Institutional Bargaining for Democratic Theorists (or How We Learned to Stop Worrying and Love Haggling)

Contemporary political science takes bargaining to be the central mechanism of democratic decision making, though political theorists typically doubt that processes that permit the exercise of unequal power and the use of threats can yield legitimate outcomes. In this review, we trace the development of theories of institutional bargaining from the standpoint of pluralism and positive political theory before turning to the treatment of bargaining in the influential work of John Rawls and Jürgen Habermas. Their ambivalence about bargaining gave rise to a new focus on the value of negotiation and compromise but this literature constitutes an unstable midpoint between the justificatory ambitions of deliberative democracy and the desire to provide plausible models of political decision making. Instead of advocating changes in mindset or motivation, we argue that a fair bargaining process requires institutional reform, as well as a justificatory framework centered on the preservation of egalitarian decision making.


INTRODUCTION
In 1953, Robert Dahl and Charles Lindblom identified political bargaining as indispensable to polyarchy and a predominant feature of the American polyarchy in particular. They described bargaining as a form of "reciprocal control" among leaders under conditions of social pluralism, interdependence, and partially resolvable disagreement (Dahl & Lindblom 1953, p. 324). Postwar pluralists such as Dahl, Lindblom, and David Truman, who shaped political science at the dawn of the behavioral revolution, took bargaining to be the central means by which social groups and parties produced policy, particularly in the United States. Indeed, in his 1956 landmark A Preface to Democratic Theory, Dahl wrote that in America, "Decisions are made by endless bargaining; perhaps in no other national system in the world is bargaining so basic a component of the political process" [Dahl 2006[Dahl (1956]. However, the ambition of the pluralists was not merely explanatory but normative-Dahl, most famously, sought to defend polyarchical regimes on the grounds that even small groups could make their claims heard and secure some responsiveness from elected officials. Contemporary political science has retained the pluralists' emphasis on bargaining, and scholars developing formal models of political decision making have pushed the field forward in crucial ways. Most political scientists tacitly assume that some form of bargaining pervades all political decision making. Although political theorists may concur with this assumption as an empirical or descriptive matter, they also tend to believe that this merely demonstrates the pervasiveness of unfreedom and inequality in political life, and they eschew bargaining as a source of justified political outcomes for several reasons. First, as Dahl himself recognized, bargained-for outcomes reflect asymmetries in power among various actors and groups. Disparities in power may themselves be justifiable, of course; parties that receive a greater share of citizens' votes may have more power than those that receive support from a minority of voters. But asymmetries in bargaining power may also result from unjustified forms of inequality, and theorists-including Dahl himself, in later years-have been appropriately concerned about the effects of such inequalities on political outcomes. Second, "bargaining" is treated as an antonym for "arguing" in the conceptual vocabulary of contemporary political theory (Elster 1986). The unreasoned and unreasonable quality of bargaining renders it unavailable as a source for political legitimacy for many theorists, who believe legitimate outcomes require justification in the form of reason giving. To be sure, some democratic theorists today seek to rehabilitate negotiation, rooted in mutual concessions, as a source of justified political decision making. But few defend horse trading as such.
Our aim in this review is twofold. Both positive political theory and contemporary normative political theory partially derive from a response to social choice theory. We begin by tracing the development of positive political theory, highlighting some of the key models and results from this literature across a variety of domains. We then turn to the challenges that political philosophersmost notably, Jürgen Habermas and John Rawls-pose to bargaining models, focusing on deliberation and reason giving. We then examine the state of contemporary democratic theory on negotiation and compromise, which we regard as an unstable waystation between deliberation and bargaining. Finally, we sketch the promise of an account of a fair bargaining process that we hope will be able to incorporate many of the core insights from the formal literature on institutional bargaining, while remaining sensitive to reasonable worries about unequal bargaining power.

BARGAINING IN POSITIVE POLITICAL THEORY
Bargaining is central to research about collective decision making in the social sciences, though research on issues related to institutional bargaining adopts a variety of approaches. In this article, we focus on a particular approach: work in the positive political theory (PPT) of institutions. This research offers some important advantages for our normative analysis of bargaining in a democratic society. First, it focuses explicitly on the ways in which the outcomes of the decision-making process differ from what we might expect in a collective decision-making process of equal political actors. In doing so, it clearly establishes an expected result and seeks to explain how and why a particular decision deviates from it. Second, PPT seeks to identify systematically the contextual and institutional mechanisms that allow the bargaining advantages enjoyed by some members of the group to influence the collective outcome. Third, due to the focus on these mechanisms, PPT provides us with a way to assess the likelihood that the decision-making process is a product of asymmetric bargaining as opposed to deliberative persuasion. Our review primarily focuses on the factors identified in the literature that highlight the implications of asymmetries of power in institutional bargaining.

