Abstract
The persistence of policy switches—whereby presidents renege on campaign promises shortly after winning elections—30 years after Latin America’s redemocratization defies established notions of democratic representation, and poses a puzzle to analysts and voters alike. In this article, I advance current explanations for switches, by arguing they can only be understood in the context of currency booms and crises, typical of Latin American economies after their reintegration into world finance in the 1970s. To test my propositions empirically, I examine elections held in the region between 1978 and 2006 and find evidence consistent with the claim that switches are more likely to occur in periods of dollar scarcity, when the need to attract financial capital to the economy pushes leftist presidents into adopting policies opposed to the programs they announced during campaign.
After almost three decades of neoliberalism in Latin America, the motivations and conditions under which governments implement market reforms in the region are still a matter of scholarly debate (Corrales, 2000; Kurtz & Brooks, 2008; Schneider, 2004).
The prevalence of an inward, state-led model of development started to be seriously challenged in Latin America in the mid-1960s, when countries suffered successive crises often associated with an exhaustion of the model of import substitution, at the same time that developed economies moved toward economic liberalization (Hirschman, 1968; O’Donnell, 1973).
In this period, governments’ resistance to liberalizing was mostly attributed to the difficulties democracies encounter in advancing unpopular policies. This led analysts to conclude that prospects for reform should increase under the military governments that plagued the region after the mid-1960s. It turns out that for a number of reasons, among them the favorable conditions of international financial markets during the 1970s, most authoritarian regimes failed to liberalize and instead furthered state-led development strategies on the basis of massive borrowing in international capital markets.
With the return of democratic regimes, the conventional wisdom remained that if dictatorships had not implemented market-oriented reforms, the new governments were even less likely to do so (Stallings, 1989). This time skepticism proved unfounded, as the vast majority of presidents elected in Latin America since the early 1980s managed to implement at least a significant part of the neoliberal agenda (Geddes, 1995).
Although the past three decades marked a period of economic liberalization and democratic consolidation, they were not necessarily a period of liberalization through democracy in Latin America (Haggard & Kaufman, 1995; Stokes, 2001). To the contrary, in many countries newly elected governments implemented neoliberal programs after having explicitly rejected them during campaign, in what Drake (1991) denominated strategies of “bait-and-switch.”
Despite the centrality of switches to understanding the diffusion and consolidation of neoliberalism in Latin America, and the normative significance of this phenomenon for electoral accountability and the quality of democracy in the region, the factors that account for their striking persistence remain to be established.
In the first attempt to systematically study policy switches, Stokes (2001) argued they result from divergent policy preferences between incumbents and voters, in a scenario where voters’ preferences are uncertain and reelection prospects depend not only on the content of government policies but mostly on their material consequences. Presidents, aware of the uncertainty of voters’ preferences, pursue the policies they expect to carry the best economic results, even if this implies reneging campaign promises.
Although this rationale offers a starting point to explaining policy switches, it still leaves us with the challenge of elucidating why, during almost 30 years, switches have consistently occurred in only one direction—from state-oriented policies to a market-oriented agenda. 1 Unless one assumes the latter to be always preferable to other alternatives, divergent preferences do not satisfactorily account for switches.
This article advances and empirically tests an explanation for policy switches that accounts for their unidirectional nature. I contend that switches reflect leftist presidents’ decision to adopt policies likely to attract inflows of financial capital in periods of dollar scarcity. Presidents inaugurated in the midst of currency crises are subject to strong incentives to renounce the state-oriented policies announced during campaign and instead switch to a market-oriented program aimed to attract financial investment to the domestic economy.
It follows that switches should be less frequently observed when state-oriented candidates are inaugurated during currency booms. The lower need to attract foreign finance in “good times” boosts governments’ capacity to diverge from investors’ preferred agenda and to advance their original program.
Finally, if this reasoning holds, it further explains why market-oriented presidents should not switch programs, either during currency crises or booms. The election of presidents who promise neoliberal policies suggests there is no marked contradiction between voters’ and investors’ policy preferences, as both expect a market-oriented agenda.
I examine these hypotheses in a sample of presidential elections held in 15 Latin American countries between 1978 and 2006, and results strongly support my expectations. Of all presidents elected promising state-oriented policies, those inaugurated under intense currency pressures display the highest probability of switching to a neoliberal program. Empirical tests reject the alternative hypothesis that presidents switch in the process of negotiating with opposition, as I find that not the weakest but the strongest presidents are the ones most likely to switch. Neoliberal candidates, on their part, regardless of whether they are elected under currency crises or booms, never switch.
These findings bear important theoretical and empirical implications for understanding the politics of economic policy making in Latin America, with potential extensions to other less developed regions subject to similar cycles of scarcity and abundance of foreign currency. First, they shed light on the conditions that determine when and how financial investors are likely to influence policy making in less developed economies. They also build on previous work on investors’ negative responses to the election of left-wing governments in these countries (Block, Vaaler, & Schrage, 2006; Leblang, 2002; Santiso & Martinez, 2003), by pointing to one mechanism through which markets “discipline” governments, and determining the conditions under which this discipline is effective.
