Doomed to disagree? Party-voter discipline and policy gridlock under divided government

Abstract

This article explains the existence of policy gridlock in systems with divided government, even when there are policies that are universally preferred to the status quo. It is shown analytically that one dimension of party institutionalization (the degree of party-voter discipline) may create incentives for veto players to block policies that, ideologically, they might like. This is the case because when party attachments dominate voters' behaviour across different electoral arenas, veto players in the opposition might find it in their electoral interests to prevent popular policies from being adopted. We illustrate our argument by analysing the recent experiences of two Latin American democracies living under divided government but with opposite levels of party-voter discipline: Mexico and Brazil. Contrary to the received wisdom, the low degree of party institutionalization in Brazil may have helped the passing of comprehensive policy reforms, whereas strongly institutionalized parties in Mexico might have been partly responsible for the persistence of policy gridlock.

Keywords

divided government gridlock institutionalization Latin America party discipline presidentialism

The puzzle of bargaining failure in divided government settings

Why do veto players often reject policy proposals that, ideologically, they prefer to the status quo? In divided government settings, it is common for policy proposals to be systematically blocked by veto players, even if, on ideological grounds, these veto players prefer those policies to the status quo.¹ President Vicente Fox of Mexico, for example, was unable to pass major fiscal reforms that would increase state revenues, because he did not enjoy a majority in Congress, albeit the opposition parties might also have preferred a higher level of taxes and revenues. In Germany, attempts by the red-green governing coalition (1996-2006) to lower taxes and reform social security were defeated in the Christian-Democrat-controlled Bundesrat, even though, ideologically, these reforms should have been welcomed by the centre-right opposition. Observers have singled out electoral considerations as being responsible for this puzzling behaviour on the part of veto players. Indeed, it seems that in presidential systems presidents' legislative coalitions around their policy platforms are much more unstable as the next election approaches (Altman, 2000). But why are those perverse electoral incentives more prevalent in some contexts than in others?

Tsebelis's (2002) veto players theory and most of the divided government literature cannot explain legislative gridlock when mutually beneficial improvements are spatially possible. Although there is no dominant theory of legislative bargaining failures, some explanations have been proposed as to why strategic disagreements occur, emphasizing the role of electoral motivations (Gilmour, 1995; Groseclose and McCarty, 2000). By blocking policies, politicians may be sending a signal to rationally ignorant voters (Magar, 2007) or blaming the other veto player for being out of step with voter preferences (Groseclose and McCarty, 2000). Our model concurs with these approaches in emphasizing electoral motivations and imperfect information as the driving forces behind bargaining failures. Our contribution is to show that veto players' incentives to play the electoral ‘blame game’ may be related in systematic ways to certain traits of parties.

Strongly institutionalized parties are typically considered a precondition for a well-functioning democracy. A large political science literature has linked its presence to many salutary outcomes.² They are instrumental in channelling people’s demands into the policy process, they reduce the uncertainty about politicians' behaviour when in office and might help keep leaders accountable (Cox, 1987; Stokes, 1998). But one key dimension of party institutionalization, namely the degree of party-voter discipline, we shall argue, makes it difficult for voters to differentiate among arenas of electoral competition in systems of divided government. When voters' partisan alignments are so strong that they do not differentiate between arenas of electoral competition, political actors whose support is necessary to pass policies might find it in their interests to oppose the adoption of mutually beneficial policies proposed by the executive. By so doing, these secondary veto players increase their chances of winning in the next election.

To illustrate the argument, we study the legislative performance of two Latin American countries, Mexico and Brazil, which happen to have experienced divided government recently, but have radically different levels of party-voter discipline. As we shall argue, whereas the loss of unified government in Mexico brought about the inability of the executive to implement reforms which were ideologically favoured by all relevant political actors, in Brazil minority presidents have been relatively successful in building broad coalitions for similar kinds of policy reforms.

The remainder of the article is structured as follows. In the second section we discuss what we mean by party-voter discipline in the context of the party institutionalization literature. Next, we show analytically why high levels of party-voter discipline might prevent mutually beneficial reforms from being approved in divided government settings. The last section of the article develops an analytical narrative by which to interpret the divergence in the degree of gridlock experienced by similar policy packages in Mexico and Brazil in the light of our argument. We draw our conclusions in a final section.

Weak and strong parties

According to Mainwaring (1999), an institutionalized party system is one in which ‘actors develop expectations and behavior based on the premise that the fundamental contours and rules of party competition and behavior will prevail in the foreseeable future’. This corresponds with the common understanding of ‘strong’ parties, i.e. parties that have clear policy platforms, persist over time and prevail over their specific leaders. Mainwaring identifies four dimensions in his definition: stability of patterns of inter-party competition, roots in society, legitimacy at the social level and resources.

