How to make democracy self-enforcing after civil war: Enabling credible yet adaptable elite pacts

Abstract

While many recommend electoral democracy as a way to avoid or resolve civil conflict, the empirical record of electoral democracy as an alternative to civil conflict is decidedly mixed. We apply recent work from new organizational economics on the nature of elite pacts to add to both sides of the debate. On the one hand, we argue that we should be more pessimistic about the ability of existing electoral institutions to help rather than hurt the prospects for a stable peace. We argue that the new organizational economics reveals a design dilemma—a forced trade-off between the credible commitment to an elite pact in the short term and the adaptability of an elite pact in the long term—that plagues the most commonly considered alternatives. On the other hand, we tentatively argue for optimism if institutional designers work with criteria that explicitly take the dilemma into account. We propose novel design criteria that would allow a polity to address the design dilemma.

Keywords

Civil war institutional design new organizational economics post-conflict elections

Introduction

When attempting to resolve these conflicts, enforcement does matter, but only in the short term. If outside states truly expect these settlements to last over time, they must consider how the institutional parameters of any new government shape groups’ expectations about their future security and factor into decisions to fight or cooperate. Military force may be crucial for demobilization, but creative institutional design matters far more in the long run. (Barbara Walter, 2002: 168)¹

Over the last 30 years the vast majority of battle deaths—whether combatant or civilian—have been caused by civil wars, not international wars.² Given the high costs of war, it is puzzling that would-be belligerents cannot strike a bargain to avoid violent conflict (Fearon, 1995). This is especially true in the case of civil war, where participants have democracy as a potential mechanism for conflict resolution.³

The empirical record of democracy as a peaceful substitute for civil war, however, is not encouraging (Gurses and Mason, 2008; Lyons, 2002; Wantchekon and Neeman, 2002). The “elections as conflict resolution” hypothesis has been undermined by violence triggered by post-conflict elections in polities ranging from Iraq to Sierra Leone to Rwanda. Recent scholarship has debated whether post-conflict elections usher in prolonged periods of peace or contribute to further instability. At the highest level, some urge cautious optimism (e.g. López-Pintor, 2005), while others exhibit reluctant pessimism (e.g. Paris, 2004). For most scholars, the question is not “are post-conflict elections helpful?” but rather “when, where, and why” are they more or less helpful (Flores and Nooruddin, 2012; Matanock, 2012)? Others emphasize that the “what” matters a lot too; both what types of electoral institutions (Hartzell and Hoddie, 2007; Hyde and Marinov, 2014; Walter, 2002) as well as what complementary institutions—in particular, constraints on the executive (Brancati and Snyder, 2011; Flores and Nooruddin, 2012; Strasheim and Fjelde, 2012)—are already in place.

We attempt to add to both sides of the debate. To do so, we draw on the new organizational economics (Moe, 2005; Williamson, 1998), especially recent applications explaining “limited access orders” in the shadow of violent conflict (Bates, 2008; Bednar, 2009; North et al., 2009, 2013). On the one hand, we argue that we should be more pessimistic about the ability of existing electoral institutions to help rather than hurt the prospects for a stable peace. We argue that the new organizational economics reveals a design dilemma—a forced trade-off between the credible commitment to an elite pact in the short term and the adaptability of an elite pact in the long term—that plagues the most commonly considered alternatives. On the other hand, we tentatively argue for optimism if institutional designers work with criteria that explicitly take the dilemma into account. Specifically, rather than taking complementary institutions—in particular, constraints on the executive—as an exogenous factor that allows elections to work better, we define design criteria for elections that would make the complementary institutions work better, which would in turn allow elections to work better. This virtuous cycle would make it possible for electoral politics to escape the shadow of violence.

The next section discusses elite pacts and the commitment/adaptability dilemma. The third section discusses conditions required to escape it. The fourth section discusses novel design criteria that create those conditions. The fifth section discusses a hypothetical application to the Democratic Republic of the Congo. The sixth section discusses a path to develop and test the design criteria, and concludes.

Civil conflict, elite pacts and electoral democracy

Constitutional design faces a dilemma: the federation needs sufficient structural integrity—solving the compliance problem—to work in the short run, but the rules upheld must adapt to changing needs. (Jenna Bednar, 2009: 3)

For citizens in democratic countries in the developed world, it is often a puzzle why other countries cannot simply make stable democracy work. What prevents electoral democracy from fulfilling its promise as a peaceful alternative to civil conflict? What prevents polities from using ballots—not bullets—to decide who should rule? Of course, it is the developed, democratic countries that are the outliers in this regard.

The much more historically commonplace “solution” to civil conflict is an elite pact on a mutually satisfactory allocation of spoils, including property rights, monopoly rights, tax revenues, import licenses, social privileges, political offices and government contracts (Bates, 2008; Hardin, 1999; North et al., 2009).⁴ It is important to understand this solution in its generic form before we introduce the specifics related to electoral democracy.

Elite pacts and the commitment/adaptability dilemma

Any elite pact to end civil conflict must overcome a design dilemma. The design dilemma consists of the interaction of a short-term challenge—enabling credible commitments to a workable agreement—and a long-term challenge—enabling adaptation of that agreement over time (Bednar, 2009). Requirements for overcoming the short-term challenge (i.e. allotting vetoes over changes to an agreement for the major players and then agreeing on streams of benefits that will flow to each in perpetuity) are in tension with requirements for overcoming the long-term challenge (i.e. enabling flexible responses to shifting circumstances, and therefore making only short-term streams of benefits credible).

In the short term, the primary challenge is enabling credible commitments to the peace reflected in the pact. It is understood that dissatisfied elite segments have the power to exercise their “outside option” and reignite civil conflict. To make commitments credible, benefits over and above what would otherwise be possible are created by limiting access to valuable resources (e.g. land, labor and capital) or activities (e.g. trade, education, politics, courts and education). To make the pact mutually satisfactory, the distribution of payoffs to each segment must be balanced to be roughly proportionate to their capacity for violence.⁵ This “limited access order” is so commonplace that North et al. (2009) call it the “natural state”.⁶

In the long term, the primary challenge is adaptability. Over time, both sides of the balance—the value of particular privileges and the capacity for violence—shift for each segment of elites (see Table 1). This makes the status quo agreement inappropriate or obsolete, in particular to rising elites whose privileges pale in comparison to their de facto power, or to established elites who find the value of their privileges collapsing owing to exogenous trends or shocks (Powell, 2012). As an example of the former, the “2000 families” of Bengali elites expropriated and then controlled key assets in Bangladesh after the partition that had been controlled by the “22 families” of Pakistan from 1947 to 1971 (Khan, 2013). As an example of the latter, recessions in developed countries in the early 1980s led to a fall in taxes collected on trade, which had been ∼40% of public revenues for many countries across Sub-Saharan Africa. This contributed to civil conflict in Sierra Leone, Uganda and Zaire (Bates, 2008).