In the particular case of Latin America, the need to build investors’ confidence in periods of crisis helps explain the long-lasting persistence of neoliberalism and also how its consolidation is associated with the low responsiveness of third-wave democracies to voters’ electoral demands (Kurtz, 2004; Weyland, 2004). The analysis further highlights the conditions that determine leftist governments’ capacity to deviate from a neoliberal agenda, as happened during the region’s widely debated “move to the left” in the 2000s (Castaneda, 2006). Contrary to previous occasions in which left-wing presidents were elected amid crises and were forced to renege on their promises, those who took office in this period experienced an unprecedented currency boom in the region, which boosted their ability to implement a state-oriented agenda.
This article is organized as follows. The next section discusses unidirectional policy switches in Latin America and provides preliminary evidence of how they are associated with currency crises. Then, I introduce my theory and develop testable hypotheses for the effect of currency crises and booms on governments’ likelihood of advancing the policies promised during campaign. The section that follows details the research design and presents empirical results, and the last one concludes.
Currency Crises and Policy Switches
Even though parties’ strategies of “bait-and-switch” were first documented by students of Latin American politics in the early 1990s (Drake, 1991; Roberts, 1996), Stokes (2001) was the first to study the phenomenon systematically. Analyzing a sample of 44 elections held between 1982 and 1995, the author noted that 12 out of 22 presidents advanced a program opposed to the one promised during campaign, shortly after inauguration. Notably, in Stokes’s sample, all policy switches occurred from a state-oriented campaign to a neoliberal government.
A decade later, unidirectional switches remain a striking regularity. Figure 1 updates Stokes’ original database until 2006, and shows that from a total of 32 candidates elected on a state-oriented platform, only 13 advanced these policies in office; the remaining 19 switched to a neoliberal agenda. Also consistent with previous findings, none of the 57 candidates who announced the intention to follow market-oriented policies switched to a state-oriented program within the 1st year in office.
Policy switches and currency crises.
According to Stokes (2001), switches result from diverging economic policy preferences between politicians and voters, in a scenario where politicians know that “voters’ beliefs about which policy is best are not fixed, and may change once policies are implemented and their results observed” (p. 61). In this scenario, presidents have incentives to advance the policies they expect to bring about the best outcomes, even if these policies diverge from those promised during campaign.
There are many reasons why politicians might consider the program announced during campaign not to be the most adequate once they are in office. Candidates may be misinformed about the state of the economy, and as they learn about it they conclude that a different set of policies is necessary. Economic conditions may also change after election and demand a different program. Finally, a candidate might be convinced that one program is better but, knowing that citizens will not vote for it, decides to announce policies that will get her elected and switch after inauguration.
Although all these mechanisms potentially illuminate politicians’ electoral behavior, they still leave us with the challenge of explaining why, over 30 years, policy switches have consistently occurred in one single direction. We still do not understand why, when candidates disguise their real intentions during presidential campaigns, these intentions are always to advance a neoliberal program. Or, in case economic conditions change after inauguration, why they always do so in a way that a neoliberal program seems more appropriate. Finally, it is not clear why candidates who campaign on a neoliberal agenda do not ever change their minds and switch to a state-oriented program in office.
Unless one assumes neoliberal policies to be always preferable to other alternatives, though, none of the mechanisms described above satisfactorily unfolds the unidirectional nature of policy switches. Even if neoliberal policies might have seemed the only option in the historical context of Latin American democracies in the post-debt-crisis era, which is itself controversial (Edwards, 1989; Rodrik, 1996), it is still important to note that “while economic theory can tell us a lot about policy alternatives, unless our economics contains an understanding of power, it will not tell us enough to understand the choices actually made” (Gourevitch, 1986, p. 17).
Figure 1 further illustrates what I claim to be the intervening factor missing in current explanations of policy switches; in the elections marked in bold, the winners were inaugurated during a currency crisis. The figure reveals that there is a substantial difference between policy switchers and nonswitchers with respect to their exposure to currency crises—about 80% of policy switchers are inaugurated under such crises, more than twice the percentage observed in the case of presidents who effectively advance a state-oriented program. 2 The figure also shows that the behavior of presidents elected on the promise of market-oriented policies is not attributable to these same pressures; even though they are frequently inaugurated during currency crises, they never switch. This preliminary evidence bolsters the claim, examined in the next sections, that switches reflect left-leaning governments’ attempts to attract foreign capital in periods of dollar scarcity.