Basic Majority Rule
The PPT analysis of institutional bargaining began with the analysis of basic collective decision making. The early literature was built on the pioneering social choice research of Arrow (1963). According to the social choice results, in a legislature that uses standard voting rules for selecting among a set of alternatives, there are very few situations under which one alternative defeats all the other alternatives in the feasible set. This means that there is no unconstrained voting equilibrium in legislative decision making. Instead, the legislature faces an unstable cycling problem. Regardless of the voted-upon alternative, there are alternatives in the feasible set that are preferred by at least a majority of the members of the legislature (McKelvey 1976). Many scholars have interpreted this result as a serious problem for democracies because it suggests that there is no stable outcome that can be interpreted as the collective group preference on any specific policy question.
The simple model of the early literature captured a fundamental feature of basic democratic decision making: majority rule with a binary choice among equal voters. The predicted result of such a vote, under reasonable assumptions about the distribution of voter preferences, would be the policy that matched the ideal preference of the median voter. The problem is that there are commonly more than two alternative ways to resolve any policy question. When we change one of the alternatives and redo the vote, the distribution of preferences may change and often produces a different median voter-hence the instability of simple majority rule with equal voters.
In response to this general result, later PPT scholars began to employ other approaches, primarily noncooperative game theory, to analyze how legislatures use institutional rules and procedures to create stability in the decision-making process. By adding institutional detail to their analyses, they hoped to resolve the instability problem. In doing so, they also developed an impressive research agenda about the effects of various legislative institutions on the substance of legislative decisions. From this research we can learn a lot about bargaining in legislatures.
The following discussion takes up the evolution of the PPT analysis of legislative bargaining. Like any mature research field, PPT has produced a rich and robust literature that delves into legal and political institutions in great detail. We focus primarily on a few general trends that emphasize the institutional effects that influence bargaining.

Institutional Effects: Agenda Setting
In any legislative setting, there are certain rules that define the general procedures by which the group makes decisions. Research on agenda setting grew out of the earliest attempts by PPT scholars to resolve the instability problem in the social choice literature. Romer & Rosenthal (1978, 1979 establish the idea that an agenda setter can, under the right circumstances, produce stability in a collective decision-making process. In their simple model, the agenda setter makes a policy proposal that is then compared to the status quo in a simple majority vote. The model is characterized by an important feature of many democratic political institutions: authority over agenda control that is constrained by majority approval. With this model, they show that if the agenda setter has complete information about the preferences of the members of the group, she can propose an alternative that would defeat all the other alternatives including the status quo. That alternative would be a stable equilibrium outcome.
The model simplifies important aspects of the decision-making process. It offers a single decision on a tax-expenditure issue. It does not consider any dynamic implications of the political process, and it imposes the agenda-setting rule exogeneously. However, it does demonstrate that the institutional structure affects the voting behavior of the actors and thus influences the nature of the collective outcome. Two features of Romer & Rosenthal's (1978, 1979 result are especially relevant to an analysis of legislative bargaining. We can see this by comparing the stable equilibrium outcome in the agenda-setting case with what we would expect in a case of equal political actors. First, in a group of equally situated voters with a random proposal process (meaning no agenda setter), we would expect the collective vote between the proposal and the status quo to reflect the preference of the median voter. However, in the agenda-setter case, the collective decision can deviate from the median voter's preference and instead favor the preferences of the agenda setter. The agenda setter has the power to influence the outcome by proposing the alternative to the status quo that is the closest alternative to her ideal preference that can also win a majority of the votes. Here the unconstrained equal voter case serves as the baseline for assessing the implications of asymmetries of power for the collective decision.
Second, in addition to the importance of the distribution of voter preferences and the agendasetting rule, Romer & Rosenthal (1978, 1979 emphasize that the collective outcome is also affected by the location of the status quo in the policy space. This captures the importance of the concept of the reservation wage. The reservation wage is the value that voters receive if they fail to accept the new proposal. Many bargaining models use this reservation wage as a measure of the bargaining power that actors enjoy in a collective decision-making process. On those accounts, the larger the reservation wage, the greater the bargaining power enjoyed by the actors (Rubinstein 1982). In these analyses of legislative bargaining, the status quo serves as the reservation wage for the legislators. In assessing the two alternatives before the legislature, each individual voter will compare the value of the new proposal with the value of the status quo. If her ideal preference is close to the status quo, she will have more bargaining power in the sense that she will only accept new proposals that are closer to her ideal preference. This power enhances her influence over the content of any new alternatives because it increases the likelihood that the agenda setter will need to take account of her preferences in order to gain her vote.