Table 1.

Sources of fluctuation in the value of spoils and the capacity for violence

	Elements	Sources of fluctuation
Value of spoils	Rights to control:	Expected GDP
	Land	Expected market prices
	Tax revenue	Resource depletion
	Resource extraction rights	Expected payments from corruption
	Monopoly privileges	Shifts in donor agenda
	Political appointments
	Government contracts
	Aid revenues
Capacity for violence	Size of coalition	Demographic change
	Organization of coalition	Public opinion change
	Technologies for violence	Change in social technologies (i.e. internet, social media)
	External allies	Change in coalition leadership (and with it personal charisma and social networks)

Despite the threat of civil conflict, renegotiation is difficult. First, asymmetric information regarding power and payoffs—both relative capacity for violence as well as the value of various privileges—makes it hard to calibrate who should be newly included in or excluded from the dominant coalition, as well as how benefits should be allocated within the coalition.⁷ Second, even if information about the relevant participants, their capabilities and the available spoils were available and shared by all, participants face a coordination problem. A variety of potentially available privileges can be divided or bundled in a number of ways and participants each prefer arrangements biased in their favor.

If renegotiation fails, civil conflict ensues. This can take the form of outright civil war (e.g. Biafra against the rest of Nigeria, Bangladesh against the rest of Pakistan), ethnic violence (e.g. Kenya’s Rift Valley in 2007, Hindu–Muslim violence in India) or coups to prevent particular policies of democratically elected governments (e.g. Chile in 1973, Egypt in 2013).

We can illustrate this dilemma more precisely with standard concepts from game theory. In the short term, the limited access order has the structure of a divisible prisoners’ dilemma game with asymmetric information. Such situations involve multiple cooperative equilibria to align on ex ante (like a coordination game), opportunities to cheat ex post (like a prisoners’ dilemma game), and asymmetric information about one another’s capacity for violence (which creates incentives for players to engage in costly “signaling” of their strength).⁸ In the long term, elites must adaptively renegotiate to accommodate ongoing fluctuations in (a) the value of spoils and (b) the capacity for violence, with some participants keen to signal that the status quo is no longer acceptable, and others keen to hold on to traditional entitlements or recent gains. Shifts in the value of spoils and the capacity for violence make it a dynamic game rather than merely repeated game. Only shorter cycle conditional cooperation is possible; all-or-nothing grim triggers are off the table.

To illustrate the logic, the design dilemma consists of the fact that (a) it may take X years for a pact to pay off once it is set in motion, (b) the negotiation or renegotiation of complex elite pacts may take Y years once the need to do so is triggered by shifting circumstances, and (c) shifting circumstances of the relevant magnitude may happen every Z years. For a particular situation, if elites expected that Z < X+Y, then it will be difficult to get agreement on a pact; if a pact is reached, but further down the road it seems that Z < X, then the agreement could unravel even before the shock appears. How many years is Z? As a reference point, consider Figure 1. The median number of years between violent regime change in the poorest half of the world’s countries is seven years, compared with 60 years in the richest decile of countries (Cox et al., 2013). Even the richest developing countries (i.e. those in the 75–90th percentile of per capita gross domestic product (GDP) overall) only have a median regime duration of 12.5 years, which is much closer to other developing countries than it is to the developed countries. The expectation of shifting circumstances outside the control of elites creates an expectation of a need for renegotiation, and this shortens the time available to set in motion an agreement and for it to pay for itself once it is set in motion (Fearon, 1998; Powell, 1999).

Figure 1.

Regime duration as a function of GDP.

Electoral democracy and the tension between commitment and adaptability

While elections are often proposed as non-violent alternatives to violent regime change, the prospect of elections can actually increase the likelihood of violent civil conflict. This happens when de jure electoral results would not or do not reflect the de facto distribution of the capacity for violence. There are two primary sources of disconnect between de jure and de facto power: over- and under-reaction to shifts in electoral support.

Those with few veto gates make possible dramatic swings in outcomes based on small swings in votes.⁹ This “over-reaction” can undermine the elite pact. If the losers do not like the result, they may resist violently. For example, in Pakistan in 1970 the Awami League, a Bengali nationalist party, won a majority of seats yet was denied its victory by the President of Pakistan. That denial sparked violent conflict and the ultimate secession of Bangladesh from Pakistan.¹⁰

Electoral systems with more veto gates—typically power-sharing systems—allow the short-term stability necessary to enable credible commitments, but run into trouble if the value of spoils or the violent capability of potential spoilers shifts too much. This “under-reaction” can undermine the elite pact as well. As Roeder and Rothchild (2005: 41) note, “among the 16 experiments with power-sharing institutions cited by Horowitz (1985) in Ethnic Groups in Conflict, 12 … were subsequently discarded”.¹¹

The case of Lebanon is tragically instructive, demonstrating the dangers of demographic change undermining an agreement. An elite power-sharing agreement signed in 1943 established an inflexible formula for political representation of relevant groups. As the Muslim community grew, grievances over Christian political representation heightened. Intragroup rivalries over leadership led elites to renege on intergroup commitments, contributing to an ultimate resumption of the conflict in the 1970s (Deeb and Deeb, 1995: 125; Roeder and Rothchild, 2005: 40). Power-sharing failures in Ethiopia, Iraq, Nigeria and the Sudan can be attributed in part to changing circumstances that required an agile reallocation of spoils to potential spoilers.

