for small parties to care more about victories in multimember districts, where the incentives for fragmentation are stronger, than single member ones. The composition of the chamber gives a greater weight (3/5) to single member districts, and votes cast on those races anyway count towards the allocation of multimember seats. This implies that when the margin of victory is slim, the value of a vote cast for a candidate in a single member district is greater than the value of a partisan vote in any district, since the single member district vote still counts for the multimember ones.

If all parties have equal chances of winning the congressional elections, they should devote most of their efforts to the single member district races. But the fact is that they do not. The PRI devotes significant financial and political resources to single member districts, while the PAN and the PRD only do so in specific regions. While the PRI could arguably benefit from some incumbency advantage, the fact that there is no immediate reelection suggests that this is not a reason for opposition parties to abandon races where that party has won in the past.

The reasons as to why opposition parties fail to coordinate have varied across time, depending on their electoral strength and rules of the game. Originally, the most important issue for opposition parties was not so much winning control of the chamber, which was not really an imaginable prospect at the time, but to be players in the political arena, to enter the system.[21] So the first priority was to obtain enough votes so as to be above the electoral threshold, and hence retain the official registry. Winning a single member district implied concentrating resources in obtaining a payoff of, let us say, 51% of the vote in 1/300 of the races. Following such strategy, 10 deputies could be obtained with 1.66% of the national vote concentrated in ten districts. However, those same 10 deputies could be obtained having only 1.8% of the vote in all 300 races.[22] Since party registration laws already required that parties had activists regionally scattered all over the country, concentrating in the multimember districts was an easier strategy. A national campaign emphasizing visibility to the party label was thus preferred over 300 localist candidate centered races.

After a few rounds, however, opposition voters started to concentrate votes into the more viable parties (the PAN and the Partido Socialista Unificado de Mexico, PSUM), and politicians in opposition parties from the left were willing to coordinate under a single label (the PSUM). Since two ballots were cast, one for the single member district candidate and another for the multimember district lists, it was probably just a matter of time for voters to start splitting their votes to sophisticatedly cast their vote only for the front runners in district level candidate centered races, and to vote sincerely their preference for the multimember closed party lists. In fact, there is evidence that voters started to divide their votes, concentrating in the most serious contenders in the single-member races and voting for the smaller opposition parties, including the so-called “satellite” parties, in the multicandidate races.[23] The major opposition party at that time, the PAN, feared that the “satellite” parties would eventually overgrow this party by taking advantage of the two-ballot system. The “satellite” parties, that is, could receive support in the multicandidate races from PRI voters, who would vote for the ruling party in the single-member district races.[24]

The 1986 reformed modified the structure of the ballot with the PAN’s tacit approval.  The ballot was to be cast as a straight ticket, and the seats were no longer strictly compensatory. This meant that the party whose candidate unified the vote in a given district would be rewarded not only with the single member district victory, but also with votes that counted towards the multimember seats. During the years of PRI hegemony, the limitation probably made not much of a difference because the ruling party could win in single-member districts regardless. Nonetheless, in Mexico’s current competitive environment, the single-ballot rule strongly impacts the dynamics of party competition for it prevents opposition voters’ from coordinating behind the strongest challenger. The reason is that the single-ballot system creates few incentives to cast a strategic vote since it implies that no single vote is actually wasted. Electors, that is, can vote for a party that has no real chances of winning in the single-member plurality race and, nonetheless, obtain a pay-off in the form of seats coming from the multicandidate race. The result is lack of coordination, as figure 1 makes explicit, which holds even in districts where the PRI actually could lose if the opposition came together. [25]

That opposition voters tend to focus on the multicandidate races, instead of the single-member ones, is further reinforced by the campaign strategies of the parties. Opposition parties encourage voters to focus on the partisan races. Few voters, for example, know the names of the candidates running in their districts. Congressional campaigns are party-centered. Individual candidates, moreover, can’t develop personal electoral machines since there is no consecutive reelection and all of the resources for the campaigns, including money and activists, come from the party. 