The existence of a reservation wage as the institutional opportunity of resisting proposed alternatives provides one potential approach for distinguishing deliberation and bargaining. The status quo of a majority-rule environment itself typically constitutes a reservation wage. While the existence of a reservation wage does not conclusively determine whether the collective decision is a product of persuasive deliberation or asymmetric bargaining, reservation wages do provide some evidence of the likelihood that bargaining has an important role to play.
From this original analysis by Romer & Rosenthal (1978, 1979, other PPT scholars began to investigate the implications of agenda setting and amendment rules for majoritarian political institutions. An important paper by Baron & Ferejohn (1989) expands the analysis of the role of the agenda setter. They model a unicameral, majority-rule legislature whose task is to decide on a policy that will distribute benefits among various constituencies. Each member of the legislature is motivated to support a policy that is best for his or her own district. The institutional structure of the legislature includes an agenda-setting rule that determines which member of the legislature can make a proposal, an amendment rule that defines the process by which amendments, if allowed, are made, and the simple-majority voting rule. The authors also assume that legislative preferences are common knowledge and that all actions of the legislators are observable.
The crux of the analysis focuses on the implications of different amendment rules. Baron & Ferejohn (1989) compare the effects on the decision-making process of open and closed amendment procedures. Under a closed rule, the legislature reaches equilibrium outcomes that strongly favor the author of the winning proposal. The resulting distribution gives benefits to a majority of the districts but the largest share goes to the proposer's district. Under an open rule, the equilibrium outcomes vary depending on both the size of the legislature and the extent to which the legislators are concerned about the costs of their deliberations. If the legislature is small and the members are significantly concerned with the costs of delay, the outcome can provide benefits for all the districts. If the legislature is large, the likely equilibrium outcomes can provide benefits to only a majority of the districts but the distribution is more equal than that achieved under the closed rule.
In Baron & Ferejohn's (1989) model, the legislators have no control over which amendment rule governs their decision making, but the authors offer some suggestions about which type of amendment rule the legislators would prefer. Under a closed rule, the legislature would reach a collective decision in the first round of deliberations, while under an open rule, the decisionmaking process could extend for several stages. Therefore, if delay in reaching a decision is costly to the individual members, they will unanimously prefer the closed-rule amendment process.
The primary result of the Baron & Ferejohn (1989) analysis reinforces the Romer & Rosenthal (1978, 1979 insight that agenda setters wield considerable power in a majority-rule collective decision-making process. The degree of the power asymmetry created by the agenda-setting rule depends on the extent to which the initial proposal is subject to subsequent amendment. But even if there is no limit to the number of possible amendments, there is still a power advantage to being the initial proposer. From these early studies of agenda setting, PPT research continued to analyze the implications of agenda setting on an array of decision-making procedures. Most of this work relies on a basic feature of the original models: that the agenda-setting rule is imposed exogeneously by the researcher and that it involves what Diermeier & Fong (2011, p. 948) describe as a "universal proposal protocol, in which every policy maker or legislator has a chance to make a policy proposal in every round of proposal making and voting." In the last decade, PPT scholars have produced important papers varying this universal proposal protocol and enhancing our understanding of the institutional effects of agenda setting.
Consider, for example, the analysis of persistent agenda setting offered by Diermeier & Fong (2011). As opposed to positing a rule that gives every legislator a chance to be the agenda setter in the next session of the legislature, their dynamic model of majoritarian decision making gives the authority for future reconsideration of a policy to the same agenda setter who originally proposed the policy. They can investigate the implications of persistent control by the same agenda setter, a common characteristic of many democratic political institutions.
Their intuitive working assumption is that, if agenda setting is a valuable role to have, then a guarantee of continuing control over the policy in the future is a desirable addition to the agenda setter's power. This would seem to follow from the Baron & Ferejohn (1989) analysis of the different implications of open and closed rules. On their account, the possibility of reconsideration through amendments proposed by new agenda setters in the future, a feature created by the open rule, would serve to constrain the strategies of the initial agenda setter. Baron & Ferejohn's model demonstrates that the initial agenda setter, cognizant of the possibilities of future amendments, would make her original offer of a distribution of benefits such that it would minimize the willingness of other legislators to support such amendments in the future. This accounts for the different distributions that we find in closed-and open-rule situations.