Escaping the dilemma: pulling adaptability with opportunity rather than pushing it with civil conflict

To escape the dilemma, elites must multiply the benefits of peace so that they prefer peace even when their share of the capacity for violence is large relative to their share of the gains. To multiply the benefits of peace, elites must find or create shared opportunities before the need for renegotiation is present. This means adaptability must be “pulled” by opportunities rather than just “pushed” by the threat of civil conflict. While there are cases where opportunities arise exogenously—e.g. responding to external threats—one way to find or create opportunities endogenously is to foster the “rule of law for elites” (North et al., 2009). The rule of law for elites consists of standardizing certain public roles (e.g. the duke of X, the minister of Y) for members of the dominant coalition: what actions are required, permitted and forbidden within the role, as well as how one enters into and/or exits from the role. Before we consider the rule of law for elites in particular, we should understand why multiplying the benefits of the peace helps at all.

Consider an arithmetic example. First, suppose political and economic institutions do not tend to multiply the benefits of peace much, so US$1 worth of decision rights after conflict—which we stipulate to be the baseline—is worth only US$1.50 in peace (i.e. it is still worth more than in conflict as a result of the demand, infrastructure and security available during peace). Suppose there is an elite segment that only receives 40% of the benefits during peace—that is, US$ 0.60 (i.e. 40% of US$1.50)—but expects a 50% chance of winning a conflict, which would pay only US$ 0.50 for each baseline dollar (i.e. 50% of US$1.00). The proportionality of their payoff to their power is enough that they will still prefer peace. If, however, that segment suddenly sees its capacity for violence shift (e.g. because the martyrdom of a leader has temporarily mobilized domestic and international supporters), so that the probability of winning goes from 50 to 80%, then the expected payoff of going to war is US$ 0.80 per baseline dollar. If their share cannot be rebalanced to reflect the sudden shift in their probability of winning a conflict—rather than 40%, it would have to be a minimum of ∼54%—then they will opt for conflict.

Second, suppose political and economic institutions do tend to multiply the benefits of peace, so that US$1 worth of decision rights after conflict is worth US$3 (e.g. because of reduced uncertainty in certain policy domains, reduced investments required to defend rights, increased specialization or investment). If that same segment—which now receives US$1.20 (i.e. 40% of US$3.00) per baseline dollar—experiences the same shock as before, with their probability of winning a conflict shifting from 50 to 80%, then they will nonetheless prefer peace to war, because war would reduce their expected payoff to US$ 0.80 per baseline dollar.

The more general case is illuminating. Suppose that p is a segment’s probability of winning a conflict and controlling all the benefits, that s is the expected share of benefits during peace, and that b (where b ≥ 1) is the multiplier of benefits during peace relative to conflict. In this case, a segment will prefer peace as long as p/s < b. When b = 1, so that peace does not multiple the benefits at all, then the only question is whether one’s current share of spoils is commensurate to one’s expected probability of winning everything. When p shifts, then s must be reactively renegotiated as well. So long as p has not shifted, if s shifts then a new mix of spoils must be agreed to that brings s back in line with p. If the rebalancing fails, conflict breaks out. Notably, when 1/s < b, then the benefits of peace are large enough to keep peace even for a segment that would win a conflict with certainty. The key to escaping the need to reactively rebalance benefits is multiplying the benefits of peace.

Clearly, exogenous factors can amplify or dampen the value of the benefits of peace. If an external threat arises (e.g. the threat of a unified Germany to the feuding monarchists and republicanists in France, or the threat of Britain to the feuding states under the Articles of the Confederation), then the benefits of peace will be greater. A divided country is more readily conquered. Conversely, the lack of external threats makes internal strife more acceptable. Bates (2001) proposes that the perverse consequence of enforcing existing national boundaries—but not taking sides in civil conflicts—is to provoke more civil conflicts.

What we seek, however, is a pathway to multiply the benefits that can be chosen endogenously. One way to do this is to choose a set of policies that create a rapid postconﬂict economic recovery (Bigombe et al., 2000; Collier et al., 2003: 152–153; Flores and Nooruddin, 2009). This can be difficult when policies that are pro-growth put the payoffs for key elites at risk (Acemoglu and Robinson, 2000). Another commonplace way to multiply the benefits of the peace is to create the “rule of law for elites” (North et al., 2009, 2013). A segment of elites prefers the rule of law when the benefit of regulations on the behavior of others outweighs the cost of regulations on their own behavior. Critically, this must begin with impersonal standards around senior political leaders, then extend incrementally to other elites, and from there to the broader populace. If the rule of law does not cascade in this way, then executives can encroach on the standards meant to protect those below them, and those below them can encroach on one another by bypassing the standards by appealing up the hierarchy to a patron with latitude to act outside the law. Thus “constraints on the executive” enable the “rule of law for elites”, which is one of the necessary conditions to allow a limited access order to transition to an open access order, where economic and political competition is not destructively destabilizing (North et al., 2013).

For many polities, however, the starting place is reactively renegotiated pacts that fit special circumstances and personalities (Higley and Gunther, 1992; Linz and Stepan, 1996). The question is how to find or create opportunities to agree on constraints on the executive, abstract and impersonal rules of the game that govern interactions of elected leaders, special circumstances and personalities aside.¹²

The next section proposes novel design criteria for political institutions that can pull elected executives into creating constraints on the executive, then the rule of law for elites, thereby multiplying the benefits of peace.

Design criteria for multiplying the benefits of the peace

A key challenge of institutional design … is to create an environment in which cooperative interaction and mutually beneficial “win–win” exchanges are possible, so that norms of cooperation and negotiation can become habituated amongst political actors. (Reilly, 2001: 7)

In this section, we apply insights from the new organizational economics to articulate institutional design criteria that cumulatively address the commitment/adaptability dilemma, and do it in a way that is compatible with electoral democracy. That is, our design criteria need to jointly enable elite pacts and mass democracy. Specifically, on the elite pact side, we are looking to create a situation (a) where pacts are pulled by immediate opportunities rather than pushed by the threat of civil conflict and (b) where the role of elected executives is defined by increasingly impersonal standards. On the mass democracy side, we are looking to create a situation where (a) electoral results do not over-react to small differences in coalition size (as with many winner-take-all systems) and (b) nor do they under-react to large differences (as with many power-sharing systems). This section proposes three novel design criteria intended to cumulatively create these conditions.

Informal account of the three design criteria

The expectation-symmetry criterion requires that two presidential candidates with roughly equal support be given roughly equal expectations of being “in” and “out” of office within the existing term by taking alternate turns in office several times before the next election. As an example, if two candidates go to a run-off and split the vote 51 to 49, then they would take alternate years in office over a four-year term, with the highest vote-getter taking the first turn. On the elite pact side, this will create immediate opportunities to coordinate on the rules of the game governing their behavior, which then cascade down, restraining how the office is used to advantage other elites or how they advantage themselves. On the mass democracy side, this keeps electoral results from over-reacting to small differences in coalition size; as long as two sides are roughly equal, they expect to win roughly equal time in office in the immediate future. This lowers the incentive for both sides to target benefits at potentially pivotal segments to maximize their short-term electoral advantage.

