Introduction

Virtually everything we study in political science involves thinking about how individual preferences and/or actions reconcile with those of the group in which the individual is situated. Peace is not possible without both the government and the rebel group laying down their arms. A legislator needs the votes of other colleagues in order for his bill to pass. The budget cannot be passed without the support of the executive and legislature. A coup d’état cannot successfully displace the government unless various factions of the military coordinate their actions. As Pierson (2000: 258) observes: “In politics, the consequences of my actions are highly dependent upon the actions of others. What I get depends not just on what I do, but (mostly) on what others do.”

In this chapter, we begin by discussing the ways in which this dependence takes different forms. We provide a variety of examples from across the subfields that, in turn, highlight the fact that the salient problems confronting political actors in different substantive contexts are generic in nature. We then discuss how institutions play a role in these strategic dilemmas by either creating or addressing them (intentionally or not) and clarify the implications of thinking about institutions in this way. Understanding how institutions function leads us to consider their origins. We highlight three particular challenges that emerge in trying to understand institutional origins and that represent some of the areas in which our knowledge of institutions is thin. We conclude by emphasizing the generality of the issues faced in trying to understand the functioning and emergence of institutions, providing good reasons for scholars across the subfields to engage more with each other if they want to have a better understanding of institutions.

The problem of collective decision-making

One thing political actors—whether they are members of a legislature, a court, a junta, or a party—must do is make collective decisions. However, collective choice is wrought with many problems that can lead to undesirable outcomes. For example, Arrow’s Theorem teaches us that there exists no preference aggregation rule—such as simple majority—that simultaneously satisfies a small number of fundamental desiderata, except dictatorship. At the same time, the various chaos theora, especially those from McKelvey (1976) and Schofield (1977), show that collective choice in a multi-dimensional space suffers from perverse problems—especially cycling and intransitivity—even when actors’ preferences are individually rational. Scholars have pointed to political institutions as a way of understanding solutions to those troubling results.

Implicit in much of this analysis is that cycling is normatively undesirable. It results in policy instability, leading to inefficiencies in governing and unpredictability for citizens conditioning their behavior on government actions. In such situations, institutions—such as parties or legislative committees—can enable actors to escape this trap and arrive at stable collective decisions.

Another way of arriving at this point is to assume directly that political actors prefer most of all to cooperate (rather than not cooperate) with others. In the classic coordination problem, the critical difficulty lies in finding a focal point. The illustrative example that captures this dynamic is that of Bach vs. Stravinsky: I like Bach more than Stravinsky; you like Stravinsky more than Bach; but we both prefer to go to either concert together. Hardin (1989) argues that individuals within a society considering the organization of a polity may find themselves in a similar situation. They may have preferences for different kinds of constitutive rules, but they most prefer to come to some agreement on any set of rules. Constitutions provide these rules and serve as a focal point that enables us to avoid living in chaos or in a world in which foundational rules are constantly amended. Settling on some set of constitutive rules enables us to make other decisions and laws that lead to more efficient governance (Holmes 1995).

When actors are confronted with the task of preference aggregation or coordinating on outcomes, institutions can help them settle on policies or outcomes, which presumably is better than living with policy instability or chaos. But the efficiency-enhancing properties of institutions do not absolve them of their distributive implications (Knight 1992). Long coalitions in the form of political parties may reduce transaction costs in passing legislation, but they also enable majority parties to act as gate-keepers. Legislative committees organized by policy jurisdiction may help reduce cycling but also bestow agenda-setting power on select members of the assembly. Gate-keeping and agenda control exerted by select actors within a group often result in policy outcomes that deviate significantly from the median group member’s preference (e.g. Romer and Rosenthal 1978). In short, there generally exists a frontier of institutional efficiency. There are usually no perfect solutions to collective dilemmas, only different ways of balancing alternative tensions. What institutions do in this situation is create a mechanism to trade off one inefficiency against another. If one worries about policy instability and undue transactions costs, there exist institutional solutions to alleviate those problems, but they may entail accepting potentially undesirable outcomes, such as non-median policies.