However, Diermeier & Fong (2011) find that this intuition is not borne out by their analysis. They find instead that the addition of a guarantee of persistent control over reconsideration of a policy weakens the present power of the agenda setter. The constraint on that power, like that of the open-rule situation in the Baron & Ferejohn (1989) analysis, does follow from the future opportunity of reconsideration, but the mechanism of that constraint is quite different. In the Diermeier & Fong analysis, the effects come from changes in the behavior of the nonpowerful legislators. It turns out that purely self-interested voters choose to look out for each other in assessing policies in the initial stage, in order to avoid a situation in future stages in which the nonpowerful voters can be played off of each other. By sticking together, these weaker legislators collectively force the agenda setter to offer an initial proposal that is more egalitarian than she would have had to offer if she did not have the persistent authority to control the agenda over the policy. Thus, Diermeier & Fong (2011) conclude that persistent agenda setting can produce outcomes similar to an open amendment rule. They also conclude that agenda setters would prefer to commit themselves to no reconsideration of policy outcomes in the future.
The Diermeier & Fong (2011) analysis is a good example of how efforts to make models more realistic can enhance our understanding of the complexity of the task of discerning the institutional effects on legislative bargaining. Sometimes more formal authority may mean less actual bargaining power. Other recent research adds realism of another important form. The rules governing legislative decision making in general and agenda setting in particular are usually chosen by the legislators themselves, and the endogeneity of the institutional rules has been incorporated into a number of recent studies. These studies involve stepping back a stage in the legislative process. The analyses consist of a rule-selection stage and then a policy-determination stage. They allow us to assess the relationship between the outcomes in the different stages and to investigate the implications of the policy stage on the bargaining over institutional rules. Consider two recent analyses of procedural choice.
First, Diermeier et al. (2015) analyze the process by which a majority-rule organization makes decisions about the procedures that will govern its future decision making. Using a standard noncooperative game theoretic approach similar to the Baron & Ferejohn (1989) model, they find that from a feasible set of procedural alternatives the group, under reasonable assumptions about the member's preferences, will choose a persistent agenda-setting rule that gives an asymmetric power advantage to some members of the group.
Diermeier and his colleagues raise the question of why a majority of the organization's members would endorse this asymmetric rule when it clearly has the effect of diminishing the substantive benefits some members of the majority would receive from the policy decisions made under the rule. Their explanation rests on the idea that the organizations' members may be risk averse and thus would prefer the stability of a collective decision-making environment to an alternative process that lacks agenda-setting asymmetries but is characterized by constant policy uncertainty. Risk aversion is, in fact, one possible reason why democratic majorities might accept the distributional effects of institution-creating power asymmetries on their collective decisions.
A second possibility, more directly related to the effects of already-existing power differentials, has been offered by Eguia & Shepsle (2015). They directly address the role that pre-existing power asymmetries may play in the establishment and maintenance of agenda rules. They incorporate the fact of elections into their model, seeking to understand the implications of electoral politics on both the choice of rules and the distribution of benefits from collective policy decisions. Their model consists of a game with an infinite number of periods. Each period involves a rule-selection stage, a policy-determination stage, and then an election. In the next period, the three stages are repeated with the newly elected legislators.
The equilibrium outcome of the game is characterized by the choice of an agenda-setting rule that favors incumbent legislators. The rule creates an asymmetry of proposal power among the legislators, with the greater probability of exercising proposal power going to the legislators with the greatest seniority. In the policy stage, the senior legislators use their proposal power to enact policies that disproportionately favor their own constituents. The authors also demonstrate that in the next election stage, voters predictably vote for incumbents over their challengers. Eguia & Shepsle (2015) make the relationship between the choice of rules and the substantive policy outcomes quite explicit. Powerful incumbents prefer the asymmetric power-enhancing rule because they substantively benefit from the choice, in terms of both distributional benefits to their communities and their own electoral success. They prefer it over an equality-enhancing rule that would give an equal proposal chance to all the legislators, and they prefer it over any other asymmetric agenda-setting rule that might be considered. Here the political nature of the choice of institutional rules and practices becomes clear.
From this research on agenda setting and the implications of institutional rules on collective decision making under majority rule, PPT scholars and like-minded students of American politics have employed these basic ideas to study the full range of democratic politics. Rodriguez & Weingast (2003, p. 1441) put the general PPT research question concisely: To the extent that positive political theory contributes the insight that legislative bargaining is ubiquitous, occurring not only during floor consideration but throughout the period prior to the bill's floor consideration, we need to explore more thoroughly how legislators shape legislation in the enactment process. We need to understand the ways in which coalitions take form and how they operate in combination and competition with one another with the objective of bargaining toward a final version of the proposal to be considered by the entire body.