We are concerned with party characteristics rather than party systems, and we focus on a plausible dimension of party institutionalization, namely the degree of party-voter discipline. A high degree of party-voter discipline simply means that in elections, voters select parties, not candidates, and this preference extends to all types of election. Put differently, voter-disciplined parties are those that manage to get the electorate to vote in the same way independently of the type of election.³

The literature has generally interpreted the institutionalization of parties and a high degree of party-voter discipline as desirable features of a party, particularly in new democracies. First and foremost, strong parties are seen as necessary instruments for accountability purposes. The argument goes back to Downs (1957: ch. 7). Parties provide expectations about politicians' behaviour in office, which makes it easier to hold them accountable (Stokes, 1998). When parties do not exist, people make choices based on personalistic appeals, and electoral democracy becomes ineffective (Mainwaring and Scully, 1995). Strong political parties may also bring stability to the legislative process (Casillas, 2001; Figueiredo and Limongi, 2000; Neto et al., 2003).

We do not regard these salutary effects of strong parties as unimportant. On the contrary, a complete normative assessment of the consequences of party institutionalization should take these considerations into account. However, we argue that the dimension of institutionalization we consider - party-voter discipline - might jeopardize effective policy-making in situations of divided government.

The problem of strong parties under divided government: An analytical illustration

In this section we show analytically how policies can systematically deviate from the ideological preferences of all veto players under situations of divided government if party-voter discipline is high.

Assume that voters' and parties' preferences can be represented in a one-dimensional policy space. There are two types of election: one selects the dominant veto player (which we call President), the other a secondary veto player (which we call Legislature). Both veto players obtain ideological benefits from moving policy closer to their ideal point, but only the President enjoys the spoils of power, i.e. what we call the residual (Miller and Hammond, 1994). This encapsulates the idea that there are intrinsic gains from controlling implementation of the policy after the content of the policy has been agreed upon. Finally, the information about the actors' preferences is common knowledge to all veto players.

The sequence of the game is as follows: two parties compete in elections, and the party closest to the median voter in the dimension of each election gets elected (as shown in Figures 1 and 2).⁴ There is a preference shock that makes the status quo undesirable. The President proposes a policy change, and the Legislature either acquiesces or vetoes the proposal. Voters elect a new President after observing the new policy, and the game ends.

Electoral accountability is imperfect, so voters have to infer each party’s true preferences. They use the following rule: associate the observed policy with the incumbent party and use the platform proposed by the opposition party to infer its policy preference.

Imagine a situation of divided government: different parties have won in different elections (party A, Figure 1; party B, Figure 2). There is a shock in the preferences determining the outcome of the presidential election, and voters' preferences shift towards party B’s original ideal point, as shown in Figure 3.B. Party preferences also move (Figure 3.C). If the incumbent changes policies accordingly (that is, party A implements its ideal point), the median voter would keep enjoying the best policy among those offered. However, under divided government, voters infer party A’s preferences based not on the policies that A would pursue on her own, but on the policies that A has to agree with party B. Ideologically, party B prefers A’s new policy to the status quo. But knowing that voters will derive party A’s new preferences based on this new agreement, or lack thereof, party B will be tempted to oppose that policy change in order to increase its chances of winning in the next presidential election.⁵

Party A will anticipate B’s obstructionist strategy, and the undesirable status quo will remain in place. In the next first-order election, voters face the scenario depicted in Figure 3.D. They compare the positions of party A (the status quo) and party B. In this case, B might win the next election (thanks to the gridlock, B is now closer to the median voter) and impose its preferred policy.

When deciding whether to acquiesce or not to A’s proposal in the first round, B must choose between accepting a socially beneficial policy today and tomorrow (the one proposed by A), and keeping the undesirable status quo today and increasing the possibility of imposing her most favoured policy and enjoying the residual tomorrow. Formally, B will endorse the proposed reform whenever the following inequality holds:

- {(P_{P} - P_{L})}^{2} - δ {(P_{P} - P_{L})}^{2} \geq - {(S Q - P_{L})}^{2} - α (δ {(S Q - P_{L})}^{2}) + (1 - α) δ (0 + ρ)

where P refers to policies (points in the policy space), the subscripts indicate whether the policy is the (P)resident’s or the (L)egislature’s ideal point, δ is a discount factor (denoting that the present is more valuable than the future whenever δ < 1), ρ is the residual and α (0 < α < 1) measures the degree of party-voter discipline (low if α = 1, high if α = 0). If B accepts the President’s policy proposal, it will enjoy a moderately desirable outcome both today (-(P_P - P_L)) and in the future (-δ(P_P - P_L)). If B rejects the proposal, it gets the (undesirable) status quo today (- (SQ - P_L)) plus a lottery in the future with an expected value contingent on the degree of party institutionalization α.

When party-voter discipline is high (α close to 0), voters do not distinguish across arenas of electoral competition and will either reward or punish the dominant veto player depending on how close the observed policy is from its ideal point. In this case, the undesirability of the status quo gives party B an incentive to block new policies. The reason is that, if gridlock prevails, the President’s party platform in the next election will correspond with the status quo. The more undesirable the status quo becomes, the more difficult it will be for party A to win. If party B wins, it gets in the second round its maximum pay-off, i.e. its ideal policy plus the residual.

When the parties have low levels of voter discipline (α close to 1), voters' attention focuses instead on election-specific factors. Because obstructionism does not increase B’s chances in the presidential election, party B’s incentive to block mutually beneficial policy proposals disappears.