The problem of interdependent action

Political actors do more than make collective decisions; they also take actions on behalf of or in conjunction with others. A variety of strategic dilemmas fit this characterization—collective action, delegation, commitment—with their canonical structures having been imported from economics. But as the substantive examples illustrate, the political contexts that we study usually require us to combine and complicate these frameworks.

The prisoners’ dilemma, the collective action problem, and the common pool resource problem share a specific feature: they are situations in which individual rationality is not sufficient for collective rationality. A cooperative outcome could make all participants better off, but individual incentives lead them to an inferior position (Sandler 1992).

The problem of enforcing the appropriate limits on state power provides such a dilemma for citizens (Weingast 1997). The state has the power to transgress against the rights of some citizens, and citizens must decide whether to acquiesce or challenge such transgressions. The twist is that the state can expropriate from some individuals and give a portion of those assets to others. The result is that there are non-targeted citizens who support the state in violating the rights of others even if citizens collectively would be better off if they were to resist all state transgressions. Because individuals have incentives to deviate from any collective challenge of such violations, the state cannot commit to not carrying them out.

Political parties in a coalition government face a problem with a similar structure, but in a very different context. Budget-making under coalition governments is a common pool resource problem. The coalition would be better off if it could adhere to limits on the aggregate budget, but each party, via its control over ministries, has an incentive to overspend for its own projects since it can claim credit with its own constituency for such spending and blame the other parties for “overspending.” Indeed, coalition governments typically engage in higher fiscal expenditures than single-party governments, and the amount of spending increases as more parties join the coalition (Bawn and Rosenbluth 2006). Some types of fiscal institutions may resolve this problem. When the formulation of the budget is centralized within the finance ministry or if floor amendments on budget bills are prohibited, for example, parties may not be able to “free ride.” In the presence of such restrictive budgetary institutions, government expenditures do not increase with each additional party in the coalition (Martin and Vanberg 2013).

Introducing the element of time creates its own set of problems for political actors. Their preferences may be time inconsistent because what might maximize their utility in an individual period may not be equivalent to what would maximize it if the whole time path were considered (Shepsle 1991). When there is a tension between consistency and optimality, two key issues emerge: first, verifiability—the one to whom the commitment is made must be able to verify the other actor’s upholding of the commitment; second, enforcement—there must be some mechanism by which the actor who made the commitment can be punished if he/she reneges.

The end of a civil or interstate conflict is often marked by belligerents signing a peace agreement that may entail promises to disarm, to cede territory, or to share power—all actions that must be taken in the future. The problem is that a bargain that is satisfactory for the parties now may not be in the future due to changes in the balance of power among actors (Fearon 1998) or the emergence of spoilers (Heger and Jung 2014). This time inconsistency incentivizes actors to renegotiate or renege on the agreement, making peace unlikely in the future and at the current time since parties may be unwilling to settle if these issues remain unresolved. One potential solution comes in the form of third party mediators that can verify whether actors are following through on the agreement and exact costly consequences for cheating (Walter 2002). By increasing the costs of renewed conflict, mediators may increase the likelihood of peace, but their effects in the long term are less clear (Beardsley 2008).

Similarly, in developing democracies, vote brokers may purchase votes for their candidates and parties. Before the election, they offer cash and in-kind benefits in exchange for a promise from the voter that she will support the candidate or party that provided these goods at the polls. Since the advent of the secret ballot, however, it can be difficult to insure that the voter actually honors his or her promise on election day. Brokers may target communities with strong social ties in order to verify individual voter behavior and use social sanctioning or the withholding of benefits as punishment (Stokes 2005; Magaloni 2006). The voters’ commitment problem, however, appears in tandem with a collective action problem. As Rueda (2013) observes, voters who want to receive bribes but also to vote for their preferred candidates have incentives to free ride on other voters’ compliance. Such free-riding is easier to get away with the larger the number of voters casting ballots at the same polling station. This may account for the inverse relationship between levels of aggregation of electoral results and vote buying (Chandra 2004; Schaffer and Schedler 2007; Birch 2011).

For actors seeking to make their commitments more credible, delegation is another solution. Rather than manipulate the consequences of actions to discourage reneging, actors seeking to make a credible commitment can simply place authority in the hands of others. For the King of England after the Glorious Revolution, the temptation to expropriate was avoided by delegating to parliament the power of the purse (North and Weingast 1989). Similarly, a government can desist from manipulating the economy by handing control over monetary policy to a central bank.