Conceived of in this way, the explanation of how legislative decisions are made entails following the path by which the legislative body arrives at a final collective outcome. Following this path involves an analysis of the various institutional stages of the legislative process. This puts the focus of the analysis on the importance of the institutional rules created by legislatures to govern their collective decision making. Rodriguez & Weingast (2003, pp. 1435 capture the complexity of the legislative process in the following description: For our purposes, the most important institutional details of Congress concern the complex set of institutions granting individuals or groups special powers. Not only must legislation command majorities, it must also gain the approval of committees in each house, the majority party, and the Rules Committee in the House, and succeed against any attempt at filibuster in the Senate. The majority party also retains numerous controls in each chamber to make sure that legislation serves its interests; for example, in the House, the majority party caucus serves this function. Legislation must also be approved by the President, subject to the veto-override provisions of the Constitution.
We highlight three main directions this research has taken, focusing on pivotal voters, political parties, and interbranch bargaining.

Pivotal Voters
The analysis of the effects of institutional rules and procedures on collective decision making takes the basic logic of the agenda-setting research and applies it more generally to various aspects of the www.annualreviews.org • Institutional Bargaining for Democratic Theorists legislative process. The underlying logic of this broader research is to identify pivotal voters whose support is necessary for the successful enactment of legislation (see, for example, McNollgast 1992McNollgast , 1994Krehbiel 1998). Various studies have investigated the importance of those legislators who, in Rodriguez & Weingast's (2003) terms, are given "special powers" by the rules of Congress.
Consider one important example, congressional committee chairs. Given the complex nature of congressional committee systems, certain members of Congress enjoy agenda control over all the policy proposals that are introduced within their substantive purview. Without their support, proposals fail, regardless of the degree of support the proposal may have in the legislative body as a whole. Thus, one of the primary tasks for the supporters of any proposed legislation in Congress is to secure the support of those committee chairs who oversee the policy areas most relevant to that legislation.
To adequately understand, more generally, the bargaining among legislators in the congressional decision-making process, we have to take special account of the actions of these pivotal voters. In the process of seeking their support, policy proposers commonly are forced to change their initial proposals in ways that will moderate the effects of their proposed legislation. In an array of studies of substantive areas of the law, PPT scholars have demonstrated that a full explanation of major legislative initiatives requires tracing their historical path through the negotiations with the relevant pivotal voters.

Political Parties
The PPT analysis of legislative bargaining extends outside the realm of the direct effects of institutional rules and procedures and considers how individual members of the legislature organize themselves into formal coalitions. The most obvious example is the political party. PPT researchers are involved in ongoing debates on the effects of political parties on legislative decision making. The disagreement centers on the extent to which political parties independently influence the nature of legislative outcomes.
One side takes the view that partisan political affiliations are an important factor in legislative bargaining (Cox & McCubbins 2005). One variant of this approach emphasizes party leaders' use of selective incentives to enforce party coalitions over substantive policy questions. Another variant highlights the ways in which party leaders assume control of agenda-setting procedures (Aldrich & Rohde 2001). The other side contends that political parties exercise no independent influence. In support of this view, Krehbiel (1993) argues that there is no evidence that the actions of party officials change the substantive nature of legislative outcomes from what legislators would produce if they acted on their own. A prominent study by Diermeier & Vlaicu (2011) assesses the relative merits of these two variants, finding some support for both positions. Using a simple bargaining model, they demonstrate that a type of behavior that is usually identified as partisan influence, grabbing control of procedural mechanisms like agenda setting, can be an equilibrium outcome even when it is merely the product of the independent choices of unconstrained legislators. That is, like-minded legislators can agree to a collective strategy even without the leadership influence of party officials. Diermeier & Vlaicu's (2011) analysis highlights some of the ways in which like-minded legislators can influence legislative outcomes even without formal party leadership. The key conditions upon which they focus are concern about opportunity costs and the polarized distribution of the legislators' policy preferences. They show that the traditional median voter outcome will not occur except under conditions of zero opportunity costs and no polarization. However, one interesting result of this analysis is that both the weak and the partisan party theories will hold under different sets of conditions-and both are different from the primary findings that Diermeier & Vlaicu choose to emphasize. It turns out that all three outcomes may arise under different sets of conditions about opportunity costs and preference polarization.
The real task for the researcher is to determine the conditions under which party officials can actually influence individual legislators to support policies that they would not choose to support under unconstrained conditions. The large body of research in support of the partisan affiliation variant highlights the ways in which parties not only offer selective incentives but also serve as coordinating mechanisms for like-minded legislators.