Rationale
The reasoning proposed here is straightforward; presidents elected on a state-oriented platform are, different from those who promise neoliberal policies, confronted with a trade-off between advancing interventionist and redistributive policies their voters expect and attracting investors who prefer a market-oriented agenda. Put simply, they confront the extensively studied trade-off between redistributing income and guaranteeing investment to generate economic growth (Alesina & Rodrik, 1994; Przeworski & Wallerstein, 1982).
This trade-off is especially critical in unequal democracies, where the gap between the income of the poor and that of the wealthy is substantial. Because redistributive policies boost the former in a way that economic growth alone cannot do, leftist governments are subject to particularly strong electoral incentives to pursue such programs. These incentives are countered, however, by those created by investors’ capacity to redirect resources to countries where policies are deemed more friendly. The open confrontation of investors can provoke capital flight, depress the economy, and make the poor worse-off, as the first presidency of Alan Garcia in Peru exemplifies.
In this context, presidents elected on a state-oriented program have to decide which policies to advance considering both voters’ and investors’—in this case opposed—preferences. It follows that, the more dependent governments’ are on foreign finance, the harder it becomes for them to advance policies that substantially depart from investors’ agenda.
Currency crises—reflected in substantial decreases in countries’ international reserves and/or sharp currency devaluation—potentially exacerbate this dilemma. These crises often follow sudden changes in terms of trade, especially in economies reliant on commodities exports, and can also result from severe capital flight. Losses of international reserves and/or a depressed exchange rate force governments to take measures to boost the inflows of foreign currency to the economy, to avoid a major collapse. Given the relative stickiness of trade flows, governments most likely respond to currency crises by attempting to attract short-term inflows of financial capital, which frequently involves abandoning a leftist agenda in favor of investor-friendly programs.
Note that what is particular about this reasoning is not the nature of the crisis that forces leftist presidents to adopt neoliberal reforms—currency versus inflationary crises, as it has been commonly claimed in the literature (Stokes, 2001; Weyland, 1996). I am also not arguing here that neoliberal reforms are necessarily the only or most adequate strategy to respond to a currency crisis, and in fact many countries in the past increased state control over international flows exactly in response to such crises (Eichengreen, 1996; Rodrik, 1996). My point is that this specific type of crisis, particularly in the case of less developed economies where capital inflows are anticyclical (Wibbels, 2006) and dependency on commodities limits export-led recovery, forces governments to adopt neoliberal policies with the immediate goal of attracting financial capital in the short run.
The strategy of adopting investor-friendly policies to attract foreign capital—sometimes referred to as confidence building—has been extensively described in the literature that studies the determinants of financial investment in the developing world (Calvo, Leiderman, & Reinhart, 1993); accordingly, Santiso and Martinez (2003) observe that “Latin America’s reform fever of 1990s must be seen in the context of the urgent need for new capital inflows” (p. 27).
The trade-off between redistributive efforts and responsiveness to investors’ demands is also present in Drake’s (1991) assertion that switches “reflect the contradiction between the immiseration of the majority of the population and the imperatives of neoliberal economic restructuring to favor market mechanisms and honor the foreign debt” (p. 36). Both claims suggest that the logic proposed here is not restricted to the aftermath of the 1980s debt crisis—the period Drake was referring to—or to the early 1990s, as in Santiso and Martinez (2003) or Calvo et al. (1993), but encompasses periods of dollar scarcity in general, when governments are subject to the urgent need of attracting foreign finance.
The same logic implies that presidents elected on a state-oriented discourse should be less likely to pursue a neoliberal agenda during currency booms, as dollar abundance reduces the need to attract foreign capital, enhancing their ability to deviate from financial investors’ preferences in favor of their original program. It comes as no surprise, thus, that Latin America’s “left turn” occurred during a currency boom, when historically high international commodity prices widened left-leaning presidents’ room to reject neoliberal policies (Kaufman, 2011; Murillo, Oliveros, & Vaishnav, 2011).
Finally, as presidents who campaign on a neoliberal platform receive a mandate to liberalize and boost investment climate to promote economic development, there are no reasons to expect neither dollar scarcity nor abundance to substantially affect their behavior. It is not unlikely that neoliberal presidents might advance some level of pro-poor policies in “good times” more than in “bad times,” but they should do so without renouncing to macroeconomic discipline and market liberalization. The absence of a trade-off between investors’ and voters’ policy preferences in the particular case of market-oriented candidates, thus, explains the unidirectional nature of policy switches.
The theory just presented is compatible, and builds on the multiple mechanisms Stokes devised to explain policy switches. In cases where currency crises occur before election, leftist candidates may hide their decision to switch if they believe that the promise of state-oriented policies will guarantee their victory. It is also possible that left-wing presidents experiencing currency crises after inauguration find themselves forced to switch to reattract capital to the economy and limit the consequences of the crisis. The probability that presidents change their minds with respect to their capacity or will to depart from investors’ preferences during a crisis, or even about investors’ responsiveness to prospects of a state-oriented agenda during a crisis, is also not ruled out.
