Abstract
I argue that when presidents are able (or forced) to cobble together broad-based coalitions to win an absolute majority, their administrations are less likely (and less able) to violate human rights, in comparison to presidential administrations whose victories are the result of a narrow plurality. Consistent with this argument, I find cabinets comprised of a higher percentage of individuals from parties other than that of the president to be associated with greater government respect for human rights. Additionally, I find that in the years after a presidential election won by an absolute majority, states are more likely to experience an increase in government respect for human rights, in comparison to the years after a presidential election won by a mere plurality. Utilizing an original dataset of cabinet composition for 35 presidential democracies spanning from 2001 to 2011, this study concludes that it may prudent for non-majoritarian systems to consider adopting a mandatory majority rule so to encourage the types of conciliatory alliances that appear to promote high human rights respect.
Introduction
Examining the relationship between democratic presidential elections and government respect for human rights, Richards and Gelleny (2007: 517) find “presidential elections to be associated with lesser levels of respect.” They argue that their “findings appear to lend credence to the argument that the rigidity and winner-take-all structure of many presidential systems can indeed contribute to the abuse of human rights.” (Richards and Gelleny, 2007: 520). But are all presidential elections winner-take-all? While it is but a single office, what if the incoming president had opted (or been forced) to negotiate away some of spoils of the presidency in order win the presidency? If these negotiations dilute some of the power from the typically winner-take-all system, would said system still have the same effect on human rights?
Like Richards and Gelleny (2007), most scholars treat those presidents that were elected with an absolute majority as essentially the same as those that were elected by a simply plurality; however, I argue that the administrations of these two types of presidents should be treated differently. Essentially, I argue that when presidents are able (or forced) to cobble together broad-based coalitions to win an absolute majority, their administrations are less likely (and less able) to violate human rights, in comparison to presidential administrations whose victories are the result of a narrow plurality. In this article, I empirically test the possible effect of absolute majority administrations versus plurality administrations by estimating and analyzing two ordered probit models that utilize an original dataset of cabinet composition for 35 presidential democracies spanning from 2001 to 2011. I find cabinets comprised of a higher percentage of individuals from parties other than that of the president to be associated with greater government respect for human rights. Additionally, I find that in the years after a presidential election won by an absolute majority, states are more likely to experience an increase in government respect for human rights, in comparison to the years after a presidential election won by a mere plurality.
Theory
In Linz’s seminal critique of presidential perils (Linz, 1990: 56), he argues that “[p]residentialism is ineluctably problematic because it operates according to the rule of ‘winner-take-all’—an arrangement that tends to make democratic politics a zero-sum game, with all the potential for conflict such games portend.” Agreeing with Linz, Horowitz (1990 :79) argues that “[h]e is right to worry about winner-take-all outcomes and their exclusionary consequences.” He goes on to add that “a directly elected president may think he has a popular ‘mandate,’ even if he has been elected with only a small plurality of the vote, perhaps even less than 40 %” however, in such cases when a president is elected with such a low level of support, “[t]he potential for conflict is accordingly enhanced.” In contrast to the dangers of presidential rule by plurality, Linz touts “the superior historical performance of parliamentary democracies” (Linz 1990: 52). Horowitz (1990: 79), this time with a rebuttal, suggests that Linz’s presidentialism-versus-parliamentarism “thesis boils down to an argument not against the presidency, but against plurality election, not in favour of parliamentary systems but in favour of parliamentary coalitions.” I argue that the parliamentary-style coalitions favored by Linz are more common in presidential democracies where the president was elected with more than 50% of the vote.
According to Bueno de Mesquita et al. (2003: 351), “[r]espect for human rights is common in systems with large winning coalitions,” while “oppression of political opponents is common in systems with small winning coalitions.” Davenport (2007: 57) adds that “if the repressive process were subject to Veto, then decision makers would hesitate to engage in the relevant behavior” because “as ‘transaction costs’ are increased by discussion, negotiation, compromise, and side payments, authorities…view repression less favorably.” In essence, when there are more people that can veto repressive behavior, repression is less likely to happen. Furthermore, when there are more people that can demand side payments in return for not otherwise vetoing (or exposing) repressive behavior, repression is also less likely to happen. I argue that a cabinet comprised of many individuals from parties other than that of the president resembles the parliamentary-style coalitions favored by Linz, as well as the systems with large winning coalitions that Bueno de Mesquita et al. (2003) expects to better respect human rights. In the case of the former, this is because a cabinet-member from a party other than the president is likely more willing to veto (or otherwise bring to light) repressive policies that the presidential administration may want to continue (or implement). In the case of the latter, a president elected with more than 50% of the vote likely achieved such a majority by comprising with parties beyond their political base. Such compromises can lead to cabinet-level appointments, which in turn increase the number of veto players.