What is critical, of course, is that the actor to whom power has been delegated is, in fact, independent from the actor seeking to make his commitment credible. Yet delegation creates its own set of problems (Sánchez-Cuenca 1998; Miller 2005). The principal–agent framework highlights two broad classes of problems—those of hidden actions (moral hazard) and of hidden information (adverse selection). Whether the agent has superior information about his actions or his type, the principal experiences some degree of agency loss. If delegation occurs in an effort to tie hands, such agency loss may be evidence of the credibility of the commitment. If the principal delegates authority as a remedy for his own lack of time, expertise, or information, then he trades off this solution for some loss of control over the agent. The principal–agent framework is flexible enough to capture the various types of trade-offs political actors must consider and the conditions under which they occur (Gailmard forthcoming). As the following examples illustrate, this flexibility is critical since political actors frequently encounter situations in which agents possess more than just superior information about actions or types. Compliance by the agent cannot be taken for granted, and agents sometimes are in the position to force principals to comply with their decisions—both significant deviations from the standard principal–agent model.

Courts, for example, can use their rulings to resolve core problems related to policymaking and institutional prestige (Staton and Vanberg 2008). Justices want to achieve policy outcomes as close as possible to their ideal points. Yet they may have limited policymaking abilities in an uncertain world, providing them with incentives to issue vague opinions so that others with more expertise can implement policy outcomes. The vagueness or specificity of rulings, however, has implications for the court as an institution. On the one hand, vague decisions leave room for other actors to avoid compliance with the court’s ruling—a problem especially vexing for an institution that does not have enforcement powers. On the other hand, a vague decision can mask open defiance; so if it is unlikely that the other branches of government will abide by the court’s ruling anyway, the court may use vague language in an attempt to protect its institutional prestige.

Representative democracy is centered around a critical act of delegation: citizens charge their representatives to make decisions either as they would or on their behalf (Manin 1997). As such, citizens must deal with both problems of adverse selection and moral hazard: they must choose among agents who possess superior information about their own competence, integrity, etc.…, and they must control these agents in the absence of full knowledge about their actions. That citizens have one instrument—elections—to solve both problems may make things worse: in using elections to distinguish among candidates by type, citizens may undercut their ability to optimally motivate whoever is elected once in office (Fearon 1999). The manner in which citizens select and control their elected officials is not a trivial concern, especially given the fact that in this instance of delegation agents actually have the authority to make decisions that bind their principals (Moe 1990).

The role of institutions

Given that actors face many different types of strategic dilemmas when they are making collective choices or taking interdependent actions, institutions fit into this picture in a variety of ways. In a somewhat trivial sense, institutional structures sometimes create the strategic dilemma in the first place or serve as a forum in which these dilemmas arise. The problem of cycling emerges, for example, in a setting of group choice (e.g. a legislature) with specific proposal and decision rules (i.e. pairwise alternatives under majority rule). The very nature of representative government creates the agency problem between citizens and elected officials. Similarly, if we did not have elections with a secret ballot, then the commitment problem for voters vis-à-vis vote brokers would not exist.

Analysis of political institutions also has a long tradition in which scholars evaluate how institutional arrangements resolve problems of coordinated action and strategic interdependence. One of the most well-known and insightful examples of this tradition is Milgrom, North and Weingast’s (1990) study of the law merchant. Mercantile law, they argue, evolved to facilitate economic activity in the context of changing world conditions that exacerbated commitment dilemmas that plague economic transactions. Specifically, they show that whereas in a small community repeated interaction among traders creates an incentive for honest behavior, in larger communities resembling modern society, relative anonymity among traders due to the ability to find new trading partners undermines that incentive. Just as traders in large communities face a relatively low risk of encountering the same trading partner again, they also recognize the limited capacity any cheated partner would have to spread word of their misdeed. As a consequence, a trader does not fear the consequences of dishonest behavior. To compensate for the adverse consequences of larger societies on economic activity, the law merchant developed. This institution created an information clearinghouse of which potential traders could make use to learn about each others’ trading records and permit to render judgments in the event of alleged dishonesty. Milgrom, North and Weingast show that it is possible to design an information clearinghouse in such a way that traders would make use of the institution and would comply with its adjudicative decisions.