The phase-substitutability criterion requires that periodic elections (a) recognize the outcome of elite competition if sufficient consensus exists, or else (b) create a situation conducive to elite pact-formation if sufficient consensus does not exist. As an example, the threshold may be set at 55% in a run-off; a candidate with “sufficient consensus” wins the whole term, but if no candidate meets the threshold, the turn-taking phase kicks in. On the mass democracy side, this keeps the electoral results from over-reacting to small differences in coalition size as well as from under-reacting to large differences. On the elite pact side, this pushes office-holders to balance both elite and mass appeal. Any party that is too narrowly self-serving will be vulnerable to mass accountability, but so long as the electorate is divided, reciprocal accountability between elites is triggered.

The two-stage decisiveness criterion requires the periodic election to both guarantee that there will be a winner in the electoral stage and that there will be one decisive office-holder at any given time in the policy-making stage. The process for filling the office and the structure for making decisions within it must each rule out formal deadlock (i.e. yielding no winner in the first place, or yielding no executive choice in the second). The examples proposed above would meet this criterion insofar as there is either one officeholder the whole term, or else one decisive officeholder at a time, taking turns throughout the term. Both officeholders should be present, consulted and informed about decisions throughout the term, but only one is decisive at a given time. On the elite pact side, this limits the bargaining required to adapt the rules of the game, allowing for much more agile decision-making than existing power-sharing alternatives. On the mass democracy side, this limits bargaining and uncertainty around coalition-formation.

Notably, these three criteria do not appear anywhere else in the institutional design literature (e.g. see Reynolds, 2002 for an overview). Our example of a mechanism that meets the three criteria is a “turn-taking institution”, in which the whole executive term (e.g. four years) goes to a sufficiently inclusive supermajority coalition (e.g. 55% in a run-off); if no coalition qualifies, major coalitions take short, alternating turns (e.g. alternate years) several times around before the next election. There are other specifications that meet all three criteria: we could vary the term length, the supermajority threshold, and the number and length of turns per term in the turn-taking phase. Clearly, a turn-taking institution with a six-year term, a 60% supermajority threshold and a turn-taking phase with 12 six-month turns would be different than one with four years, 55%, and a turn-taking phase with 4 one-year turns. For now, we believe these differences are small relative to the difference between these and existing mechanisms.

Even with our four-year, 55%, alternate year example, there are institutional design questions that we leave deliberately unspecified, both for elections and once in office. When it comes to elections, there are many ways to specify candidate nomination processes, ballot type, the role of third-party observers, etc. When it comes to the office itself, there are many ways to specify the process and structure through which ministry or cabinet appointments are made, the policy agenda crafted and daily operations run. All of these specifications would matter substantively. For now, we believe that these differences are small relative to the differences between these and other criteria.

There are mechanisms that meet some but not all of the three criteria. One could meet expectation-symmetry and two-stage decisiveness with a turn-taking rule that always requires alternation between the top two candidates for the duration of the term, regardless of the degree of consensus on the front-runner. This is not unlike the two consuls in the Roman Republic who took month-long turns presiding over the Senate. One could meet expectation-symmetry and phase-substitutability by setting a 55% threshold, but making alternation in the turn-taking phase a matter of who is in a “proposer” role and who in the “accept/reject” role for each turn during the term; in this case, there would be no single decisive executive in office.¹³ One could meet phase-substitutability and two-stage decisiveness by electing the majority winner in either case, but using 55% as the threshold to determine whether some other conflict resolution mechanism—i.e. other than turn-taking—should kick in to induce moderation and cooperation when the electorate is divided. Some might propose a “divide and choose”-like mechanism for allotting cabinet ministries (Brams and Kaplan, 2004), more deliberative democracy to inform the executive (Fishkin, 2009), more delegation to lower levels of government (de Figueredo and Weingast, 2005), and so forth.

As these examples are meant to illustrate, we believe the three criteria together are jointly sufficient to (a) address the commitment/adaptability dilemma for elite pacts and (b) do so in a way that is consistent with electoral democracy.

On the elite pact side, we can use the simple model from above. Regular turn-taking shortens X, the years it takes for elite pacts to pay off once set in motion. If an elite pact is working fine at its current scope, office-holders would have the opportunity to increase the scope. In most cases, acting with restraint for one turn would pay for itself once the other reciprocates, so X would be twice the length of a turn (i.e. one to two years). Each pair of turns would offer the opportunity to multiply the benefits of peace, either by fixing what was not working within the existing scope or else by broadening the scope. The scope of the pact would broaden from merely taking turns as prescribed (i.e. not grabbing on to power) to behaving a certain way while “in” (i.e. consulting and informing the other executive with regard to certain decisions) to regulating the behavior of other elites (i.e. requiring coalition-members to pay their taxes or respect the property rights of other elites). This creates the cascade from constraints on the executive to the rule of law for elites.

Meanwhile, the combination of symmetry and decisiveness shortens Y, the number of years it takes to coordinate or re-coordinate in the first place. Most changes could simply be initiated with consultation of the other executive, rather than requiring formal ratification. This would allow new rules to be explored and abandoned with less grandstanding up front or loss of face after the fact. This would reduce the renegotiation process from years to months. Reducing both X and Y makes it more likely that X+Y < Z, the number of years between major shifts in circumstances.

Finally, the elite pact is done in a way that acknowledges major shifts in the electorate without over-reacting to minor shifts. In effect, the turn-taking institution is a mix between a single-member district (when supermajority consensus exists) and a multi-member district (when it does not). Notably, the lower the district magnitude, M, the more pressure there is to consolidate into a few large parties to maintain electoral viability.¹⁴ On the spectrum, there is little attention to the possibility where M = 2 or 3; most alternatives are clustered at the extremes, where M = 1, or else M > 5. Taagepera (1984: 91–92) proposed that M = 2 systems could combine the best of both, as “the two-seat districts could, apparently, achieve everything the single-seat districts can for assuring stable two-party alternation—and they could reduce the possibility of partisan gerrymander and of one-party hegemony”. Since Taagepera wrote, Chile has offered a prominent example of M = 2, which has helped foster the incremental and mutually productive adjustments in policy that have made Chile peaceful and prosperous over the last 20 years, despite a legacy of civil conflict (Siavelis, 2006).