This reasoning not only accounts for the unidirectional nature of policy switches but also explains the circumstances under which these presidents should or not be expected to advance them, providing a theoretical framework to explore policy switches beyond Latin American politics.
The Ecuadorean 2002 and 2006 presidential elections illustrate this claim. Both Lucio Gutierrez and Rafael Correa won promising typical state-oriented, leftist policies, including increased social expenditures, renegotiation of foreign debt, and limits on economic liberalization. Both candidates experienced capital flight during campaign, signaling financial investors’ rejection of their agenda, but only Gutierrez endured sustained currency pressures in a scenario of high foreign indebtedness and low export prices. Correa was elected during an oil price boom that limited his government’s need to attract international finance, enhancing his room to deviate from investors’ preferences. Not surprisingly, Gutierrez adopted a neoliberal program soon after his inauguration, whereas Correa advanced the policies announced during campaign.
The hypotheses that follow summarize the arguments above.
Hypothesis 1: Currency crises should increase the likelihood that presidents who campaign on state-oriented policies switch to a market-oriented program in office.
Hypothesis 2: Currency booms should decrease the likelihood that presidents who campaign on state-oriented policies switch to a market-oriented program in office.
Hypothesis 3: Neither currency crises nor booms should affect the prospects for policy switches in the case of presidents who campaign on market-oriented policies.
Political Systems and Policy Switches
Presidents who promise to advance a state-oriented agenda may decide to switch programs while confronting currency crises, but they are less likely to do so if they are constrained by the domestic political system (Haggard & Kaufman, 1995).
Strong presidents are usually thought of as those who have the capacity to influence legislation, which arguably rests on two categories of presidential powers: partisan and constitutional (Mainwaring & Shugart, 1997). Partisan powers reflect the incumbent party’s strength in the legislature and the need to foster alliances. Constitutional powers, inherent in the office of the presidency, allow incumbents to have their preferences taken into consideration in the passage of legislation (Mainwaring & Shugart, 1997).
In Latin American political systems, the president is elected independently from the legislature, and it is quite common that the incumbent party does not control a majority in congress. In these cases, the lack of legislative support might significantly limit presidents’ capacity to influence policy making. It follows that, other conditions fixed, the larger the proportion of the incumbent’s party’s seats in the legislature, and therefore the lower the need to foster political alliances with potential opponents, the more capable a president becomes of initiating her preferred policies, including switching to an alternative program if she deems it necessary.
Constitutional powers—mainly veto and decree powers—also affect presidential strength, allowing the executive to shape policy output even without a legislative majority. These powers determine the ability of the incumbent to influence (or even dominate) the law-making process that results from the president’s standing with respect to the legislature.
Presidents with strong constitutional powers can initiate and veto legislation and should be in a better position to push for their preferred agenda, either by influencing the adoption of policies that represent a change in the status quo or by blocking unfavorable policy changes promoted by the opposition.
Even when a legislative majority can rescind a decree, presidents might still be able to play a major role in shaping legislative outcomes. Unlike a bill passed by a legislature, a presidential decree is already law—not a proposal—before the other branch can react to it. Thus, Latin American presidents frequently resort to the strategy of overwhelming the legislative agenda with a flood of decrees, making it difficult for the congress to consider measures before they have a possibly irreversible effect. Finally, presidents can always use decree power strategically, attempting to discern a point in the policy space at which a congressional majority is indifferent between the status quo and the proposed change.
For all these reasons, presidents with stronger constitutional powers are more likely to switch programs, as they are better positioned to shape policy outcomes according to their own preferred agenda. In sum,
Hypothesis 4: The likelihood of policy switches should increase with the strength of the incumbent party in the legislature.
Hypothesis 5: The likelihood of policy switches should increase the more constitutional powers the executive has.
Finally, an extensive literature has developed since Mainwaring and Scully’s (1995) influential analysis of the political implications of party and party system institutionalization. According to this literature, a major consequence of institutionalization should be to foster governments’ responsiveness to electoral demands (Jones, 2005).
The institutionalization of party systems is conceived as a process by which a practice or organization becomes well established and widely known, and where actors develop expectations, orientations, and behavior, based on the premise that this practice/organization will prevail into the foreseeable future (Mainwaring & Torcal, 2006).
In institutionalized party systems, thus, political actors should develop clear and stable expectations about the behavior of other actors. Stable patterns of electoral competition, the presence of party roots in society, citizens’ recognition of the legitimacy of party politics, and the institutionalization of party organization, as opposed to parties working as electoral vehicles for personalistic leaders, are all necessary conditions for establishing what is considered “good representation.” This notion subsumes representatives who do not work on the basis of leaders’ voluntarist will and who are elected and govern in response to voters’ programmatic preferences.