As our examples from the previous section illustrate, others in this tradition have tackled nearly every imaginable problem of social and political interaction. Among other things, scholars have evaluated how political parties can prevent cycling in complex policy spaces. They have investigated how a system of separation of powers can enable more credible commitments by leaders (North and Weingast 1989), how committees resolve informational (e.g. Krehbiel 1991) and distributive (e.g. Weingast and Marshall 1988; Shepsle and Weingast 1995) problems inherent in legislative settings, and how investigation and litigation can resolve agency loss in the bureaucracy, the judiciary, and federalism. In other settings, scholars have studied how legislative institutions allow governments to manage the interests of their constituent factions. In places with coalition governments, for example, the different parties in the government have potentially conflicting pressures from their coalition members and their electoral constituents. Some types of institutional arrangements enable the government to allow member parties to serve their electoral constituencies without undermining the coalition itself (e.g. Martin and Vanberg 2011).

In other cases, institutions constrain the possible solutions to the problem. Whereas in the previous examples institutions create incentives for people to work together and overcome their collective social dilemmas, in other examples institutions simply foreclose behavioral patterns that are counter-productive or otherwise problematic. For example, in much of the law there are clearly specified institutional rules for what types of disputes courts resolve. For disputes that do not qualify for judicial resolution, those rules foreclose using the judiciary as a means for resolving a conflict. Similarly, the U.S. Constitution specifies a division of powers among the branches of government and a set of authorities specifically reserved to some political entities. Those rules proscribe, for example, the House of Representatives from confirming presidential nominations. Of course, these examples are of rules that could themselves be changed, but the crucial point is that the institution serves to constrain behavior, rather than induce desired choices.

Whether one views institutions as solutions to social dilemmas or mechanisms to facilitate the choice of solution—e.g. by constraining the options available to political actors—the crucial point is that institutions allow actors to manage trade-offs. They structure the options available to actors, provide a framework for evaluating competing interests, and allow for a mechanism to choose among competing solutions to social dilemmas. Of course, this does not mean all solutions to strategic dilemmas are institutional. However, the tradition of institutional analysis provides rich, extensive insight into how institutions shape social behavior.

Institutional origins

These lines of inquiry are primarily about the effects of institutions on social coordination problems. However, the social science model for making inferences about causal effects relies on the construction of counter-factuals. What would have happened if the institution were not adopted? The strongest model for making such an inference relies on the assumption that the institutions are randomly created—some places have independent courts and property rights, whereas others do not, and which ones do and which ones do not is random. Unfortunately, that institutions are randomly created is implausible. As a consequence, we must face an a priori question about the origins of institutional arrangements.

Specifically, we must ask ourselves why an institution exists. Are the institutions we observe in the world created specifically to optimally resolve the dilemma we are studying? In some cases, the answer is probably yes. To the extent that political actors want to avoid the inefficiencies resulting from uncertainty or renegotiation over the basic structures of government, for example, they may purposefully settle on a set of constitutional rules to serve as a focal point (e.g. Ginsburg, Chapter 8). Yet the specific provisions of constitutions are usually a product of political bargains, designed by actors who are keenly aware of their distributive implications and only secondarily (if it all) considered about their implications for collective welfare (Elster 1993). Similarly, the rules a society selects for legislative policy making can aid in resolving problems of collective choice, but also have important consequences for which groups in society tend to fare better in the legislative process. We know that institutional arrangements have important distributive implications (Knight 1992: ch. 2). The implications are twofold. First, even an institution that maximizes collective welfare is not free of distributive consequences. A variety of different possible institutional solutions lie on the Pareto frontier, and each one of them implies different distributional consequences. Second, and perhaps more commonly, distributive concerns may lead to a set of institutions that is not optimally designed for other roles the institutions play.