This balance puts us in between two major threads of the institutional design literature. To simplify, one side recommends preferential ballots and/or low magnitude districts (Horowitz, 1985; Meisburger, 2012; Reilly, 2001), while the other recommends proportional representation in higher magnitude districts (Carey and Reynolds, 2011; Lijphart, 2004). Relatively speaking, the result of the former is meant to be a smaller number of large, more moderate parties, while that of the latter is meant to be a greater number of small, more representative parties. Both sides argue that complementary institutions matter a lot as to how well electoral institutions will work. Our proposal is different in that (a) we steer a middle path in terms of district magnitude and (b) we focus on creating incentives for office-holders to form, enforce and adapt elite pacts to create those complementary institutions endogenously.

Formal account of expectation-symmetry

As discussed above, expectation-symmetry has the effect of (a) making it possible to form, enforce and adapt informal pacts on the impersonal standards that should regulate executive behavior as well as the behavior of other elites, while (b) reducing the stakes of small changes of opinion in the electorate when parties are roughly equal in size. Many would say that electoral democracies that exhibit turn-over in office already have something like “expectation-symmetry”, in that both parties expect to spend some time—perhaps equal time—in and out of office. As we have argued elsewhere, the level of time-discounting that people exhibit—even in stable, predictable environments—suggests that turn-over in the distant future (e.g. four, eight or twelve years out) has little to no influence on behavior. This is why we focus on equalizing expectations within the immediate term. It is helpful to formalize this further, as it gives us a way to assess where alternative institutions stand, both relative to the ideal (“equal expectations” of being in and out) and relative to one another.

We begin with the simple proposition that access to spoils and influence over policy and policy-making institutions is proportionate to time in office.¹⁵ Consider the simple case where we treat “years in office” as the unit. To minimize the stakes of small changes of opinion, the number of years that first and second place candidates expect to be in office should be equal, so that there is zero difference in years between first and second place.

To maximize the incentives that each side has to invest in impersonal standards, the first and second place candidates should be equal, so that the ratio of years in and years out is one. As long as the benefit of restraint to the “out” is greater than the cost of restraint to the “in”, both sides will prefer it. Conversely, a side that always expects to be “in” will see only the cost and never the benefit of regulation, and a side that always expects to be “out” will see only the benefits and never the costs of regulation. Equalizing expected years in office prevents either side from having a bias for/against regulations based on that expectation.

It is a happy coincidence that both targets are met when years in office are expected to be the same for first and second place candidates, or equivalently, when each side’s years “in” are equal to their years “out”. The “symmetry” is the equality both within a party across in/out roles, and across parties for the same role.

To assess how time-discounting, incumbency advantages and short commitment horizons affect the “symmetry” of any institution, we introduce several related concepts to measure the symmetry (or lack thereof) of expectations to control an office: the weighted win-sum, weighted loss-sum, weighted win-share and weighted loss-share. We refer to each coalition’s sum of expected, discounted years in a particular office as its weighted win-sum, $W_{t, H}^{k} (J)$ , and its weighted win-sum as a proportion of the total as its weighted win-share, $w_{t, H}^{k} (J)$ . The weighted win-sums and shares are a function of the discount factor, d, the incumbent party’s survival probability (i.e. their probability of reelection), p, the party’s ex post rank k, the institution J, the years until the next election t, and the horizon of years considered H. As complements, we also have the weighted loss-sum, $L_{t, H}^{k} (J)$ , the sum of discounted years that a coalition expects to lose, and its weighted loss-sum as a proportion of the total as its weighted loss-share, $l_{t, H}^{k} (J)$ .

To factor in effects from time-discounting, the advantages of moving first, and so forth, we can think of the office as a stream of payoffs per-year, weighted according to the discounting function. The simplest and most common model of time discounting is that of exponential discounting, so that is the one we will use here.¹⁶ Suppose that participants apply a discount factor, d, to each year out from the present, so that the first year is worth 1, the next d, the third $d^{2}$ , the fourth $d^{3}$ and so on.

We now translate our normative target into the language of weighted win-sums and win-shares. When $w_{t, H}^{1 st} (J) = w_{t, H}^{2 nd} (J) = 0.5$ , then stakes of winning, $w_{t, H}^{1 st} (J) - w_{t, H}^{2 nd} (J)$ , reduce to 0, and the ratio of time in to time out, $w_{t, H}^{1 st} (J) / w_{t, H}^{2 nd} (J)$ , reduces to 1. The per-year cost to citizen i when on the outside is c_i, and the per-year benefit to the same citizen is b_i. When the cost–benefit ratio of non-cooperation is sufficiently high, cooperation can be sustained as a subgame perfect Nash equilibrium:

\frac{c_{i}}{b_{i}} > \frac{w_{t, H} (J)}{l_{t, H} (J)}

(1)

The closer to 1 is the ratio of weighted win-share to loss-share, the more robust the ability to sustain cooperation as the cost–benefit of non-cooperation varies.

Figure 2 shows how closely the turn-taking institution stays to that target while d and p vary. The upshot is that the turn-taking institution provides incentives to elites to form, enforce and adapt mutually productive agreements, even when elites are impatient, when they have substantial incumbency advantages, and when their commitment horizon is limited. By contrast, the winner-take-all majoritarian institution is brittle. With empirically plausible parameters, that institution would not allow elites to form, enforce and adapt mutually productive agreements. Our account adds formal warrant for the skepticism of North (1990: 12–16), Robinson (1998) and Bates et al. (2002: 609) regarding the robustness of conditional cooperation in this context. In the majoritarian case, the binding constraint on behavior is not the prospect of reciprocity but rather the calculus of immediate political advantage.

Figure 2.

Weighted win-shares for the relevant future with turn-taking and winter-take-all institutions for different values of p, d and H after an election (t = 4).