For all that, both party and party system institutionalization should lower the incentives presidents face to openly betray campaign promises. First, voters’ programmatic preferences should increase the electoral costs of doing so. If these constraints are not sufficient, institutionalized parties should be willing and more capable of vetoing presidents’ attempts to switch policies, at the risk of losing support from activists, legislators, and voters otherwise. Therefore,
Hypothesis 6: The likelihood of policy switches should decrease with the institutionalization of party systems.
Hypothesis 7: The likelihood of policy switches should decrease with the institutionalization of the incumbent’s party.
The next section presents an empirical analysis designed to test the hypotheses just stated.
Empirical Analysis
The research design advanced in this section tests the impact of currency crises and booms on the probability of policy switches. I estimate a probit model where the dependent variable is Switch, and the explanatory variables capture currency crises and booms that occurred in the months immediately before and after presidential inauguration (named the “electoral period”), as well as political conditions deemed relevant to explain presidents’ probability of switching. The sample includes elections held between 1978 and 2006 in 15 Latin American countries (see Table A1 in the appendix).
Dependent Variable: Switch
To examine the effect of currency crises and booms on presidents’ decisions to deviate from their original program, it is necessary to identify (a) presidents’ original program and (b) the policies they initiated in office. Following Stokes (2001), I use campaign promises as a proxy for the first and the policies reported in the media during the 1st year of government as an indicator of the second.
Candidates who promise to increase the role of the state and regulation in the economy, prioritize employment and wage increases over control of inflation, commit to maintaining strategic restrictions to trade and to advance industrial policies, as well as to impose limitations on cross-border capital movements and on the payment of the external debt were classified as campaigning on a state-oriented agenda.
Oscar Arias’s campaign in Costa Rica 1986 presidential election exemplifies a typical state-oriented discourse; Arias advocated the renegotiation of the country’s foreign debt and promised to seek “better treatment” from the International Monetary Fund (IMF) and other international financial organizations. He also stated as his main objective “the creation of a welfare state, instead of a garrison state,” and announced plans to create 25,000 jobs and to build 20,000 new houses a year if elected.
Candidates who commit with reducing the state’s role in the economy through privatization and deregulation, prioritize anti-inflationary shocks and inflation targets, and promise to eliminate subsidies and tariffs and to launch financial liberalization measures and central bank independence were classified as campaigning on a market-oriented, or neoliberal, agenda.
Cesar Gaviria, running for the Liberal Party in the Colombian 1990 presidential election, was a typical neoliberal candidate. He leaned strongly toward opening the economy to international investment and foreign competition, guaranteed investors that debt payments would continue, and demonstrated intentions to be amenable to IMF and World Bank policy recommendations, as well as to follow the regional trend toward privatization.
I obtained information regarding electoral campaigns from newspaper data available in LexisNexis academic (all sources in English) released in the 6 months prior to each presidential election, and extended Stokes’s sample from 44 to 89 cases, including elections held until 2006.
The same criteria and sources were used to classify governments’ policy initiatives. The information used to code each government as state-oriented or market-oriented was also obtained from newspaper sources and completed with case studies when data were scarce (see Conaghan, Malloy, & Abugattas, 1990; Mauceri, 1995; Wilson, 1994). As I am interested in postelectoral switches, I restricted the coding to the 1st year of a president’s term.
The limitation of categories into two groups surely fails to capture a more complex reality where these policies are a matter of degree. Yet, as Stokes (2001) notes, the difficulty in quantifying these degrees “counsels against continuous codings and in favor of reducing positions to a dichotomous choice” (p. 29). The categories adopted capture quite well the basic message of presidential campaigns, often framed between statist and promarket orientations. 3
The dependent variable here is the dummy Switch, which takes the value of 1 when the program advanced in a government’s 1st year differs from the policies promised during the campaign and 0 otherwise.
Explanatory Variables
Crisis and boom
Currency crises/booms, reflecting dollar scarcity/abundance, are captured by the index of exchange market pressure (EMP), created by Eichengreen, Rose, Wyplosz, Dumas, and Weber (1995). The index aggregates changes in international reserves and exchange rates for each country included in the sample, weighted by their volatility, which is measured by the standard deviation of the distribution of EMPs.
EMPs are reflected in falling reserves (s) and/or in a depreciation of the exchange rate (r) (positive EMP). The opposite applies to exchange market booms, when international reserves increase and/or the exchange rate appreciates (negative EMP):
Although Eichengreen et al. (1995) use the same standard deviation for the entire sample, and Leblang (2002) uses one value for each country, I adopt 1-year moving averages of standard deviations for each country. This is arguably a better measure, as earlier observations of both indicators are highly volatile, with volatility significantly decreasing over the three decades studied. The use of a single standard deviation for the whole period would overestimate crises occurred in the early 1980s, and underestimate those observed in the late 1990s and 2000s.
Still following Eichengreen et al. (1995), I create two dummy variables designed to reflect dollar scarcity/abundance, which occur whenever EMP assumes extreme—adopting a cutoff of one standard deviation from the mean—positive/negative values. The inclusion of two dummies increases the flexibility of the analysis, compared with single categorical variable encompassing booms and crises.