Formally, the dominant veto player (A) will propose the closest policy to its ideal point that satisfies both (1) above (so that B will not veto it), and the following inequality (the reform’s pay-off has to be positive):

ρ (1 - δ) - {(P_{L} - P_{P})}^{2} - δ {(P_{L} - P_{P})}^{2} \geq ρ - {(S Q - P_{P})}^{2} + α (δ (- {(P_{P} - S Q)}^{2} + ρ)) - (1 - α) δ {(P_{P} - P_{L})}^{2}

For the simulations that follow, we assume that the status quo is located in point 2, the presidential veto player’s ideal point (party A) in 3, and the legislative veto player’s (party B) in 7, as shown in Figure 4. In this example, the winset (the range of policy points that is preferred by both veto players to the status quo) ranges from 2 to 4 and the Pareto frontier (the range of policies for which there is no policy alternative that makes the two veto players simultaneously better off) from 3 to 4.

Figure 5 represents the equilibrium policy for different values of the model parameters. The equilibrium policy is defined as the one closest to the presidential veto player’s ideal point that satisfies both inequalities, if there exists one, or the status quo if both inequalities cannot simultaneously be satisfied.

The equilibrium policies (shown in the vertical axis) change as a function of the ρ parameter, the value of the residual (represented in the horizontal axis) and the value of the party-voter discipline parameter α, which varies from 0 (highest degree of voter-party discipline) in the northwest graph, to 0.2 in the northeast, 0.5 in the southwest and no party-voter discipline at all (α = 1) in the southeast. For sufficiently large values of the residual, strong levels of voter-party discipline prevent veto players from reaching agreements within their original Pareto frontier (denoted by the shaded area). Depending on the degree of party-voter discipline, after reaching a certain value of the residual, the status quo always prevails. As party-voter discipline decreases, mutually beneficial agreements are reached more easily. At the extreme, when voters are not coordinated across elections (α = 1), there always exists an agreement within the Pareto frontier, irrespective of the size of the residual. The first-mover advantage created by the set-up of the game guarantees that the policy chosen within the Pareto-frontier is the one closer to the President’s ideal point.

Interestingly, when residuals coexist with intermediate levels of party institutionalization, the secondary veto player manages to approximate the executive’s proposal to its ideal point. Socially undesirable policy gridlock is less likely, but the outcome is not party A’s ideal point as before, but a policy closer to that of party B. The reason is that the President is willing to grant ideological concessions in exchange for not being removed from office. Indeed, the magnitude of the concession increases with the size of the residual.

It is also interesting to explore the effect of the discount factor (δ). Figure 6 shows the equilibrium policy for different values of α and δ, for an intermediate value of the residual (ρ = 30). When veto players heavily discount the future (δ = 0.5), policy outcomes within the winset are obtained relatively easily. However, as the future becomes more valuable (δ = 1 or, especially, when δ = 2, for instance, in the run-up to an election), the socially undesirable status quo tends to prevail. The intuition behind that result is straightforward: because obstructionism only pays off in the future, it is less attractive the more myopic veto players are.

To sum up, when party-voter discipline is high, when actors care more about the future and when there are large residuals that parties can seize from being in office, divided government tends not to function efficiently in the sense that policies that would be preferred to the status quo by all relevant actors are not adopted.

It is beyond the scope of the article to test all the empirical implications of the model. What we offer next is an illustration of our argument by looking at the recent experiences of two countries under divided government, but with opposite levels of party-voter discipline: Mexico and Brazil.

Party-voter discipline and policy gridlock in Mexico and Brazil

There are a number of ways to go about measuring party-voter discipline. One is to examine the degree of party volatility across elections under the assumption that party support changes less the more institutionalized parties are. This measure would be contaminated, however, by any change of preferences in the electorate: any change in vote levels for different parties is wrongly interpreted as a change in the degree of institutionalization of these parties.

A more appropriate indicator is the degree of ‘party coherence’ of the electorate in two different elections held simultaneously. If vote-shares for a given party across elections are roughly similar, we infer that party-voter discipline is high. Figure 7 plots the values of this variable for ten Latin American democracies.

Mexico and Brazil stand out as two extreme cases along this dimension in the region: whereas Mexican parties do just as well in presidential as in lower chamber elections, which we interpret as an indicator of strong party-voter discipline, in Brazil there are enormous differences in the vote-shares of the same party for different concurrent elections, probably the consequence of some electoral institutional features such as open party lists (Ames, 1995; but see also Mainwaring, 1998). Figure 8 plots the vote-share in congressional elections against the vote-share in presidential elections for the four parties that have held the presidency in the past decade in these two countries: PRI (Partido Revolucionario Institucional) and PAN (Partido Acción Nacional) in Mexico, and PT (Partido dos Trabalhadores) and PSDB (Partido da Social Democracia Brasileira) in Brazil.

In Mexico, differences in vote-share between the two types of election are very small, both for PRI and PAN. In contrast, Brazilian parties' levels of support across elections differ greatly. For instance, in the two states where the PSDB presidential candidate did best in 2002, Rio Grande do Sul and Tocantins, the PSDB was barely competitive in elections to the Senate. The graph also reports the coefficient of the bivariate regression. A positive coefficient of 1 implies that the vote-share in congressional elections is translated into the same proportion of votes in presidential elections. Only for the two Mexican parties can this hypothesis not be rejected at conventional statistical confidence levels. For PT, the coefficient is positive and significant (indicating that there is some relation between the support to a party between different elections), but is less than 1. In the case of PSDB, levels of support in congressional elections provide no information at all about the expected level of support for PSDB in the presidential election (the coefficient is statistically indistinguishable from zero). This evidence is consistent with the general view that PT is a highly institutionalized party by Brazilian standards (Mainwaring and Pérez Liñán, 1997; Samuels, 2004), and with previous evidence that there are few partisan ‘coat-tail’ effects in Brazilian presidential elections (Samuels, 2000). Overall, we take this as evidence of strong party-voter discipline in Mexico and weak party-voter discipline in Brazil.