Case and counterfactual: the Democratic Republic of the Congo

We discuss one application of the limited access order framework and our design criteria in the context of the Democratic Republic of the Congo (DRC) and its attempt to escape the Second Congo War (1998–2003), the deadliest war in modern African history.¹⁷ The Second Congo War began in 1998 after President Laurent-Desire Kabila of the newly named DRC (formerly Zaire) dismissed the Rwandan and Ugandan military advisors that—in large part—had enabled him to seize power from the corrupt Mobutu Sese Seko regime the previous year. As a result, Rwandan and Ugandan-backed rebel groups (the Rassemblement Congolais pour la Démocratie (RCD) and the Mouvement de Libération du Congo (MLC)) rose up and began to capture towns in the eastern regions of the DRC. At the request of Kabila, the governments of Namibia, Zimbabwe and Angola sent military aid, and helped to reverse some of the offensive gains of the rebels. In 2001, Laurent-Desire Kabila was assassinated by one of his bodyguards. His son, Joseph Kabila, took power as president. In 2002, the Intercongolese Dialogue organized peace talks between the new Kabila government, the armed and unarmed opposition, civil society and the Mai Mai community-based militias. A peace accord signed in December 2002 set up a transition government led by Kabila and four vice-presidents (one from the RCD, one from the MLC, one from the unarmed opposition, and one from the Kabila regime), with key ministerial positions distributed according to the strength of each component.

This transitional power-sharing arrangement successfully kept the peace. The need for constant negotiation, however, made it hard for the transition government to get traction on key issues like the reform of the army, the disarmament of armed groups, the elaboration of a new constitution, and the planning of elections. The peace held—according to Kaiser and Wolters (2013: 90)—because “all the players had a seat in government, and, therefore, an opportunity to access its resources … for most parties, and the individuals running them, this was a vast improvement on their previous circumstances, and there was therefore no incentive to rock the boat and risk losing a cushy position”. After several years of deliberation and negotiation, a constitutional referendum was held in 2005. With the help of the international community (and the United Nations peacekeeping mission in particular), presidential elections were held in 2006 (Gambino, 2011). Joseph Kabila’s L’Alliance pour la Majorité Présidentielle (AMP) had a plurality in the first round, but not a majority. In the run-off, Kabila won with 58% of the vote compared with 42% for Jean-Pierre Bemba’s MLC. Critically, the Parti Lumumbiste Unifie (PALU) aligned with Kabila in exchange for the prime ministerial position and several key ministries; had the PALU aligned with the MLC, the election may have swung to Bemba. After the election, Kabila awarded ministries to the parties within his AMP according to a formula adjusted to reflected their strength, and defined a developmental agenda focused on improving infrastructure, health and education, water and electricity, housing and employment. Broad-based progress in service delivery was not, however, made over the term. Kabila opted out of holding elections for governors or mayors, choosing them by appointment instead. In early 2011, Kabila amended the constitution so that no run-off would be required in 2011, publically justifying the change as a way to save the additional expense of a second round (Bafilemba, 2011). Kabila won the 2011 election with a plurality of 48% of the vote.

For the time being, the Kabila regime holds. Our framework would predict that the regime will continue to deliver benefits to elites in Kabila’s coalition, and fail to deliver broader services to the electorate—especially in the west—until an external shock to the relative power and privileges of opposition elites provokes renegotiation or, failing that, another civil conflict.

Counterfactual

Consider the counterfactual where a different set of electoral institutions are chosen during negotiations and ratified through the constitutional referendum. When it comes to electing the president, so long as no candidate has 60%, the top two candidates go to a run-off. In the run-off, the first-place candidate must have 60% to win a whole five-year term. If neither candidate qualifies, the first- and second-place candidates take six-month turns in office over the five-year term.

The 2006 election goes forward. Given the splintered primary results, neither side expects to win 60% of the vote. The PALU consider it less likely that Kabila will win outright, making a bargain with him less attractive; they end up splitting their votes, with 3 percentage points switching to Bemba. Even so, Kabila wins 55% and Bemba wins 45%. Kabila and Bemba take office, with Kabila in the decisive role for the first six months.

Neither trusts the other. They each consider the other’s entourage incompetent and inefficient. They are both wary as to whether the new agreement will stick; as a countermeasure, Kabila delays disarming in the east, and Bemba in the west. If either side fails to concede the other his turn, then both sides remain ready to return to war. To avoid turn-over in appointments, Kabila and Bemba agree on a formula to divide up the ministries. For certain key ministries, they agree to find a candidate who is jointly satisfactory. They agree to hold elections for governors and mayors.

As the term progresses, there are many occasions for mutual frustration. There are conflicts between their respective clients, a lack of coordination across ministries, a lack of delivery on development projects overseen by each. Fortunately, both sides are on the giving and receiving end in short order. This leads to them to reconsider the process and structure they use as part of the office. Informally, they agree to a norm of jointly reviewing the contents of the executive agenda each week, minimizing surprises and avoidable mistakes. An additional two days each week are spent adjudicating conflicts between their clients. In the process, they begin to develop a shared impression of what rules should govern elite property rights and privileges, as well as who is competent and who is not within their administrations and their broader client networks.

When the 2011 election comes, neither side mobilizes for violence. Both tone down promises to their own coalition, and the threats posed by the other. As a result, both face challengers within their coalition, but intra-party majorities on both sides see the alternation of the more extreme candidates as likely to be more problematic. In the general election, Kabila wins again, but by less. In the subsequent term, Kabila and Bemba appoint more competent members of their administrations to create judicial institutions to help resolve conflicts between their clients, so that they can spend less time on it themselves. They establish standards for how governors and mayors can campaign for election, and for regulating their behavior once they are elected.

As elections come and go, preferences for policy and policy-making processes are shaped by the fact that the choices they make will have relatively immediate and symmetric repercussions. Policy options that look attractive from the insider’s perspective look less so when they expect to be insiders and outsiders in equal measure in the near future. When leaders move to prune back the privileges that would otherwise flow to elites within their coalition, elites become less loyal to one side or the other, and more inclined to float between coalitions depending upon who can more credibly deliver better public policy.¹⁸ Over time, voters follow suit, shifting from voting based on ethnic patronage to a mix of ideological and retrospective voting.