In the empirical model, I test a set of variables that capture episodes of currency crises and booms, calculated as sums of the dummies Scarcity and Abundance, respectively, during different time periods. Crisis sums the values of Scarcity in the 12 months that surround an inauguration (plus the month of inauguration), capturing pre- and postelectoral turbulence in currency markets. Crisis.pre does the same but only for the preinauguration period (6 months prior to inauguration and the month of inauguration) whereas Crisis.post sums Scarcity only for the postinauguration period (6 months after inauguration). These variables apprehend the persistence of pressures governments experience in the periods of interest. Governments under extreme pressure for 2 months (Crisis = 2) should be less likely to switch than those that experienced the same pressures during a full year (Crisis = 13). The same was done for periods of dollar abundance, creating the variables Boom, Boom.pre, and Boom.post.
Inflation
Scholars have long studied the effects of inflationary crises as determinants of market reforms in Latin America. Arguably, hyperinflation should not only motivate incumbents to advance painful adjustments (Haggard & Kaufman, 1995; Stokes, 2001), but it should also make citizens more willing to accept the risks they carry when finding themselves in the “domain of loss” (Weyland, 1996). In addition, even though currency and inflationary crises do not always coincide, it is noteworthy that largest currency crashes “are similar in timing and order of magnitude to the profile of inflationary crises” (Rogoff & Reinhart, 2009, p. 6). Due to that, and for the reasons presented in the theoretical section, it is important to separate these effects in the empirical analysis. I do so by including a variable Inflation (the log of the average annualized inflation rate in the 12 months that surround presidents’ inauguration) as a control.
Political Factors
Presidents might prefer to switch programs when facing currency crises, but they are more capable of doing so under specific conditions of the political system. For this reason, variables that reflect institutional and political incentives/barriers expected to have an impact on incumbents’ capacity to switch programs are included in the model:
Executive
This variable reflects the constitutional powers of the executive and was obtained from the Inter-American Development Bank and International Institute for Democracy and Electoral Assistance (IDB/IDEA) database (Payne, Zovatto, Florez, & Zavala, 2002). The index ranges from 3 to 15 and includes measures of package veto, partial veto, decree power, exclusive initiative, convocation of referendum/plebiscite, and power to define budget and to default budget.
Legislature
The share of seats controlled by the incumbent’s party in the lower house is the measure of government strength in the legislature. Sources were also IDB/IDEA, the Political Database of the Americas (PDBA), and Psephos electoral database. 4
Volatility
Electoral volatility is the easiest variable to measure and possibly the most important dimension of party system institutionalization, because institutionalization is conceptually very closely linked to stability (Mainwaring & Torcal, 2006). Although party system institutionalization encompasses other dimensions besides the stable intraparty competition captured by volatility, the latter is still the most widely used proxy for institutionalization. Electoral volatility (lower house) is calculated, for the sample used in this article, using data obtained from the PDBA and Psephos databases.
Party
Party age is used here as proxy for party institutionalization (Mainwaring & Torcal, 2006). Despite the fact that old parties might not be institutionalized, age is arguably a necessary condition for institutionalization and therefore strongly associated with it. Ease to measure is an additional motivation for using age to capture the effects of party institutionalization. Information on party age was obtained from the Database of Political Institutions (DBPI) as well as from political dictionaries of Latin America (Gunson, Chamberlain, & Thompson, 1989).
Results
How do currency crises and booms affect the likelihood that Latin American presidents switch programs after inauguration? 5
As currency crises and booms are expected to affect candidates elected on the promise of market- and state-oriented programs in different ways, I analyze these cases separately. This strategy has the additional advantage of controlling for potential endogeneity in the results, driven by the fact that a component of the dependent variable (Campaign) might be influencing the values of the most important independent variable (Crisis). As figure 1 shows, currency crises are far more likely to occur when a state-oriented, rather than a neoliberal candidate is elected.
Market-Oriented Campaigns
The null impact of currency crises and booms on policy switches from market- to state- oriented programs becomes evident as we observe that, although the treatment varies (the mean value for Crisis and Boom are 3.7 and 2.3, with standard deviations of 3.7 and 3.2, respectively), the effects remain constant (candidates elected on a market-oriented platform never switch).
In case neither the treatment nor effects varied, it would be impossible to determine the influence of currency crises and booms on market-oriented candidates’ propensity to switch. As it stands, the evidence supports the claim that neither currency crises not booms have any effect in this group.
State-Oriented Campaigns
Empirical analyses provides strong support for the hypothesis that state-oriented presidents elected in the midst of currency crises are more likely to switch to a neoliberal program soon after inauguration. Table 1 displays the effect of changes in the main explanatory variables (first differences) on the likelihood of switches from state- to market-oriented programs, and shows that the effect of Crisis is positive, substantial, and consistent across different model specifications. 6
Effects of Changes in Explanatory Variables on the Probability of a Policy Switch.