For instance, prior to the second round of the 2017 French presidential election, the Democratic Movement’s (MoDem) leader, François Bayrou, endorsed Emmanuel Macron. Shortly after winning the second round, Bayrou was appointed as the Minister of Justice in Macron’s new cabinet. Similarly, after the first round of the 2010 Colombian presidential election, Germán Vargas Lleras of Radical Change (CR) came in third place behind Juan Manuel Santos of the Social Party of National Unity (Party of U). After the election, a victorious Santos appointed Lleras as his Minister of Justice. In both of these cases, the president’s post-election cabinet contained at least one member of an opposing party in return for that party’s support prior to the majoritarian runoff election. However, such deal-making is likely not only a characteristic of two-round elections. Equally plausible is that deals could be struck prior to the first (and sometimes only) round of a presidential election as candidates seek to increase their chances of winning the first (and sometimes only) round. Regardless, ministers of an opposing party who find themselves in cabinet-level positions within the new administration effectively wield a veto because if their president proposes a new policy (or the continuation of an old policy) that they disagree with, they can make the implementation of said policy much more onerous. While the president (if aware), could ask for a resignation, that itself could bring a high political cost.
In short, I argue that when presidents are able (or forced) to cobble together large broad-based coalitions in order to win a majority, their administrations are less likely (and less able) to violate human rights, in comparison to presidential administrations whose victories are a result of a narrow plurality (and who therefore have more freedom to fill appointments entirely with party loyalists). This leads me to my two hypotheses:
Years where the president won the previous election with an absolute majority are more likely to be associated with greater government respect for human rights versus years where the president won the election by a mere plurality.
Having a greater percentage of the cabinet comprised of individuals from parties other than that of the president is likely to be associated with greater government respect for human rights.
Research design
In order to test my two hypotheses, I have estimated two models, the first of which is nearly identical to that used by Richards and Gelleny (2007), albeit using a dataset that takes advantage of new data that has been collected over the last decade. For instance, for human rights data, I use Version 2014.04.14 of the Cingranelli–Richards (CIRI) Human Rights Dataset (Cingranelli, Richards and Clay, 2014), as opposed to Version 2005.10.12 (Cingranelli and Richards, 2005). This allows access to data through the year 2011. The specific data in this dataset that I make use of is the CIRI Physical Integrity Rights Index, which is an aggregated nine-point index of four ordinal indicators of government respect for physical integrity rights: the rights of all human beings to be protected from torture, summary execution, disappearance and political imprisonment. CIRI scores range from ‘0’ (no respect for any of the four physical integrity rights) to ‘8’ (full respect for all of them). 1
While Richards and Gelleny (2007) used the 1946–2000 version of the Democratic Electoral Systems Around the World dataset (Golder 2005), I make use of their more recent dataset (Bormann and Golder, 2013). Using this data, I have constructed one of my primary independent variables: president elected with a majority (versus by plurality). For this variable, every country-year where the president had been elected with more than 50% of the vote has been coded as ‘1’. For all country-years where the president had been elected with less than 50% of the vote, this variable is coded as ‘0’.
The remaining control variables used are all expanded versions of those used by Richards and Gelleny (2007). For domestic conflict, I use Version 4-2014a of the UCDP/PRIO Armed Conflict database (Pettersson and Wallensteen, 2014), as opposed to Version 2-2005 (Gleditsch et al., 2002). For population size and GDP per capita (which is the per capita value of a state’s gross domestic product), I use more recent World Bank (2015) data, and both of these variables have been logged. Finally, for executive constraint, which takes into account “the extent of institutional constraints on the decision-making powers of the chief executive,” I use the most recent version of Polity IV (Marshall et al., 2014: 62). The 35 democracies in my dataset include states from all geographic areas and all levels of economic development, all of which have directly elected their presidents in multicandidate, democratic elections. 2
Because CIRI scores are ordinal, an ordinary least squares regression would lead to biased inferences as distances between the categories of the dependent variable are not equal (McKelvey and Zavoina 1975). Instead, the ordered nature of CIRI lends itself to the use of ordered probit estimation, as was used by Richards and Gelleny (2007). Theoretically, a state’s previous year’s level of government respect for human rights could likely influence the following year’s level. Since at least Poe and Tate (1994), the norm within human rights scholarship has been to address serial correlation by including a lagged dependent variable. However, since CIRI data is both nonlinear and categorical, the inclusion of a lagged dependent variable to address serial correlation would have been inappropriate. As a result, Richard and Gelleny (2007) did not address serial correlation beyond the use of sandwich ‘robust’ standard errors. Since the publication of their piece, however, it has become common practice to address serial correlation when using CIRI data. As such, I follow a technique used by Hafner-Burton (2005), among others, to address serial correlation by including a series of binary variables for each category of the lagged dependent variable. Standard errors are clustered on country.