The resulting equilibrium in that the opposition parties have been extremely unsuccessful at winning single-member races, which count the most in the distribution of assembly seats. Winning more single-member races would require coordination. If opposition parties were regionally specialized in such a way that they did not have overlapping areas of support, an arrangement of coordination could presumably be possible to achieve – although with high transaction costs. But this is not the case - there is considerable regional overlap between the PAN and the PRD vote. While each party tends to specialize in some regions, the former being more popular in the North and the Bajio region, and the latter in the South, both are equally strong in the more populous states of the Center, in the metropolitan area around Mexico City, and in the relatively urban and modern districts.[26] A coordination dilemma hence emerges: there is no side payment that a party can give to another party’s candidate bowing out in a single member district, so as to get support for its lists or candidate in another district. It should be noted, moreover, that unlike the Canadian case, coordination failure at the national level does not stem from local bipartyisms, which once aggregated, produce a three party national race. SF ratios in figure 1 show that there is not a bipartisan pattern of competition at the district level.

Existing rules thus have facilitated the entry of opposition parties, but make it hard for the PRI’s dominance to be seriously challenged. The opposition parties have entered and established themselves as relevant institutional players; they however have failed at producing an alternative majoritarian alliance to defeat the PRI. We want to highlight that the PRI actually did not need to use the “safety thresholds” discussed above to “manufacture” its majority control of the Chamber between 1960 and 1994. The ruling party’s vote share and its total number of single-member districts were above 50 percent in all those years –in fact, much above such threshold with the only exception being 1988. Not until the 1997 mid-term elections did the PRI’s vote share fall below 50 percent and, paradoxically, the rules failed at manufacturing a majority for this party.

Thus, we find an important instance of miscalculation in the logic of institutional design: the unexpected loss by the PRI of the majority in the Lower Chamber of Deputies after the midterm elections of 1997. The 1996 electoral reform established a rule for translating votes into seats that, in practice, had the effect of manufacturing a majority for the PRI as long as it’s vote share remained above a certain threshold. If the PRI won 43% of the national vote and 165 seats from the single-member district races, the electoral rule would automatically manufacture a majority for the party (Crespo, 1999). The actual result was only slightly below such threshold: the PRI won 39% of the national vote and 164 single member races. Such result was induced by the economic recession brought about by the 1994 peso crisis, which caused a dramatic change in the preferences of voters, who voted retrospectively away from the PRI in favor of the opposition. The ruling party miscalculated when designing the electoral rule because it failed to anticipate that its electoral support would fall below the established vote threshold.

With the results of the 1997 midterm elections and the ensuing loss of an absolute majority in the lower chamber, the PRI’s unilateral control over the policy agenda collapsed. The dynamics of policy making has thus changed. Not counting with the PRI’s majority in the Lower Chamber, the Mexican president appears now as a far less powerful institutional player in the system than in the past. All major legislations have to be negotiated with the opposition. Legislation has to be approved by a majority, which in practical terms means that the opposition, especially the smallest parties with seats in the assembly, the PT and the PVEM, now possess veto power over legislation. The opposition has also gained an agenda setting power: if the parties come together in a majority coalition, they can approve legislation against the PRI’s will, generating the effects of divided government.[27]

However, except for budget bills, most legislation requires the approval of both the Chamber of Deputies and the Senate. Hence, the PRI still retains significant veto power thanks to its majority control of the Senate. The midterm election of 1997 poised no challenge to the PRI control of the upper chamber.[28]

 

3.2 How to retain veto power if you lose the majority: the Senate.

 

It does not come as a surprise that the PRI still controls the majority of seats in the Senate. What is striking, however, is that through the careful crafting of the electoral rules for the filling of seats in the Senate, the PRI managed at the same time to reduce entry costs for the opposition parties, increase coordination dilemmas among them, and protect its majority even as its electoral support fell. A common use of upper chambers, even in well-established democracies, seems to be to allow for the entrenchment of traditional or authoritarian traits inherited from the past, and Mexico seems to be no exception.

Reforms for the composition of the Senate highlight, more than any other, the logic of institutional design explicitly directed at dividing opposition parties in Mexico. Originally, two senators were elected from each state and the Federal District under a bimodal formula in statewide plurality elections with only one vote cast. The rule disproportionately rewarded majority parties, namely the PRI. In fact, the ruling party had virtually won 100% of Senatorial seats until 1988, when the PRD won four seats coming from Mexico City and Michoacan.