It is not news that the Brazilian parties are less institutionalized than Mexican parties. What this section has emphasized is that these two countries differ radically in the dimension of party institutionalization that we are interested in: while parties heavily structure the vote in elections for different tiers of government in the Mexican case, they do so to a much lesser extent in Brazil. According to our argument, we can expect divided government to lead to policy gridlock even if all veto players could potentially agree on mutually beneficial reforms in Mexico, but less so in Brazil. To study whether this has been the case, we next review the divergent fates of recent comprehensive macro-economic policy reforms initiated by presidents in these two countries under opposition-dominated Congresses. These examples should not be read as a rigorous test of the theory developed above, but rather as a way of illustrating the mechanisms highlighted by the theoretical argument.

In spite of the constitutionally mandated separation of powers between the presidency and congress, Mexico did not experienced de facto divided government until the democratizing reforms of the 1980s and 1990s led to the loss of PRI of the majority of the Cámara de Diputados in 1997. Divided government made legislative proposals less likely to succeed (the proportion of legislative proposals that ended up being approved fell from 43 percent in 1994-7 to 22.6 percent in 1997-2000) and reduced the legislative role of the president: Zedillo sent to Congress only 38 legislative proposals from 1997 to 2000, compared to 84 from 1994 to 1997. Whereas 76.9 percent of the laws approved by congress between 1994 and 1997 had been introduced by the President, this dropped to 20.4 percent between 1997 and 2000 (Nacif, 2006).

Although, arguably, inefficient policy gridlock existed in the last Zedillo years (Casillas, 2001), it was after Vicente Fox won the presidential election of 2000 that the problem intensified. For the first time in Mexican history, the largest party in the Congress, PRI, was not in the executive, opening a new era for executive-legislature relations. In spite of the apparent unanimity observed by looking exclusively at plenary votes in the Fox years (a large majority of bills sent to the plenary ended up being approved with the support of the three main parties, which probably reflected the fact that the executive sent to vote only bills for which there was enough support), legislative productivity came to a halt, notwithstanding the ambitious reformist agenda of the incoming president. In his first legislature years, Fox only managed to pass 50 laws compared, for instance, with the 83 laws approved by Zedillo in the last legislature with unified government (Delal Baer, 2004; Lawson, 2004; Pastor and Wise, 2005). Certainly, a good deal of the policy gridlock observed under Fox was due to ideological differences between the President and the majority in Congress, but, as we argue below, there are powerful reasons for arguing that at least some of the President’s proposals were not rejected because they were perceived by the opposition to be ideologically worse than the status quo.

The example of tax reform is illustrative. Mexican tax revenues are among the smallest in the OECD countries, and there is widespread consensus among economists and policy-makers about the need to expand the tax base. A survey conducted in 2001 by a newspaper (Reforma) showed that up to two-thirds of Mexican deputies agreed with the idea that the tax base should be expanded (Lehoucq, 2007). As part of his economic programme, Fox attempted several times to introduce tax reforms aimed at increasing revenues. In principle, we can expect opposition parties (ideologically to the left of the conservative PAN) to be in favour of these proposals, as they were improvements over the status quo.⁶ Yet not one of the three attempts managed to garner majority support in Congress.

The first proposal came in 2000 with the elimination of a wide range of exemptions that plagued the value added tax (VAT) and strengthening of the fiscal authorities. PAN legislators persuaded the President to abandon the plan. The next attempts took place in 2001 and 2003, and were drafted as comprehensive fiscal reforms that, in addition to the tax increase, would include programmes to compensate low-income groups that would suffer as a result of the elimination of some VAT exemptions. The 2003 proposal was even drafted by a bipartisan committee, and the PRI initially supported it. Yet, at the last minute, over three-quarters of PRI deputies decided to vote against the bill, which was rejected by Congress by a narrow 234-251 vote on 11 April 2003.

A first explanation for the observed gridlock is that, more important than the shared need to increase revenues, parties held very different views on how such an increase should be financed. More specifically, whereas the right-wing PAN aggressively pushed for an expansion of the tax base (most notably, through elimination of the VAT exemption for food and medicines), PRI and PAN strongly opposed this element of the reform on the grounds that it penalized lower income groups disproportionately. Under this view, it was (ideological) disagreements on the distribution of the cost of the reform that ultimately caused the reform to fail. This explanation of the reform is nevertheless incomplete. First, it is questionable whether the reform was as regressive as the opposition parties claimed.⁷ Second, it is hard to reconcile this view with the behaviour of the main opposition party, the PRI. In 1995, under the Zedillo administration, PRI had already approved an arguably regressive fiscal reform that increased VAT from 10 to 15 percent,⁸ and in 2001 gave its support to a reduction of the maximum income tax rate to 35 percent, with a progressive decline until 32 percent in 2005 (the only thing that was passed of the 2001 tax reform). Most importantly, however, it can be shown that in a multidimensional policy space like this one, conflict of preferences over one dimension (distributional consequences) cannot explain the persistence of the status quo when universally disliked in one dimension (state revenues).