Conclusion: limitations, omissions, contributions

We propose turn-taking as a better starting place for democratic governance after civil war when compared with existing alternatives, not as a definitive solution. We believe that the turn-taking institution has important limits. It would not help in a case where some elite segments are so disproportionately unpopular relative to their capacity for violence that any openly competitive electoral system would fail. We acknowledge that such empirical cases do exist. We acknowledge that elites who choose to act as “moderates” will have to manage intra-party dynamics from challengers who try to promise better outcomes. The prospects of turn-taking should limit what promises are credible, but this is a dynamic that requires further study. Turn-taking cannot guarantee that “cooperative” individuals will win office, that they will anticipate the consequences of their policy choices or that they will learn the right lesson when outcomes are different than anticipated. Even in the best case where the turn-taking institution successfully fosters the formation, enforcement and adaptation of elite pacts, all of the difficult technical work of identifying what rules work better or worse is left to participants. Even if we assume the theory to be sound for rational actors in the abstract, real participants need not act rationally.

Going forward, we intend to apply a multiple methods approach, working iteratively in cycles of hypothesis-generation and prototype-testing (Poteete et al., 2010), in increasingly realistic conditions. We intend to refine prototypes in the context of game-theoretic models, agent-based models, and laboratory experiments that would ultimately be tested in field experiments and pilots (Green and Gerber, 2003). Pilots are most likely to begin at the level of city or county politics; if these are successful, subsequent pilots could spread to regional or national levels. As Steven Brams (2006) and other academic institutional designers would attest, this process could take decades.¹⁹ This is especially true if we wish to enable local ownership and experimentation, to avoid the hubris of “democracy promoters” who often recommend a single institutional exemplar (e.g. mirroring the United States) without understanding the variety of alternatives or the complexity of transitions (Carothers, 2012).

We see this paper as making three contributions to the literature on institutional design and post-civil war democratization. First, by posing the commitment/adaptability dilemma, especially as it interacts with electoral democracy, we add to understanding of why it is difficult to get out and stay out of civil war.

Second, we frame the problem in a new way that suggests the criteria for its solution. Multiplying the benefits of peace in advance of a crisis allows escape from the shadow of violence. We can achieve this by creating conditions that generate mutually advantageous impersonal standards for executive behavior that cascade to bind other elites, and do so in a way that is compatible with electoral democracy.

Third, we offer a new electoral mechanism, a turn-taking institution, that satisfies these criteria even when elites are impatient, when incumbency advantages are substantial, and when commitment horizons are relatively short.

We consider this an example of “creative institutional design” in the sense of Walter (2002), quoted at the outset, but more research is required to test whether it deserves a place on the menu of electoral institutions available to sustain the peace.

Footnotes

Acknowledgements

The authors thank Anjali Dayal, Jesse Edgerton, Thomas Flores, Scott Gates, Meghan McConaughey, Ryan Oprea, Kail Padgitt, Jennifer Raymond Dresden, Erik Voeten, John Wallis, participants at the Georgetown University Comparative Government Working Group, as well as the editor and reviewers at Conflict Management and Peace Science for helpful comments. The opinions expressed in this article are solely those of the authors and do not necessarily reflect the views of those acknowledged here, of McKinsey & Company, or of Yale University.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Acemoglu

Robinson

(2000) Political losers as a barrier to economic development. American Economic Review 90(2): 126–130.

Babcock

Loewenstein

(1997) Explaining bargaining impasse: The role of self-serving biases. Journal of Economic Perspectives 11(1): 109–126.

Bafilemba

(2011) Congo’s Kabila revises constitution: Presidential candidates now only need a plurality. The Christian Science Monitor, 19 January 2011. Available from: http://bit.ly/1cRO6e4 (accessed 9 January 2014).

Bates

(2001) Violence and Prosperity. New York: WW Norton.

Bates

(2008) Things Fell Apart: State Failure in Late-Century Africa. Cambridge: Cambridge University Press.

Bates

Greif

Singh

(2002) Organizing violence. The Journal of Conflict Resolution 46(5): 599–628.

Bednar

(2009) The Robust Federation: Principles of Design. Cambridge: Cambridge University Press.

Bigombe

Collier

Sambanis

(2000) Policies for building post-conflict peace. Journal of African Economies 9(3): 323–348.

Blainey

(1988) The Causes of War, 3rd edn. New York: The Free Press.

10.

Brams

(2006) The normative turn in public choice. Presidential Address to Public Choice Society, New Orleans, LA. Available from: http://www.nyu.edu/gsas/dept/politics/faculty/brams/normative_turn.pdf

11.

Brams

Kaplan

(2004) Dividing the indivisible: Procedures for allocating cabinet ministries in a parliamentary system. Journal of Theoretical Politics 16(2): 143–173.

12.

Brancati

Snyder

(2011) Rushing to the polls: The causes of premature postconflict elections. Journal of Conflict Resolution 55(3): 469–492.

13.

Carothers

(2012) Aiding Democracy Abroad: the Learning Curve. Washington, DC: Carnegie Endowment for International Peace.

14.

Carey

Reynolds

(2011) Comparing the Arab revolts: The impact of election systems. Journal of Democracy 22: 36–47.

15.

Chhibber

Kollman

(1998) Party aggregation and the number of parties in India and the United States. American Political Science Review 92(2): 329–342.

16.

Chhibber

Kollman

(2004) The Formation of National Party Systems: Federalism and Party Competition in Canada, Great Britain, India, and the United States. Princeton, NJ: Princeton University Press.

17.

Collier

Elliott

Hegre

Hoeffler

Reynal-Querol

Sambanis

(2003) Breaking the Conflict Trap: Civil War and Development Policy. Washington, DC: The International Bank for Reconstruction and Development/The World Bank.

18.

Cox

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

19.

Cox

North

Weingast

(2013) The violence trap: A political–economic approach to the problems of development. Working paper. Center on Democracy, Development, and the Rule of Law, Stanford University.

20.

Deeb

M-J

Deeb

(1995) Internal negotiations in a centralist conflict: Lebanon. In: Zartman

(ed.) Elusive Peace: Negotiating an End to Civil Wars. Washington, DC: Brookings Institution, pp. 125–146.

21.

De Figueiredo

RJP

Weingast

(2005) Self-enforcing federalism. The Journal of Law, Economics, and Organization 21(1): 103–135.

22.

Durant

Weintraub

(2014) An institutional remedy for ethnic patronage politics. Journal of Theoretical Politics 26(1): 60–79.

23.

Fearon

(1995) Rationalist explanations for war. International Organization 49: 379–414.

24.

Fearon

(1998) Bargaining, enforcement and international cooperation. International Organization 52: 269–305.

25.