External Actors: Interbranch Bargaining
Although the authority for the creation of law in a democratic society is situated primarily in the legislative branch, there are other venues in which the substantive content of the law is determined. In the United States, the institutional structure of the separation of powers provides the other branches of government with significant opportunities to influence this content. Some of these opportunities are direct, employing institutional mechanisms like veto power and statutory interpretation; others are indirect, when the legislature anticipates potential future actions of the other branches and strategically chooses to alter its preferred policy as a way of minimizing the likelihood of undesired actions.
PPT research addresses many of these interbranch scenarios. Primarily, the analyses employ relatively simple spatial models as well as noncooperative games to assess the possible equilibria outcomes from various configurations of interbranch bargaining. In the area of legislativeexecutive relations, there are two relevant bodies of research. The first includes work focusing on veto bargaining (Cameron 2000). Most of this research analyzes how various distributions of preferences among the President and the pivotal voters in the House and the Senate determine the final content of legislation. One version of these games focuses on the bargaining between the two houses of Congress, with the President's preferences serving merely as the limit for how far Congress can deviate from the status quo. A second version focuses more on the strategic behavior of the President, analyzing the ways in which the President's behavior can both influence the internal bargaining within Congress and set limits on the terms of the final congressional bargain (Cameron & McCarty 2004). The models offer interesting insights into the strengths and weaknesses of the branches given the various preference configurations, as well as the important role of uncertainty in determining bargaining strategies and outcomes.
The second area of legislative-executive relations focuses on the influence of administrative bureaucracies in implementing legislative enactments. This research highlights a fundamental feature of this relationship, which McCubbins et al. (1987McCubbins et al. ( , 1989 famously called the delegation dilemma. This dilemma follows directly from the fact that legislatures commonly delegate policy implementation to bureaucracies. While legislators may want to turn over implementation responsibility to agencies whose members have special expertise in a substantive area, the policy preferences of the experts may differ from those of the legislators. The legislators may reasonably fear that the expert bureaucrats will implement the enacted legislation in line with their own preferences and not those of the enacting legislature.
The research in this area seeks to understand how and under what circumstances the legislature tries to address this dilemma. It has identified some important ways in which legislators and bureaucrats are engaged in both explicit and implicit bargaining. Legislators try to address the dilemma through both ex ante and ex post mechanisms. The ex ante mechanisms relate to the legislature's influence over the general structure of governmental agencies, including such issues as agenda control and budgetary decisions. The ex post mechanisms primarily involve ongoing oversight of agency actions. The success of the legislature at controlling the agencies and solving the delegation dilemma is, unsurprisingly, a function of multiple factors, such as information about policy preferences, the transparency of bureaucratic behavior, and political control of the main branches of government. The dilemma is never definitely resolved but rather is an issue for constant attention.
A newer area of interbranch bargaining research incorporates the judicial branch into the separation-of-powers spatial model. First proposed by Marks [2012Marks [ (1988], this analysis employs a model of judicial-legislative bargaining to analyze the judiciary's influence on the collective decisions of Congress. The underlying logic of his analysis is that the expectation of future judicial interpretation could constrain the internal dynamics of legislative bargaining. Marks' model assumes that legislators are aware of the preferences of the judges who will be interpreting the future statute. The task for the legislators is to find a collective decision that is as close as possible to their own preferred law but can also survive efforts by the court to interpret it differently, effectively changing the substance of the law and thus undermining the goals of the legislature. As in the earlier models of legislative-executive bargaining, the likelihood of success for the legislature is a function of the distribution of preferences among the court and the veto players within the legislature.
The Marks model led the way for important studies (e.g., Gely & Spiller 1990, Eskridge & Ferejohn 1992, Ferejohn & Weingast 1992 of various aspects of the legislative-judicial relationship, focusing primarily on the influence of the expectations of future statutory interpretation on legislative decision making. Other studies put the focus on the judicial side of the relationship, analyzing the potential ways in which courts' interpretations might be constrained by expectations of legislative or executive responses to their own judicial decisions. There is empirical support for the idea that judicial-legislative bargaining affects the collective decision making of both of these branches. In addition to the historical examples from US Supreme Court cases used in the PPT studies above, more general studies of the Supreme Court lend further support. Epstein & Knight (1998) assess the claims of these studies as they apply to both constitutional and statutory review by the Supreme Court, and find evidence in the private papers of some of the Justices that they do take account of the potential reactions of the other branches of government in making their own decisions. In contrast, Segal and his collaborators raise questions about the degree to which the Justices actually act strategically in the face of anticipated reactions of political actors. In the first study, Segal (1997) argues that in matters of statutory interpretation, where it is easiest for Congress to respond to a decision of the Court by merely enacting a new statute, the empirical evidence does not support the view that the Justices alter their decision from what they would decide without the threat of congressional reaction. But in the later study, Segal et al. (2011) analyze the possibility of strategic decision making in matters of constitutional review and find evidence of a strategic calculation on the part of the Justices in response to potential congressional reaction. The logic behind the strategic behavior is not that the Court is concerned that Congress will try to change the substantive nature of their particular decision, but rather that the Court fears that Congress will try to attack the Court more directly, seeking to undermine its basic legitimacy.