To see this clearly, one might think of fiscal reform as a policy comprising two components: the extent to which it increases the revenue of the state, and its distributional consequences, i.e. how progressive or regressive it is. While there was consensus among all relevant players in moving the status quo in the former dimension (the status quo was in one extreme), it is likely that there was no consensus in the latter - PAN’s obstinacy with the elimination of the VAT exemption on essentials could be revealing its desire to make the system more regressive. Figure 9 represents the status quo, the parties' ideal points in the tax reform debate and the expected and observed outcomes. The two circles represent the combinations of policies that provide the same utility as the status quo for Fox and the PRI, so the area in which the two circles overlap can be interpreted as the Fox-PAN winset.

The graph clearly shows why, as long as parties have Euclidean policy preferences and there is consensus about the desirability of a change in the status quo in one dimension, a winset (that is, a set of combinations of policies that the relevant actors prefer to the status quo) must always exist, regardless of the degree of policy conflict in the other dimension. The reason is that, for any given degree of progressivity, there is always a new combination of policies involving a higher level of revenues that increases the utility for all actors. Under a pure veto players model when only ideological considerations matter, the current status quo cannot be an equilibrium.⁹

Why did the status quo prevail? According to our theory, in the Mexican context the Fox administration would have been the (electoral) beneficiary of a socially optimal fiscal reform. This gave in turn incentives to PRI to block that proposal. In principle, this propensity to block could have been compensated by moving the policy proposal towards PRI’s ideal point, a result we obtained in the theoretical section of the article. However, if ideological concessions had to be too extreme (for instance, elimination of the VAT exemption), at some point the Fox administration would have preferred the status quo (i.e. the winset would eventually disappear), which is what happened in the end. In other words, differences across the progressivity dimension meant that the only proposal acceptable to PRI in this second dimension was not sufficiently attractive to PAN. But without a non-ideological element (the electoral considerations induced by the nature of Mexican parties), it is not possible to explain why an agreement to increase revenues between PRI and PAN was unachievable. Indeed, although PAN nearly achieved PRI support, electoral considerations on the part of the more electorally motivated fraction of PRI led by presidential candidate Roberto Madrazo resulted in the final failure of the reform.

Our explanation for the failure of the reform fits remarkably well with the conventional accounts of the final outcome. Former Mexican Foreign Minister Jorge Castañeda provides the following explanation for what happened:

[PRI leader Roberto Madrazo blocked the reforms] because that was in his electoral interest, and his electoral interest consists only of negative things: impede the others to move. And he has done that … thanks to the fact that the PRI is the biggest minority in Congress. With parties like this, there is nothing you can do. (Castañeda, 2004)

According to Lehoucq (2007), the reason why a majority of the PRI did not support the tax reform had nothing to do with PRI’s ideological stance (at the time most PRI governors were in favour of the reform), but with the presidential campaign strategy of Madrazo, who thought that it was in his electoral interests to block the reform. By blocking the reform, they ‘impeded the others to move’, that is, they de facto tied the President’s hands to the status quo, just as in the original illustration of our model in Figure 3. The objective of this obstructionist strategy was thus very clear: to increase PRI’s chances in the oncoming presidential election.¹⁰ According to our theory, this strategy makes sense only in the context of a high degree of party-voter discipline, in which opposition parties can capitalize on the executive’s failures to bring about socially beneficial reforms.

The success of Fox’s successor, Felipe Calderón, in passing a tax reform bill in 2007 under a similar institutional context raises questions about the validity of our theory. Under closer scrutiny, however, the specificities of Calderón’s reform actually strengthen our argument. According to most observers, Calderón’s success can be understood only in reference to two facts: first, the fiscal reform deal included a highly controversial electoral reform designed to benefit the opposition parties. PRI leader in Congress Gamboa put it bluntly: ‘The PRI is very clear on this: without the electoral [reform] there is no fiscal [reform]. The position of my party is that the two reforms go together’ (Reforma, 7 September 2007; see also Mary A. O’Grady, ‘Calderonomics', Wall Street Journal, 17 September 2007). Even if, as was widely recognized, the PRI leadership was ideologically in favour of the fiscal reform, PAN had to accompany it with an electoral reform that benefited the PRI if it was to be approved. As our model illustrates, non-ideological considerations implied that the secondary veto player could obtain more ideological concessions. Second, there were subtle but important changes between the underlying institutional settings faced by Fox and Calderón. First, albeit Calderón’s party still falls short of having a majority in Congress, PAN became the largest minority (207 out of the 500 House members), while PRI was relegated to third place. Arguably, this increased PAN’s bargaining position. More importantly for our purposes, it can be argued that party system dynamics in Mexico made PRI a de facto much less voter-disciplined party (in the sense we use that term: as a party that carries the same level of support across different types of election) in 2006 (Figure 10).