For my second model, my primary independent variable is percent of the cabinet from a party other than that of the president. This variable was constructed by using the Central Intelligence Agency’s (CIA) Chiefs of State and Cabinet Members of Foreign Governments directory. 3 The year 2001 is the first for which this directory has data, and for each year since 2001, this directory provides a monthly description of the cabinets for each country with which the United States maintains a diplomatic relationship. Using this data, I looked up who is in each cabinet for each country where data was available during the years 2001–2011 (as those are the only years that both the CIA directory and CIRI data are available). The reference month for each year was the same month as that of the last presidential election. My primary sources for identifying the political party affiliation for cabinet member were: Banks, Muller and Overstreet (2003, 2007, 2008, 2009); Banks, Muller, Overstreet and Isaccoff (2010, 2011, 2012); Lansford (2013); and finally Bértoa (2016). Once all the cabinets were compiled for countries with available data, my percent of the cabinet from a party other than that of the president variable was constructed by dividing the number of individuals in a party other than that of the president by the total number of cabinet-level positions in that country for that year. 4
Results
In Table 1, I present results of two ordered probit regressions that estimate government respect for human rights. Recall that my first hypothesis is that years where the president had won the previous election with an absolute majority are more likely to be associated with greater government respect for human rights versus years where the president had won the election by a mere plurality. Consistent with this hypothesis, Model 1 reports that my primary independent variable—president elected with a majority (versus by plurality)—is positively associated with government respect for human rights, and that this relationship is statistically significant at the 99% level. This suggests that compared to the reference category (which are years when the president was elected with less than 50% of the vote), the years when the president was elected with more than 50% of the vote are more likely to be associated with high government respect for human rights (and thus lower levels of repression).
Ordered probit estimates of government respect for human rights in 35 presidential democracies, 2001–2011.
p < 0.10, ** p < 0.05, *** p < 0.01. Figures in parentheses are robust standard errors clustered by country.
Recall that my second hypothesis is that having a greater percentage of the cabinet comprised of individuals from parties other than the president is likely to be associated with greater government respect for human rights. Consistent with this hypothesis, Model 2 reports that my primary independent variable—percent of the cabinet from a party other than that of the president —is positively associated with government respect for human rights, and that this relationship is statistically significant at the 95% level. Consistent with previous human rights scholarship, across both models all of the control variables are statistically significant and have signs pointing in the expected direction. For instance, executive constraint and GDP per capita are found to be positively associated with government respect for human rights, while population size and domestic conflict are both found to be negatively associated with government respect for human rights. Each of these variables is found to be statically significant at least at the 95% level in both models.
While identifying statistically significant regression coefficients is helpful on a superficial level, a more intuitive understanding of the substantive effects of these relationships can better be communicated through an analysis of predicted probabilities. In Table 2, I outline the parameters of three scenarios which I use to form the basis of my predicted probabilities analysis. The “Best” and “Worst” scenarios are based on the 5th and 95th percentiles of the control variables, while the “Average” scenario is based on each variable’s mean (rounded to the nearest integer if a variable is categorical). Essentially, the “Best” scenario represents a rich, less populous state that has a highly constrained presidency, a history of high government respect for human rights, and no domestic conflict. The “Worst” scenario represents a poor, highly populous state that has a powerful and relatively unconstrained presidency, a history of low government respect for human rights, and is experiencing some low-level domestic conflict. Finally, the “Average” scenario represents a state whose statistics are between the “Best” and “Worst” scenarios.
Summary statistics of scenarios used in predicted probabilities analysis.
Scores range from 1 to 7, the latter of which indicates the highest level of executive constraint.
Scores range from 0 to 2, the former of which indicates the lowest level of domestic conflict.
Scores range from 0 to 8, the latter of which indicates full respect.
In Figures 1 and 2, I present probabilities for a year-on-year change in government respect for human rights following changes in my primary independent variables. These probabilities were estimated using the Clarify software package (Tomz, Wittenberg, and King 2003) and help to provide substantive meaning to Model 1 and Model 2 from Table 1. For both figures, all control variables were held at the values designated in Table 2.