Positive political theorists and empirical political scientists recognize that interbranch and intrabranch bargains alike may pose challenges to democratic legitimacy. More powerful political actors, whether institutional or individual, may seek to strip power from weaker ones. The division of labor within political science seems to assign democratic theorists the responsibility for taking up these questions, but so far, they have generally declined the challenge. We suggest the reasons for their reluctance in the next section.

DELIBERATION AGAINST BARGAINING
Social choice theory not only influenced PPT but shaped normative political theory as well. In a year spent at the Center for Advanced Study in the Behavioral Sciences (1955)(1956), Dahl became friendly with Arrow, and he cited Arrow as a major influence on the Preface, including, "for better or worse," his "decision to present parts of the argument in a formal notational system" [Dahl 2006[Dahl (1956, p. xvii; see also Dahl & Levi 2009]. Perhaps even more significantly for the trajectory of political theory in the last decades of the twentieth century, Arrow also influenced John Rawls;Rawls, Sen, and Arrow taught a seminar at Harvard in 1968-1969. Remarkably, as Amadae (2003 has documented, in 1964, Rawls presented a synoptic draft of A Theory of Justice to the organization that later became the Public Choice Society, the "Committee on Non-Market Decision-Making," which Buchanan, Tullock, Downs, Olson, Riker, Ostrom, and Harsanyi all attended. As Amadae (2003, p. 149) suggests, "it is not widely appreciated that Rawls developed his theory of justice in active dialogue with rational choice theorists." It is easy to demonstrate the importance of Arrow's work for Rawls, in particular. In a 1964 paper, "Legal Obligation and the Duty of Fair Play," Rawls takes up the question of how we might be obliged to obey a law that we would find unjust by our own lights, or why we would accept as binding, in the circumstance of contracting, any constitutional procedure that would not accord with our own conception of justice. Rawls (1999, p. 124) says that he here "introduces the concept of justice in accounting for what is, in effect, Arrow's nondictatorship condition." In A Theory of Justice, Rawls describes the "concept of rationality" operative in the original position as the "standard one familiar in social theory," citing Sen's (1970) Collective Choice and Social Welfare and Arrow's (1963) Social Choice and Individual Values in support of the "usual" account that a rational person has a coherent set of preferences among the available options and is able to rank them with respect to their ability to further his ends (adding that he makes the special assumption that rational individuals do not suffer from envy) (Rawls 1971, p. 134).
Yet whereas PPT took the assumption of individual rationality and preference ordering and moved into noncooperative game theory and models of bargaining, Rawls (1971, p. 134) famously excluded bargaining as a source of just decision making: "I assume that each according to his threat advantage is not a conception of justice. It fails to establish an ordering in the required sense, as ordering based on certain relevant aspects of persons and their situation which are independent from their social position, or their capacity to intimidate and coerce." Indeed, the original position is designed specifically for the purpose of discerning the "appropriate status quo" from a moral point of view-to wit, "We cannot take various contingencies as known and individual preferences as given and expect to elucidate the concept of justice (or fairness) by theories of bargaining" (Rawls 1971, pp. 134-35). In the original position, the parties do not know their societal positions or endowments, and as such cannot "tailor principles" to their advantage (no one knows what principles would be in their interests), nor form coalitions that would advantage them (Rawls 1971, pp. 139-40). Hence the veil of ignorance can yield a unanimous agreement, as we can presume that "equally rational and similarly situated" parties will be convinced by the same arguments and order their preferences of conceptions of justice identically (Rawls 1971, p. 139).
At least in the initial stages of the four-stage process by which the society creates laws and institutions, bargaining is excluded on the grounds of justice. Rawls (1971, p. 4) asserts in the opening paragraphs of A Theory of Justice that "in a just society the liberties of equal citizenship are taken as settled; the rights secured by justice are not subject to political bargaining or to the calculus of social interests." But when Habermas (1995) accuses him of relegating democratic decision making to an inferior status beneath liberal rights, Rawls (1995) responds that he did not intend to claim that the shape of legislation would be entirely given by the principles of justice.