Indeed, the correlation between levels of support for the PRI across elections is substantially weakened, and is less than 1 at conventional statistical significance levels. It is now easier to understand why the congressional PRI that emerged after 2006 was much less interested in engaging in the obstructionist behaviour analysed before. For one, PRI was not even the largest opposition party in Congress, and was therefore in a poor position to capitalize on bad presidential outcomes. And second, based on these electoral results, one might expect the less voter-disciplined PRI in Congress to be more responsive to the pressures from strong local and regional PRI bosses than to the weakened party national leadership. Anecdotal journalistic accounts of developments within PRI since the 2006 election seem to support to this view.^11,12

In Brazil, under the current constitutional regime established in 1988, no president has ever coexisted with a Congress controlled by his own party. Compared to the Mexican case, the division is, if anything, sharper, since (with the exception of the unstable José Sarney term) Brazilian presidents have come from minority parties in Congress: Collor de Mello’s fellow party members represented less than 6 percent of the House of Deputies, while Cardoso’s PSDB and Lula’s PT never managed to control more than 20 percent of the seats by the time they were presidents. In such a weak institutional position, one would expect Brazilian presidents to have experienced frequent legislative boycotts as the ones documented for the Mexican case.

A recent series of studies has shown that Brazilian minority presidents are not deemed to face legislative gridlock, and that, quite the contrary, they have been successful in securing support for their proposals in Congress (Cheibub, 2002; Figueiredo and Limongi, 2000; Neto, 2005), either by providing side payments to other parties (Samuels, 2002) or by using the informal powers of the respective party leaderships in Congress (Figueiredo and Limongi, 2000). The question is why Brazilian presidents have been able to reach agreements with other parties in Congress while Mexican presidents have not.

One possibility is that presidents' strong constitutional powers in Brazil facilitate their negotiations with Congress. However, it is widely recognized that the political authority of Mexican presidents extends well beyond their strong constitutional prerogatives (Weldon, 1997), and, more importantly, it is not altogether clear that strong presidentialism is an antidote to inter-branch conflict. Shugart and Carey (1992) suggest that strong presidential powers are often associated with legislative inefficiencies and policy gridlock when compounded by strong and national-based parties in Congress.

According to our account, Pareto-improving reforms should be more easily adopted in Brazil because non-presidential parties in the legislature have little incentive to block them. In a situation with low levels of party-voter discipline, proposals from the executive that improve the status quo should be welcomed by opposition parties. To see whether the electoral motivations that prevented socially improving reforms from being approved in Mexico are not at play in the Brazilian case, we study in detail a highly visible policy with, arguably, large beneficial consequences, at least from the perspectives of a majority of members in the Brazilian Congress: the Plano Real.

The Plano Real consisted of a series of profound structural reforms initiated by the Finance Minister, later President, Fernando Henrique Cardoso, in response to the macro-economic imbalances that were affecting Brazil in the early 1990s. The plan involved a drastic realignment of the exchange rate, severe monetary tightening to attract financial flows and several reforms aimed at remodelling the entire macro-economic policy-making environment. Three politically important aspects of the plan are worth emphasizing for our purposes: the plan promised to solve a crucial problem for the Brazilian economy (hyperinflation) and thus quickly became extremely popular. In other words, it represented a clear improvement with respect to the status quo for virtually all parties. Second, the plan required the constant collaboration of Congress: most notably, because it entailed a series of constitutional reforms that, as such, required a supermajority of 60 percent of the votes in both chambers. Finally, as virtually all programmes of macro-economic adjustment, the fiscal restraint dimension of it meant that the plan was expected to be, at least in the short run, especially costly for state-dependent groups.

In principle, as the median legislators in both chambers were ideologically to the right of the President, we should expect them to be ideologically more favourable to comprehensive macro-economic and liberalizing reforms than the President and the parties on the left. In sum, the situation of Brazil roughly resembled the Mexican one described above: the undesirability of the status quo created room for mutually beneficial reforms between the President and the legislature, as Figure 11 shows.

Very much as in the Mexican case, the Plano Real could have run the risk of being blocked by the opposition-dominated legislature as a result of its expected beneficial electoral consequences for the President. In fact, it is not controversial to claim that the success of the Plano Real played a major role in the success of Cardoso’s two presidential bids in 1994 and, especially, in 1998. Although, as we show, the negotiation of the plan was not easy, and some provisions faced strong opposition (some administrative reforms did not get support from Congress until 1998), by and large the core of the plan was approved by a Congress dominated by parties different from Cardoso’s. With the Mexican counter-example in mind, the question is: how did Cardoso manage to obtain support from other parties for a policy reform that would greatly increase his chances of re-election - and therefore diminish his opponents' chances? In our view, the weak party-voter discipline of Brazilian parties, unable to influence voters' choices in different level elections, plays a major role in the explanation.