Holding all other variables in their mean values, a one standard deviation change around the mean in the variable Crisis leads to a 59 percentage point increase in the likelihood that a president elected on a state-oriented campaign switches to a neoliberal program in office, significant at a lower than 1% level (Model 5). This effect can be better appreciated in Figure 2(a).
Effect of explanatory factors on policy switches.
According to these results, if President Hugo Chávez had been elected in 1998 subject to the exact same institutional conditions but under the economic scenario observed in his 2006 election, his chances of switching to a market-oriented economic program would have been 43 percentage points lower, with 95% confidence. Peruvian president Alberto Fujimori, if inaugurated in 1990 under the circumstances of Alan Garcia in 2006, would present 82 percentage points lower chances of switching, with 95% confidence.
Interestingly, booms do not seem to reduce the likelihood of switches from state-oriented to market-oriented programs; effects never reach acceptable levels of statistical significance. This finding suggests that the absence of crises (“normal” and “boom” periods) is sufficient for these presidents to pursue their original agenda. 7
Table 1 also shows the effects of political variables on the chances of switches, which can be further observed in Figure 2. It reveals that a one standard deviation change around the mean in the variable Executive increases by 62 percentage points the chances that a left-wing president switches to a neoliberal program, significant at a 5% level (Model 5). I find no evidence that incumbent party’s presence in congress (Legislature) affects the likelihood of switches. The coefficient for party age (Party) is negative and significant, confirming that it is harder for a president from an institutionalized party to switch.
Surprisingly, Volatility presents a negative and significant impact on the likelihood of switches from state-oriented to a market-oriented program, suggesting that policy switches are more likely to occur in institutionalized party systems. Electoral volatility could be potentially capturing the negative effects of party system fragmentation on policy switches, but the low correlation between these two variables in the sample (0.14) and the fact that the inclusion of fragmentation in the model does not produce any change neither in the effect of volatility on the likelihood of switches nor in the probability of switches themselves suggests this is not the case.
It is also possible that voters punish switchers more strictly than nonswitchers for bad economic results (Stokes, 2001). If this is true, politicians might only dare to openly betray campaign promises when they know they can count on voters’ loyalty—consistent parties would have more legitimacy to switch, in a sort of “Nixon goes to China” effect.
It is important to note, however, that Volatility should capture the effects of party systems institutionalization only when the stability of electoral competition is associated with the development of parties’ stable and programmatic roots in society. Whenever this association does not hold, volatility might not be an appropriate measure of party system institutionalization. In the absence of stable programmatic linkages between parties and electorate, which is likely the case in Latin America, it should be less surprising that volatility has not the expected effect on presidents’ likelihood to switch.
In a final effort to verify the robustness of results, I performed a casewise deletion in the sample, still based on Model 5 of Table 1. Figure 3 reveals that coefficients are resistant to the deletion of each election.
Robustness check - casewise deletion.
Conclusion
This article examined the relationship between dollar scarcity/abundance and electoral accountability in Latin America. It verified whether and how currency crises and booms occurring during inauguration affected the likelihood that presidents of different ideological leanings launched their original program or, instead, switched to a different agenda.
The main interest of the analysis was to investigate whether the need to attract financial capital during currency crises (and the lack thereof during booms) affects governments’ capacity to advance state-oriented policies. I also examined the claim that, as market-oriented presidents are not subject to the same trade-off between votes and capital incurred by their leftist counterparts, their programmatic choices should not be affected by either booms or crises. Results show that state-oriented presidents inaugurated in the midst of severe currency pressures, in less institutionalized systems in which the executive retains strong powers, are the ones most likely to switch to a neoliberal program.
This finding has two major implications. First, it evidences that the main motivation to switch does not originate in the political arena itself—it is rather a response to economic conditions. If institutional factors, such as fierce opposition in the legislature or the lack of executive strength, forced presidents to switch to placate the opposition, switchers should be weak, rather than strong incumbents. Second, it highlights the predominant role that Latin American presidents have in policy making, due to their capacity to enact policies by decree.
According to the analysis, institutionalized parties impose a barrier to switches, whereas the chances of switches decrease with electoral volatility, a widely used proxy for party systems’ institutionalization. This suggests that even though low volatility likely reflects stability of interparty competition, if this competition is not based on programmatic preferences, it is unlikely to bind presidents’ policy choices. Confronted with the need to attract capital in times of crisis, institutionalized electoral competition seems insufficient to force presidents toward the implementation of the policies they were voted to advance. Under these circumstances, presidents from all ideological leanings tend to converge around market-oriented policies.