As an example, consider the U.S. Congressional committee system. There exists a deep debate about whether legislative committees are primarily concerned with resolving distributive or informational limitations in the legislature (e.g. Shepsle and Weingast 1987; Krehbiel 1991). Under the distributive account, giving committees jurisdiction over specific substantive domains enables a form of logrolling by which the farmers’ representatives can control agricultural policy and Silicon Valley’s representatives can control technology and patent law policy. The two groups can sustain a bargain in which they each get what they want on the dimensions about which their constituents care most. Related, the committee system, by breaking policy into discrete dimensions debated and acted on separately, can help resolve the social choice pathologies that exist—such as cycling—when a group collectively votes on multi-dimensional policy domains. Under the informational account, an entire legislature lacks expertise in every single policy domain and so would prefer to delegate crafting agricultural policy to the farmers’ representatives and crafting patent policy to Silicon Valley’s representatives. Because those representatives have strong interests in the policy domains, they will have an incentive, lacking for others, to collect information and develop expertise in each of their respective policy domains.

What is critical about this example is that the inferences we draw about the committee system’s ability to resolve informational problems, for example, may be different if they were intended to resolve distributional battles rather than informational limitations. Drawing an inference about whether the committee system is more about informational challenges or more about distributive concerns cannot be accomplished simply by examining the committee system and seeing whether it is stacked with preference outliers who push for extremist policy. One must begin with a theory of the problems facing a legislature before a committee system existed, what the goals were for legislators, and what choices were available to them. Once we consider the a priori choice among institutional arrangements we can begin to consider what consequences institutions have on incentives and behavior.

Indeed, this observation is the crucial point Diermeyer and Krehbiel (2003) advance. Drawing an inference about the effects of an institutional structure on political behavior requires scholars to make assumptions about actors’ preferences and then evaluate how they behave under alternative institutional arrangements. Understanding why those institutions exist, in turn, requires backing up that logic one step, making an assumption about political actors’ preferences and evaluating which institutions they adopt under varying conditions. This latter analytic goal is clearly more complicated than the former, but it offers the potential for important insight and raises a host of challenging lines of inquiry, to which we turn shortly.

First, however, we highlight a second, related, challenge in studying institutional origins. Specifically, institutions are usually not established by fiat by one actor. So part of the process to understand is how such institutions (that solve the collective problem) actually emerge. For example, a regime party may be useful for sharing spoils and avoiding internecine conflicts among authoritarian elites. But it also requires each of them to give up some part of their fiefdoms. So how does the party emerge? Do they realize that it is collectively beneficial for them in the long run? Or must it be imposed, perhaps on some but not others? Further, if any particular institutional arrangement will have long-run distributive consequences for those subject to the institution—for example by resulting in policies that tend to favor special interests over median voters—then might not the institutional founders anticipate those outcomes and have a harder time agreeing on institutions in the first place?

The upshot is that because groups must agree to institutions in the first place, forward-thinking political actors are likely to build in to their initial negotiations an anticipation of what will follow, perhaps not being able to agree to institutional solutions to short-run problems because of long-run concerns, or vice versa. When studying institutional origins, then, it is important to consider the full range of concerns that might be in play—concerns about social efficiency but also concerns about distributive politics and the political consequences of resolving the social dilemma. The rub may be that sometimes the solution is worse than the problem, at least from some people’s perspectives.

Complications of studying institutional origins

Once one opens the lines of inquiry to the origins of institutions and the extent to which they are endogenous to political preferences, distributive concerns, and contextual factors, a host of challenging opportunities emerge. Among these intellectual puzzles are: (1) unintended consequences; (2) institutional change; and (3) the effect of institutions on preferences themselves. In our view, these topics are among the most important, yet elusive, issues in current research on political institutions.

Conclusion

Scholars traditionally study their particular substantive institution of interest within their respective subfields. This segmentation emerges for understandable reasons related to professional specialization. But if we think about how institutions emerge and function as responses to a variety of strategic dilemmas facing political actors, there are good reasons for scholars to pay attention to institutions across a range of different substantive contexts. For example, scholars interested in commitment problems—whether they emerge within the context of vote-buying, peace agreements, or constraints on executive power—can and should learn more from each other. Similarly, scholars who examine principal–agent relations within a particular context can learn more about related areas and their own subject of study by engaging with work on delegation across different substantive contexts. It is only with this kind of intra-disciplinary work that we can make headway in understanding more complex questions related to institutional origins and change.