In the negotiation of the Plano Real with Congress, concerns about the electoral consequences that the plan would have in fostering Cardoso’s electoral chances were not particularly relevant.¹³ This did not mean that negotiation with Congress was going to be without problems. One constant headache for the executive’s economic strategy was the need to put local government finances under control. Although this was one of the cornerstones of the fiscal consolidation plan, the local interests controlling Congress made this a particularly sensitive issue. To achieve the acquiescence of the legislature for the whole plan, the executive had to move away from its ambitious initial attempt at fiscal re-centralization and adopt instead a solution that greatly benefited state and local governments. Not only did the central government end up assuming state debts at unprecedented levels (Dillinger and Webb, 1999), but the fiscal reform finally approved ultimately resulted in increased resources for state and local governments: states' revenues as a percentage of GDP increased in the 1990s by 19.2 percent, and municipalities' by 25.6 percent (Samuels, 2003). Moreover, with the extensive use of sunset provisions (i.e. fiscal reforms had to be constantly renewed by Congress), the local interests represented in the legislature maintained control over the future orientation of the reforms.

Once their regional and particularistic concerns were satisfied, legislators from parties different from Cardoso’s had little reason to oppose the President's plan if they were ideologically in favour of it. As in the Mexican case, these legislators were probably aware that the plan would increase Cardoso’s electoral prospects in the 1994 election, but, unlike in Mexico, this played little role in their decision whether or not to support the executive’s proposals.¹⁴ In Brazil, precisely because of the low degree of party-voter discipline of Brazilian parties, these legislators knew that Cardoso’s electoral success was going to be limited to the presidential election, and that his huge popularity would not give his party, PSDB, a clean sweep in the legislative and state-level elections in which they were competing. In spite of Cardoso’s landslide victory in 1994 (he obtained 54.3 percent of the vote in the first round), PSDB obtained only 13.9 percent of the vote in the concurrent legislative election.¹⁵ To be successful in these elections (which were the ones legislators were most concerned with), it would have helped little to block the President’s policies. They preferred instead to follow a more sensible strategy given their constraints: to give support to these policies in exchange for an increased ability to pursue their particularistic political agendas. This strategy, we contend, is only feasible in contexts such as the Brazilian one where parties are sufficiently weak in securing voter discipline.

Conclusions

In this article we have attempted to explain why some policies favoured by all relevant political actors are not adopted in certain divided government settings. We argue that one key factor accounting for bargaining failure between veto players when mutually beneficial policies exist is the degree of party-voter discipline. First, this article has shown analytically why party-voter discipline might create problems for effective policy-making in situations of divided government. Because strong parties homogenize the way people vote across different electoral arenas, voters are unable to hold each branch of government accountable independently of each other. As a result, the multiple veto players that coexist in systems with separation of powers develop incentives to block welfare-enhancing proposals in order to maximize their electoral prospects in the future.

This might explain some cases of socially undesirable policy gridlock that cannot be accounted for by a pure veto players' model. The inability of the divided government in Mexico to pass reforms that were ideologically favoured by all veto players illustrates the argument well. As the high degree of party-voter discipline prevents voters from differentiating between different arenas of electoral competition, some veto players preferred to block the reforms and damage the President’s party reputation for the next election. In contrast, weaker parties have actually facilitated policy compromises between the presidency and the legislature in Brazil.

The implications of the theoretical argument go beyond the Mexican and Brazilian cases. This new perspective might be used to explain the variation in the general patterns of executive-legislature relations in presidential countries, or coalition governments in parliamentary ones, but it is worth restating what the article is not saying. It is not argued here that weak parties or candidate-centred politics are intrinsically better than strong parties or party-centred politics. Parties perform many roles in modern democracies, most of which have not been considered here. Weak and unstable parties might facilitate the approval of mutually beneficial reforms in contexts of divided government, but strong parties might be necessary for other equally valuable ends, such as policy predictability or ideological consistency of politicians. Whether these benefits outweigh the costs we have discussed here is a question to which we cannot offer an answer.

Footnotes

Acknowledgement

Earlier versions of this paper have been presented in seminars organized by IBEI, Barcelona, and the Quality of Government Institute, Gothenburg. For their useful comments and criticisms, we owe special thanks to the participants in these seminars as well as to the anonymous reviewers of Party Politics.

Notes

References

Alesina

Alberto

Rosenthal

Howard

(1995) Partisan Politics, Divided Government and the Economy. Cambridge: Cambridge University Press.

Altman

David

(2000) ‘The Politics of Coalition Formation and Survival in Multiparty Presidential Democracies: The Case of Uruguay, 1989-1999’, Party Politics. 6: 259–83.

Ames

Barry

(1995) ‘Electoral Strategy Under Open-List PR’, American Political Science Review. 39: 406–33.

APSA (1950) ‘Towards a More Responsible Two-Party System’, American Political Science Review. 44, part 2, supplement.

Casillas

Carlos

(2001) ‘Gobierno dividido en México: Entre la pluralidad y la eficacia’, Revista de Estudios Políticos. 113: 75–99.

Castañeda

Jorge

(2004) ‘Méico: tres metas para Fox’, El País, 1 December.

Chiebub

José

(2002) ‘Minority Governments, Deadlock Situations, and the Survival of Presidential Democracies', Comparative Political Studies. 35: 284–313.

Cox

Gary W.

(1987) The Efficient Secret: The Cabinet and the Development of Political Parties in Victorian England. New York: Cambridge University Press.

Delal

Baer

(2004) ‘Mexico at an Impasse’, Foreign Affairs. 83(1).

10.

Dillinger

William

Webb

Steven B.