The 1993 reform doubled the size of the assembly, creating four seats for each state, three coming from a trinomial formula in statewide plurality races with only one vote cast and the fourth given to the first minority party in the state. The electoral reform was specifically designed to benefit the PAN over the PRD, opening up a political space that had been virtually closed, but without risking the PRI’s control of the assembly. Despite its increasing electoral strength, the PAN had only won one senatorial seat in the 1991 elections. The reform gave 24 senators to this party in the 1994 race. This meant that the formula ensured 75% of the seats to the party coming first place and 25% for that coming second. This was consonant with an idea of “governance” similar to that discussed above for the Chamber of Deputies, in which the winner was overrepresented, although not as much as under a purely majoritarian system.

The rules implied an unambiguous punishment for the third party and a reward to parties winning a state but remaining below the 66.6% threshold. For the second place party, the reward or punishment depends fully upon whether they are above of below the 33.3% threshold. Thus, the formula generated strong incentives for the two major opposition parties to compete against each other for the second place, especially in states where they were unable to win a plurality of the vote. 

This reform came into effect for the 1994 election, but it did not affect 32 Senators who had been elected in 1991 for six year terms in what turned out to be a short lived reform that attempted to stagger the Senate terms. Hence, the reform did not imply any real threat to the PRI’s control of the upper chamber. That party had 31 senators (it only lost the Baja California Senate race in 1991) who would remain for three more years, so even if the PRI had experienced a catastrophe in the election, it was enough to be second in all states, and to win one single Senate race, for that party to retain the majority. 

The second reform took place in 1996. It returned to the binomial formula, kept the first minority seat and established 32 seats by principle of proportional representation. This time the electoral reform was meant to benefit the PRD. The 1996 reform introduced the principle of proportional representation (PR), but only for one of the four senators, so that the majority of the PRI was safe, even if the party had received no single vote. Thus the makeup of the Senate involved a formula in which the winner would be assured 66% of the seats, the first loser 33%, and an additional district of 32 seats would be allocated according to pure PR (yielding an average district magnitude of 1.97).

During the 1997 elections, only PR Senators were at stake, since the majority and first minority Senators had already been elected in 1994, and in 1997 only 32 seats would be vacated upon the end of the staggered terms of Senators elected in 1991 through the short lived reform that renewed the upper chamber by halves. While no electoral result has actually tested the new rules and how the PRI fares under them, it is insightful to use the results of 1997 to see what could have happened with the Senate composition, had the full chamber been renewed.[29]

            The first thing that must be done to understand the likely effect of these reforms on each party is to distinguish in how many states a party came first, second or third in each state. Table 7 provides that information, revealing that the PRI never came third, and came first in the majority of the states. These victories and second places, coupled with a third of the PR Senators would have been enough to ensure a majority of the chamber for that party.  The table also reports the average effective number of parties calculated for the states that were carried by each party, and the prize the winner obtained, in terms of a greater representation than the percentage vote (66.6%-percentage vote). The PRD seems to do better under multiparty competition, and hence the prize it receives from winning a majoritarian district is larger. The table also shows the prize a party obtains in those districts where it won the first minority (33.3%-percentage vote). The PAN seems to get the smallest prize, which is attributable to the fact that such party is closer to the third of the votes in those states where it came second. Finally, the last column shows that the average over (under) representation of each party, taking into account the wasted votes for the third place that gets no electoral reward, is highest for the PRI. The PRD, since it comes in third place in most races seems underrepresented, at least with this measure.

 

[Table 7 about here]

 

            It is important to highlight that the formula not only discourages coordination, it further encourages competition, between the two major opposition parties. The electoral rule for the Senate implies that voters should mainly focus on electing two slots, a binomial formula and a third senator belonging to the first minority. The size of the modal district is hence 2, which given Duverger’s Law, implies that three parties can effectively compete.[30] Since in most states the PRI is the party most capable of winning the binomial formula in the plurality race, this means that PAN and PRD have strong incentives not to coordinate, but rather to compete against each other for the second place so as to capture the third seat.