The probability of a year-over-year change in government respect for human rights when the president has been elected by a majority versus a plurality.
The probability of a year-over-year change in government respect for human rights based upon the percent of the cabinet that is from a party other than that of the president.
Starting with Figure 1, again recall that my first hypothesis is that years where the president had won the previous election with an absolute majority are more likely to be associated with greater government respect for human rights versus years where the president had won the election by a mere plurality. Congruent with this hypothesis, note that across all three scenarios, there is a higher probability of an increase in government respect for human rights when the president was elected with a majority than when the president was elected by plurality. Furthermore, note that across all three scenarios, there is a lower probability of a decrease in government respect for human rights when the president was elected with a majority versus by a plurality.
In Figure 2, I seek to further explore the effect that veto players within the cabinet may be having on government respect for human rights. Again, recall that my second hypothesis is that having a greater percentage of the cabinet comprised of individuals from parties other than the president is likely to be associated with greater government respect for human rights. For this figure, note that the values for all of my control variables are held at the “Average” scenario values listed in Table 2. For my primary independent variable (percent of the cabinet from a party other than that of the president), I have five lines that respectively represent 0%, 25%, 50%, 75%, and 100% of the cabinet being from a party other than that of the president. As you can see, when more of the cabinet is from a party other than that of the president, there is a higher probably that the given year will experience an increase in government respect for human rights. Additionally, when less of the cabinet is from a party other than the president, there is a higher probably the given year will experience a decrease in government respect for human rights. In essence, the relationship illustrated in Figure 2 provides a compelling explanation for the relationship illustrated in Figure 1: the formation of cabinets where there is a greater percentage of individuals from a party other than the president’s party promotes greater government respect for human rights, and since majoritarian elections theoretically promote the formation of cabinets where there is a greater percentage of individuals from a party other than the president’s party, years where the president was elected by a majority promote greater government respect for human rights.
Conclusion
Montesquieu—as translated by Cohler et al. (1989: 69)—argues that “constant experience shows us that every [individual] invested with power is apt to abuse it, and to carry [their] authority as far as it will go…To prevent abuse [therefore], it is necessary from the very nature of things that power should be a check to power.” If, after an election, a president is able to appoint party loyalists to all top positions of power, to paraphrase Montesquieu, power would not be checking power. However, when candidates make deals for support (whether by choice or necessity), and when these deals end up empowering members of opposing parties with cabinet-level positions, the administration’s power is checked with power.
In this article, I have sought to build upon Richards and Gelleny (2007). My findings partially confirm their observation that some presidential elections lead to lower levels of respect for human rights. My results, however, add nuance to our understanding of this relationship. While Richards and Gelleny (2007) treat all presidential systems as the same, I argue that there should be a distinction between those administrations where the president won with an absolute majority and those where the president won by a mere plurality. I argue that presidents elected by plurality have less of an incentive to win over (or work with) many of their competitors. When power is more concentrated, repression is more likely (and possible). In contrast, presidents elected with more than 50% of the vote often resort to pre-election maneuvering in order to obtain the majority they need to win the election. As result, majoritarian presidents are likely to have built a broad coalition that includes individuals from other parties. The original dataset utilized by this article finds cabinets comprised of a higher percentage of individuals from parties other than that of the president to be associated with greater government respect for human rights, and this finding augments the theoretical argument that majoritarian elections are more likely to lead to diverse cabinets, which further supports the argument that majoritarian elections indeed promote greater government respect for human rights.
Horowitz (1990: 79) argues that “democratizing societies need to think, and think hard, about electoral systems that foster conciliation and governmental systems that include rather than exclude.” He adds that “[p]rominent among innovations they might consider are presidents chosen by an electoral formula that maximizes the accommodation of contending political forces” Horowitz (1990: 79). Although broad-based coalitions may be formed in non-majoritarian systems, it may be prudent for non-majoritarian systems to consider adopting a mandatory majority rule so as to further encourage conciliatory alliances. As it turns out, one of presidentialism’s greatest perils is the plurality election.
Footnotes
Acknowledgements
I would like to thank Jada Chiuchiarelli for all the help throughout this project. Your constant companionship was an invaluable and irreplaceable resource. You will be sorely missed.
Declaration of conflicting interest
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.
Notes
Carnegie Corporation of New York Grant
This publication was made possible (in part) by a grant from the Carnegie Corporation of New York. The statements made and views expressed are solely the responsibility of the author.