(1999) ‘Fiscal Management in Federal Democracies: Argentina and Brazil’. Policy Research Working Paper 2121, World Bank.

11.

Downs

Anthony

(1957) An Economic Theory of Democracy. New York: Harper and Row.

12.

Figueiredo

Argelina

Limongi

Fernando

(2000) ‘Presidential Power, Legislative Organization, and Party Behavior in Brazil’, Comparative Politics. 32: 151–70.

13.

Gilmour

John B.

(1995) Strategic Disagreement: Stalemate in American Politics. Pittsburgh, PA: University of Pittsburgh Press.

14.

Groseclose

Tim

McCarty

Nolan

(2000) ‘The Politics of Blame: Bargaining Before an Audience’, American Journal of Political Science. 45: 100–19.

15.

Lawson

Chappell

(2004) ‘Fox’s Mexico at Midterm’, Journal of Democracy. 15: 139–53.

16.

Lehoucq

Fabrice

(2007) ‘Why is Structural Reform Stagnating in Mexico? Policy Reform Episodes from Salinas to Fox’. Background paper for World Bank Mexico Democratic Governance in Mexico: Beyond State Capture and Social Polarization. Mexico City: World Bank.

17.

Lyne

Mona

(2005) ‘Parties as Programmatic Agents. A Test of Institutional Theory in Brazil’, Party Politics. 11: 193–216.

18.

Magar

Eric

(2002) ‘Making Sound Out of Fury: Using Vetoes for Posturing in Chile and in Mexico’, ITAM Working Paper, 2002–04.

19.

Magar

Eric

(2007) ‘A Model of Executive Vetoes as Electoral Stunts with Testable Predictions'. Working Paper, ITAM, Mexico City.

20.

Mainwaring

Scott

(1998) ‘Electoral Volatility in Brazil’, Party Politics. 4: 523–45.

21.

Mainwaring

Scott

(1999) Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil. Stanford, CA: Stanford University Press.

22.

Mainwaring

Scott

Pérez-Liñán

Aníbal

(1997) ‘Party Discipline in the Brazilian Constitutional Congress', Legislative Studies Quarterly. 22: 453–83.

23.

Mainwaring

Scott

Scully

Timothy

(eds) (1995) Building Democratic Institutions: Party Systems in Latin America. Stanford, CA: Stanford University Press.

24.

Marenghi

Patricia

(2001) ‘Organización de los partidos políticos de los países de Centroamérica, México y República Dominicana: Dimensiones Comparadas', in Alcántara

Manuel

Freidenberg

Flavia

(eds) Partidos Políticos de América Latina. Salamanca: Universidad de Salamanca.

25.

Miller

Gary

Hammond

Thomas

(1994) ‘Moral Hazard in Work Organizations', Public Choice. 74: 245–56.

26.

Nacif

Benito

(2006) ‘The Fall of the Dominant Presidency: Lawmaking under Divided Government in Mexico’, in Shugart

Matthew

Weldon

Jeffrey A.

(eds) What Kind of Democracy has Mexico? The Evolution of Presidentialism and Federalism. San Diego, CA: UCSD Center of US-Mexican Studies).

27.

Neto

Octavio A.

(2005) ‘En defensa del semipresidencialismo, pero sólamente para algunos países', in Arias

Ramacciotti

(eds) Hacia el Fortalecimiento de la Gobernabilidad democrática. Situación y Perspectivas del Presidencialismo y el Parlamentarismo en América Latina. Washington, D.C.: OEA.

28.

Neto

Octavio A.

Cox

Gary

McCubbins

Matthew

(2003) ‘Agenda Power in Brazil’s Camara dos Deputados, 1989-98’, World Politics. 55: 550–78.

29.

Pastor

Manuel

Wise

Carol

(2005) ‘The Lost Sexenio: Vicente Fox and the New Politics of Economic Reform in Mexico’, Latin American Politics and Society. 47: 135–60.

30.

Roett

Richard

(1997) ‘Brazilian Politics at Century’s End’, in Roett

Richard

Purcell

(eds) Brazil Under Cardoso. Boulder, CO: Lynne Rienner.

31.

Samuels

David

(2000) ‘Concurrent Elections, Discordant Results: Presidentialism, Federalism, and Governance in Brazil’, Comparative Politics. 33: 1–20.

32.

Samuels

David

(2002) ‘Pork Barreling is Not Credit Claiming or Advertising: Campaign Finance and the Sources of the Personal Vote in Brazil’, Journal of Politics. 64: 845–86.

33.

Samuels

David

(2004) ‘From Socialism to Social Democracy. Party Organization and the Transformation of the Workers' Party in Brazil’, Comparative Political Studies. 37: 999–1024.

34.

Shugart

Matthew

Carey

John

(1992) President and Assemblies. Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press.

35.

Stokes

Susan

(1998) ‘Son los partidos políticos el problema de la democracia en América Latina?’ Política y Gobierno. 5: 13–46.

36.

Tsebelis

George

(2002) Veto Players. How Political Institutions Work. Cambridge: Cambridge University Press.

37.

Weldon

Jeffrey

(1997) ‘The Political Sources of Presidencialismo in Mexico’, in Mainwaring

Scott

Shugart

Matthew

(eds) Presidentialism and Democracy in Latin America. New York: Cambridge University Press.