The empirical analysis pursued in this article advances a still recent agenda on developing countries’ responses to capital volatility and the effect of financial booms and crises on domestic policy making. Currency crises and capital flight are quite common during presidential elections in Latin America, especially when left-leaning candidates are expected to win the contest. In addition, as demonstrated in this article, these candidates have a high probability of betraying campaign promises and switching to neoliberal programs when elected in the midst of severe currency pressures. These two pieces of evidence reveal a potential mechanism through which investors “discipline” governments in Latin America; the fact that former left-leaning parties tend to increasingly campaign on market-oriented programs unfolds the long-term implications of this process for ideological convergence in the region. 8
Ultimately, this analysis contributes to explain the long-lasting persistence of neoliberalism in Latin America, despite the frustrating record observed in terms of economic growth and reduction of income inequality. It calls attention to the means through which financial liberalization, started in the 1970s, nurtures democratic systems in which free electoral competition does not necessarily lead to the free competition of programmatic alternatives—as described by Weyland (2004), systems in which the democratic practice becomes more sustainable but very limited in its quality.
Footnotes
Appendix
How Currency Crises and Booms Affect the Likelihood of State-Oriented to Market-Oriented Policy Switches, Dependent Variable: Switch, N = 32 elections.
| Variable | Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | Model 6 | Model 7 |
|---|---|---|---|---|---|---|---|
| Crisis | 0.21 | 0.21 | 0.32 | 0.31 | |||
| SE | 0.08 | 0.08 | 0.14 | 0.15 | |||
| p value | 0.01 | 0.01 | 0.02 | 0.03 | |||
| Crisis.pre | 0.22 | ||||||
| SE | 0.14 | ||||||
| p value | 0.13 | ||||||
| Crisis.post | 0.23 | ||||||
| SE | 0.15 | ||||||
| p value | 0.12 | ||||||
| Boom | −0.02 | 0.16 | 0.16 | ||||
| SE | 0.10 | 0.18 | 0.18 | ||||
| p value | 0.87 | 0.37 | 0.40 | ||||
| Boom.pre | 0.33 | ||||||
| SE | 0.35 | ||||||
| p value | 0.34 | ||||||
| Boom.post | 0.07 | ||||||
| SE | 0.20 | ||||||
| p value | 0.73 | ||||||
| Executive | 0.18 | 0.30 | 0.30 | 0.23 | 0.22 | ||
| SE | 0.12 | 0.15 | 0.15 | 0.14 | 0.12 | ||
| p value | 0.13 | 0.04 | 0.05 | 0.10 | 0.08 | ||
| Legislature | 0.01 | 0.02 | 0.02 | 0.02 | 0.02 | ||
| SE | 0.02 | 0.02 | 0.02 | 0.02 | 0.02 | ||
| p value | 0.42 | 0.28 | 0.35 | 0.34 | 0.46 | ||
| Party | −0.03 | −0.04 | −0.04 | −0.05 | −0.03 | ||
| SE | 0.01 | 0.02 | 0.02 | 0.02 | 0.01 | ||
| p value | 0.01 | 0.02 | 0.02 | 0.03 | 0.02 | ||
| Volatility | −0.09 | −0.12 | −0.13 | −0.11 | −0.11 | ||
| SE | 0.04 | 0.05 | 0.05 | 0.04 | 0.04 | ||
| p value | 0.01 | 0.01 | 0.01 | 0.03 | 0.01 | ||
| Inflation | 0.18 | 0.25 | 0.13 | ||||
| SE | 0.24 | 0.23 | 0.23 | ||||
| p value | 0.46 | 0.28 | 0.58 | ||||
| Intercept | −0.87 | −0.81 | 1.62 | −0.75 | −0.90 | 0.42 | 0.59 |
| SE | 0.49 | 0.61 | 1.14 | 1.64 | 1.67 | 1.40 | 1.40 |
| p value | 0.07 | 0.19 | 0.15 | 0.65 | 0.59 | 0.76 | 0.67 |
| Chi-squared | 7.58 | 7.61 | 14.99 | 21.96 | 22.56 | 18.63 | 19.19 |
| p value | 0.00 | 0.01 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| % Correct | 0.69 | 0.69 | 0.72 | 0.84 | 0.84 | 0.84 | 0.78 |
| Log Likelihood | −17.83 | −17.81 | −14.20 | −10.63 | −10.33 | −12.30 | −12.02 |
| N | 32 | 32 | 32 | 32 | 32 | 32 | 32 |
Acknowledgements
Thanks to Geoffrey Garrett, Ronald Rogowski, Barbara Geddes, Robert Brenner, James DeNardo, Cesar Zucco, Amaney Jamal, Carles Boix, Christina Davis, Kosuke Imai, Stephen Kaplan, Chris Achen and participants in the Princeton Comparative Politics Seminar and Princeton Junior Faculty Meeting for their valuable comments on previous versions of this paper. Thanks to David Leblang, for sharing his dataset on speculative attacks, and also to Clayton Mendonça for valuable research assistance.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.