It will hence get ever more difficult for the opposition to dislodge the PRI from the Senate: if the opposition fails to coordinate, the PRI will very likely win the overwhelming majority of the plurality binomial races and majority control of the upper Chamber. However, the PAN and the PRD approved the new rules because they significantly reduced entry costs compared to the former purely majoritarian system. Thus the logic of institutional design: reward the existing incumbent majority, reduce entry costs to the opposition and, in doing so, inhibiting opposition coordination. To highlight this logic, table 8 presents counterfactual scenarios of the electoral fortunes of each party under alternative electoral rules. The first scenario is a pure PR system, where there would be a single national district, and it assumes a 5% threshold, in order to eliminate from the analysis the other two minor parties in the Senate. We use data from the 1997 elections.

Under PR the PRI would not have controlled the Senate, since it would have obtained a very similar result to that of the Chamber of Deputies. The Senate would have become under this scenario, in Lijphart’s terms, almost completely congruent with the lower house, so that bicameralism would have become “insignificant”. This insignificance would have been compounded by the asymmetry between both chambers, since the Senate does not have authority for voting the budget (see Diaz-Cayeros, 1998).

 

[Table 8 about here]

 

The last line in the table shows, in contrast, what a pure majoritarian system would have produced. It is clear that the PRI would have done best under this system, but no opposition party would have agreed to retain the current status quo, once the agenda was opened up at the end of the 1980s of the need to make reforms in the Senate. Hence, it comes as no surprise that intermediate mixed systems were created, which tempered the majoritarian effects of winner take all districts, but ensured the PRI a comfortable majority.

The effect of the trinomial second minority system created by the 1993 reform is shown in the second column.  Apart from the binomial system existing before, this is the system under which the PRI does best and the PRD fares worst. [31] The fourth line shows the system currently in place, in which the PRD does better than under any other alternative, except PR. Under the current system, the PAN does slightly better than with the system it accepted in 1993. Hence, the introduction of the PR principle, although making the upper chamber more congruent, was well liked by both opposition parties. Compared to the status quo, however, there is no doubt that this last reform was meant to bring in the PRD. Coincidentally, it is the first time ever that this party approves an electoral reform.[32]

 

Final remarks

 

                Is the Mexican transition completed or unfinished? If the premise is that democratization will not be completed until the PRI loses the presidency, then the process will perhaps be unfinished for a long time.[33] This is because party dominance is a behavioral equilibrium that seems to be difficult to change. By this we mean that the PRI does not remain in power against the will of the voters or exercising political force. It rather remains in power because opposition forces have been unable to craft an alternative majoritarian alternative. 

The PRI’s chances of retaining power in Mexico’s more competitive environment crucially depend on the split of opposition parties. This is actually one of the key mechanisms that explains party dominance elsewhere. Coordination failure at the national level does not stem from local bipartisms, which once aggregated, produce a three party national race. Neither is failure to coordinate driven by social cleavages or ideology alone. Since party competition in Mexico is multidimensional, an argument similar to Riker (1982) on the central location of the dominant party in India driving the division among opposition parties is untenable.

We have argued that the prevailing rules of the game drive the existing non-coordinating equilibrium. These rules were not imposed through undemocratic means. Quite often, one of the major opposition parties agreed to them. In particular, the PRI followed a “divide-and-rule” strategy, offering short-term electoral benefits to a fraction of the opposition, in exchange of rules that would eventually make the incumbent party even more difficult to dislodge. In doing so, the PRI could successfully discourage opposition parties coming together behind an overarching agenda of institutional reform.

Electoral reforms in Mexico have consisted of piecemeal changes to validate existing balance of forces: the dominance of the PRI and an ever-growing opposition camp. In the long process of bargains over institutions, the strongest tension has been between disproportionately rewarding the existing majority, on the one hand, and making the rules more proportional, on the other. The rules of the game have transited from a truly majoritarian all single-member district system, to a mixed plurality single-member-PR system that has significantly reduced entry costs for the opposition. The existing system still disproportionately rewards the existing incumbent majority, but at the same time prevents potential opposition majorities to form.

That the PRI in Mexico has been particularly successful at crafting institutions to its advantage sharply contrasts with the failures at entrenchment and miscalculations at institutional design that characterized communist parties in Eastern Europe. We believe that the difference is explained by the very fact that the Mexican transition has been so slow. Through repeated electoral rounds, uncertainty about voters’ preferences and the likely consequences of different electoral rules is minimized.