Fearon

(2004) Why do some civil wars last so much longer than others?Journal of Peace Research 41(3): 275–301.

26.

Fishkin

(2009) When the People Speak: Deliberative Democracy and Public Consultation. Oxford: Oxford University Press.

27.

Flores

Nooruddin

(2009) Democracy under the gun: Understanding postconflict economic recovery. Journal of Conflict Resolution 53(1): 3–29.

28.

Flores

Nooruddin

(2012) Electing peace: The effect of elections on post-conflict reconstruction. Journal of Politics 74(2): 558–570.

29.

Gambino

(2011) World Development Report 2011 background case study: Democratic Republic of the Congo. Available from: http://goo.gl/N5y9xi

30.

Green

Gerber

(2003) The underprovision of experiments in political science. Annals of the American Academy of Political and Social Science 589:94–112.

31.

Gurses

Mason

(2008) Democracy out of anarchy: The prospects for post-civil-war democracy. Social Science Quarterly 89(2): 315–336.

32.

Hardin

(1999) Liberalism, Constitutionalism and Democracy. Oxford: Oxford University Press.

33.

Hartzell

Hoddie

(2007) Crafting Peace: Power-Sharing Institutions and the Negotiated Settlement of Civil Wars. University Park, PA: The Pennsylvania State University.

34.

Heckathorn

Maser

(1987) Bargaining and constitutional contracts. American Journal of Political Science 31(1): 142–68.

35.

Higley

Gunther

(1992) Elites and Democratic Consolidation in Latin America and Southern Europe. New York: Cambridge University Press.

36.

Horowitz

(1985) Ethnic Groups in Conflict. Berkeley, CA: University of California Press.

37.

Hyde

Marinov

(2014) Information and self-enforcing democracy: The role of international election observation. International Organization (in press).

38.

Jervis

(1988) War and misperception. The Journal of Interdisciplinary History 18(4): 675–700.

39.

Kaiser

Wolters

(2013) Fragile states, elites, and rents in the Democratic Republic of Congo (DRC). In: North

Wallis

Webb

Weingast

(eds) In the Shadow of Violence: Politics, Economics, and the Problems of Development. Cambridge: Cambridge University Press, pp. 70–111.

40.

Khan

(2013) Bangladesh: Economic growth in a vulnerable LAO. In: North

Wallis

Webb

Weingast

(eds) In the Shadow of Violence. Cambridge: Cambridge University Press.

41.

Lacina

(2006) Explaining the severity of civil wars. Journal of Conflict Resolution 50(2): 276–289.

42.

Lacina

Gleditsch

(2005) Monitoring trends in global combat: A new dataset of battle deaths. European Journal of Population 2(3): 145–66.

43.

Lijphart

(2004) Constitutional design for divided societies. Journal of Democracy 15(2): 96–109.

44.

Linz

Stepan

(1996) Problems of Democratic Transition and Consolidation: Southern Europe, South America, and Post-Communist Europe. Baltimore, MD: Johns Hopkins University Press.

45.

López-Pintor

(2005) Postconflict elections and democratization: An experience review. USAID Bureau for Policy and Program Coordination. Available from: at http://pdf.usaid.gov/pdf_docs/pnadb897.pdf (accessed 9 January 2014).

46.

Lyons

(2002) The role of postsettlement elections. In: Stedman

Cousens

Rothchild

(eds) Ending Civil Wars: The Implementation of Peace Agreements. Boulder, CO: Lynne Rienner, pp. 215–235.

47.

Matanock

(2012) Bullets for ballots: Examining the effect of electoral participation on conflict recurrence. Working paper.

48.

Meisberger

(2012) Getting majoritarianism right. Journal of Democracy 23(1): 155–163.

49.

Moe

(2005) Power and political institutions. Perspectives on Politics 3(2): 215–233.

50.

North

(1990) Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press.

51.

North

Wallis

Weingast

(2009) Violence and Social Orders: A Conceptual Framework for Interpreting Recorded Human History. Cambridge: Cambridge University Press.

52.

North

Wallis

Webb

Weingast

(2013) In the Shadow of Violence: Politics, Economics, and the Problems of Development. Cambridge: Cambridge University Press.

53.

Paris

(2004) At War’s End: Building Peace After Civil Conflict. Cambridge: Cambridge University Press.

54.

Poteete

Janssen

Ostrom

(2010) Working Together: Collective Action, the Commons, and Multiple Methods in Practice. Princeton, NJ: Princeton University Press.

55.

Powell

(1999) In the Shadow of Power: States and Strategies in International Politics. Princeton, NJ: Princeton University Press.

56.

Powell

(2012) Persistent fighting and shifting power. American Journal of Political Science 56(3): 620–637.

57.

Reilly

(2001) Democracy in Divided Societies: Electoral Engineering for Conflict Management. Cambridge: Cambridge University Press.

58.

Reynolds

(ed.) (2002) The Architecture of Democracy: Constitutional Design, Conflict Management and Democracy. Oxford: Oxford University Press.

59.

Robinson

(1998) Theories of bad policy. Journal of Economic Policy Reform 2: 1–46.

60.

Roeder

Rothchild

(eds) (2005) Sustainable Peace: Power and Democracy after Civil Wars. Ithaca, NY: Cornell University Press.

61.

Schelling

(1960) Strategy of Conflict. Cambridge, MA: Harvard University Press.

62.

Siavelis

(2006) Accommodating informal institutions and Chilean democracy. In: Helmke

Levitsky

(eds) Informal Institutions and Democracy: Lessons from Latin America. Baltimore, MD: Johns Hopkins University Press.

63.

Slantchev

Tarar

(2011) Mutual optimism as a rationalist explanation of war. American Journal of Political Science 55(1): 135–148.

64.

Strasheim and Fjelde (2012) Pre-designing democracy: Institutional design of interim governments and democratization in 15 post-conflict societies. Democratization 1–24.

65.

Taagepera

(1984) The effect of district magnitude and properties of two-seat districts. In: Lijphart

Grofman

(eds) Choosing an Electoral System: Issues and Alternatives. New York: Praeger.

66.

Walter

(2002) Committing to Peace. Princeton, NJ: Princeton University Press.

67.

Wantchekon

Neeman

(2002) A theory of post-civil war democratization. Journal of Theoretical Politics 14: 439–464.

68.

Williamson

(1998) Transaction cost economics: How it works, where it is headed. The Economist 146(January): 23–58.