Two hypotheses then emerge from the analysis regarding the logic of institutional design. First, the higher the degree of uncertainty regarding voter’s preferences and the existing balance of forces, the higher the possibility to miscalculate the likely effects of different electoral rules. In the Mexican case, we showed that there is only one important instance of miscalculation that represents the exception to the logic of institutional design in favor of preserving the existing balance of forces: the loss by the PRI of the majority in the Lower Chamber of Deputies produced by an unexpected change of voters’ preferences, which fell below the “safety threshold” above which a majority for the PRI could have been legally “manufactured”.

The second hypothesis is that only if an incumbent seriously calculates that it might become a minority electoral force in the short-run, will it have a preference for more proportional electoral rules so as to minimize the risk of losing it all. As long as the incumbent has certainty over its majoritarian support, rules will be designed so as to protect it. If this hypothesis is correct, we only expect the system to move toward true proportionality if the PRI calculates that there are significant chances of becoming a minority party in the future. As long as the PRI continues to be supported by the electorate, however, chances are slim that the electoral system will change in that direction.

References

 

Aguirre, Pedro, Becerra, Ricardo, Cordova, Lorenzo and Woldenberg, Jose (1995) Una Reforma Electoral para la Democracia. Argumentos para el Consenso Mexico: Instituto de Estudios Para la Transicion Democratica, A.C.

 

Alvarado, Arturo (1995) “Los Resultados de la Eleccion para Senadores” in Perez Fernandez del Castillo, German et.al. (eds.) La Voz de los Votos: Un Analisis Critico de las Elecciones de 1994 Mexico: M.A. Porrua.

 

Axelrod, Robert (1970) Conflict of Interest Chicago: Markham.

 

Balinski, Michael Ramirez, Victoriano (1996) “A Case Study of Electoral Manipulation: The Mexican Laws of 1989 and 1994” Electoral Studies 15(2).

 

Becerra, Pablo Javier (1996) “El Sistema Electoral y de Partidos Mexicano: La Transicion Interminable. Notas Sobre la Reforma Electoral de 1996” in Miron, Rosa Maria and Valdez, Leonardo (eds.) Partidos y Elecciones. Congreso Nacional de Ciencia Politica. Mexico: UAM/IFE/CNCPyAP.

 

Codigo Federal de Instituciones y Procedimientos Electorales (1996) Mexico: Insituto Federal Electoral.

 

Constitucion Politica de los Estados Unidos Mexicanos Comentada (1990) Mexico: UNAM

 

Constitucion Politica de los Estados Unidos Mexicanos (1999) (http://www.juridicas.unam.mx/cnsinfo/fed00.htm).

 

Cox, Gary (1997) Making Votes Count. Strategic Coordination in the World’s Electoral Systems Cambridge: Cambridge University Press.

 

Cox, Gary (m.s.) “Duverger’s Law: an Interpretative Review of Research”

 

Craig, Ann and Cornelius, Wayne (1995) “Houses Divided. Parties and Political Reform in Mexico” in Mainwaring, Scott and Scully, Timothy R. (eds.) Building Democratic Institutions. Party Systems in Latin America Stanford: Stanford University Press.

 

Crespo, Jose Antonio (1996) “Nueva Integracion del Congreso” Reforma Suplemento Enfoque No. 136 (August 11)

 

Diaz-Cayeros, Alberto (1997) Political Responses to Regional Inequality: Taxation and Distribution in Mexico Unpublished Ph.D. Dissertation, Durham: Duke University.

 

(1998) “Federalism and Endogenous Institutional Change in the Mexican Senate.” Paper presented at the Latin American Studies Association Meeting.

 

Dominguez, Jorge (1999) “The Transformation of Mexico’s Electoral and Party Systems 1988-97” in Dominguez, Jorge and Poire, Alejandro (eds.) Toward Mexico’s Democratization. Parties, Campaigns, Elections and Public Opinion New York: Routledge.

 

Dominguez, Jorge and McCann, James (1996) Democratizing Mexico: Public Opinion and Electoral Choices Baltimore: Johns Hopkins University Press.

 

Eisenstadt, Todd (1999) "Weak Electoral Institutions or Legacies of Social Conflict? Modeling Causes of Mexico's Local Post-Electoral Mobilizations, 1989-1998." Paper prepared for delivery at the 1999 Annual Meeting of the American Political Science Association, Atlanta, Ga., September 2-5.

 

Gomez Tagle, Silvia (1997) La Transicion Inconclusa. Treinta Anos de Elecciones en Mexico Mexico: El Colegio de Mexico.

 

Laver, Michael and Schofield, Norman (1991) Multiparty Government. The Politics of Coalition in Europe Oxford: Oxford University Press.

 

Lijphart, Arend (1984) Democracies. Patterns of Majoritarian and Consensus Government in Twenty-One Countries New Haven: Yale University Press.

 

Lijphart, Arend (1984) Electoral Systems and Party Systems. A Study of Twenty-Seven Democracies 1945-1990 Oxford: Oxford University Press.

 

Lujambio, Alonso (1995) “La evolucion del sistema de partidos 1988-1994” in Alcocer,  Jorge (ed.) Elecciones, Dialogo y Reforma, Mexico 1994 / II Mexico: Nuevo Horizonte.

 

Magaloni, Beatriz (1994) "Elección Racional y Voto Estratégico: Algunas Aplicaciones al caso Mexicano" (Rational Choice and Strategic Voting: Some Applications to the Mexican Case) Política y Gobierno, Vol. 1, Num 2.

 

Magaloni, Beatriz (1997) "The Dynamics of Dominant Party Decline: The Mexican Transition to Multipartyism." Unpublished Ph.D. Dissertation, Durham: Duke University

 

Molinar Horcasitas, Juan (1996) “Changing the Balance of Power in a Hegemonic Party System: The Case of Mexico” in Lijphart, Arend & Waisman, Carlos (eds.) Institutional Design in New Democracies. Eastern Europe and Latin America. Boulder: Westwiew Press.

 

Molinar Horcasitas, Juan (1991a) El Tiempo de la Legitimidad. Elecciones, Autoritarismo y Democracia en Mexico Mexico: Cal y Arena.

Molinar Horcasitas, Juan (1991b) “Counting the Number of Parties: an Alternative Index” American Political Science Review 85.

 

Moreno, Alejandro (1999) “Campaign Awareness and Voting in the 1997 Mexican Congressional Elections” in Dominguez, Jorge and Poire, Alejandro (eds.) Toward Mexico’s Democratization. Parties, Campaigns, Elections and Public Opinion New York: Routledge.

 

Ordenamientos Electorales (1994) Mexico: Tribunal Federal Electoral.

 

Pacheco Mendez, Guadalupe (1995) “Los Resultados Electorales de 1994” in Pascual Moncayo, Pablo (ed.) Las Elecciones de 1994 Mexico: Cal y Arena.

 

Pempel, T.J. (1990) Uncommon Democracies The One Party Dominant Regimes Ithaca: Cornell University Press.

 

Riker, William (1982) “The Two Party System and Duverger’s Law: An Essay on the History of Political Science” American Political Science Review Vol. 76.

 

Sanchez Gutierrez, Arturo (1995) “La Eleccion de la Camara de Diputados” in Perez Fernandez del Castillo, German et.al. (eds.) La Voz de los Votos: Un Analisis Critico de las Elecciones de 1994 Mexico: M.A. Porrua.

 

Sartori, Giovanni (1976) Parties and Party Systems. A Framework for Analysis Cambridge: Cambridge University Press.

 

Shugart, Matthew and Carey, John (1992) Presidents and Assemblies Cambridge: Cambridge University Press.

 

Taagepera, Rein and Shugart, Mathew (1989) Seats and Votes New Haven: Yale University Press.

 

Valdez Zurita, Leonardo and Larrosa Haro, Manuel (1995) “Las Reformas Electorales 1989-1993” in Alcocer,  Jorge (ed.) Elecciones, Dialogo y Reforma, Mexico 1994 / II Mexico: Nuevo Horizonte.

 

Zmitz, Hector and Hernandez, Carlos (1990) “La Composicion Politica de la Camara de Diputados” Revista Mexicana de Sociologia 136 (jan-mar).