Rather, after the society has chosen the two principles of justice in the original position (the first stage) and drawn up constitutional norms in a constitutional convention in light of those principles (the second stage), the third stage is that in which legislation (as circumscribed by the principles and the constitution) emerges. (The fourth stage is judicial review to ensure coherence of ordinary law with the constitution through public reason.) At that third stage, citizens and legislators may vote their "comprehensive views" within those bounds and when questions of justice are not at stake; they need not justify these votes by public reason (Rawls 1993, p. 235). Crucially, though, in a footnote in the reply to Habermas, Rawls (1995, p. 152) acknowledges that at this third stage there is a "formidable complication" that he "can only mention here": namely, that there is an important distinction between legislation dealing with constitutional essentials and basic justice, and legislation dealing with political bargaining between the various interests in civil society which takes place through their representatives. The latter kind of legislation is required to have a framework of fair bargaining both in the legislature and in civil society. The complication is formidable because it is a difficult task to spell out the criteria needed for drawing this distinction and illustrating it by instructive cases.
As we will see, political theorists have not yet met this challenge.
Did Habermas attempt to do so? Habermas (1995, p. 131) distinguishes his view from Rawls' by arguing that his own is more modest; it "entrusts more to the process of rational opinion and will formation" (emphasis original). Again, Rawls disputes this by appeal to his third stage. Habermas' own views over bargaining and strategic interaction more generally have evolved over his career, though nowhere does he articulate a full-throated defense. Mansbridge (2012) traces Habermas' views since The Structural Transformation of the Public Sphere, in which he lamented the move from a politics rooted in deliberative consensus regarding the general interest to compromises "haggled out" and bearing "the marks of their origins in the sphere of the market." Specifically, Habermas (1989, p. 198) wrote, "Political decisions were made within the new forms of 'bargaining' that evolved along the older forms of the exercise of power: hierarchy and democracy." Although Habermas (1982, p. 237) claimed that he aimed to treat strategic and communicative action as "two equally fundamental elements of social interaction," in fact he tended to treat strategic action as "unsavory," as Johnson (1991, p. 189) persuasively argues. Communicative action entails the pursuit of understanding, using speech acts to make "offers" which are negotiated through discourse or argumentation with the aim of coordinating actions on the basis of consent. Strategic action, on the other hand, entails a type of force against others, pursued by atomistic, egoistic members who seek to promote their aims rather than to promote understanding.
As Mansbridge (2012) notes, in Between Facts andNorms, Habermas (1996, p. 167) further softens his objections to bargaining, accepting that discursive resolution may not be directly available to certain questions facing complex societies in which "only particular-and no generalizableinterests are available." Drawing in part on Elster's (1986) distinction between arguing and bargaining, in which the latter is marked by "threats and promises" aimed at "forcing or inducing opponents" to accept one's claims, Habermas (1996, pp. 166-67) suggests that bargaining still depends on the discourse principle to provide the conditions of its legitimacy, because the procedural conditions under which bargains (or "compromises") may be presumed to be fair themselves require justification. Habermas' (1996, p. 107) discourse principle-requiring for norms' validity that "all possibly affected persons could agree as participants in rational discourse"-specifically invokes "procedures that provide all the interested parties with an equal opportunity for pressure, that is an equal opportunity to influence one another during the actual bargaining, so that all the affected interests can come into play and have equal chances of prevailing. To the extent that these conditions are met, there are grounds for presuming that negotiated agreements are fair" (Habermas 1996, pp. 166-67). In other words, as Mansbridge (2012) emphasizes, the principle of equal power as such cannot legitimate either the process or outcome of political bargaining; rather, the principles underlying a fair bargaining process depend for their legitimacy on the possibility that all affected parties could agree to them in rational discourse. In Habermas' (1996, p. 166) words, "non-neutralizable bargaining power should at least be disciplined by its equal distribution among the parties." In "Three Normative Models of Democracy," Habermas (1994, p. 4) in fact suggests that discourse theory differs from both liberal and republican models of democracy in its effort to legitimate democratic will formation both through the "communicative" presupposition that allows for the emergence of "better arguments" and "from the procedures that secure fair bargaining." The liberal model aggregates plural values and interests, its normative grounding provided solely by basic rights securing equal weight in the political process; discourse serves only as a means to enable the advocacy of competing interests. In contrast, the republican model depends on citizens' orientation toward the common good, ever constituting itself anew as a political community, and relies on a too-thick, "ethical construction" of political discourse. Instead, Habermas (1996, p. 6) argues, under the discourse model, "deliberative politics should be conceived as a syndrome that depends on a network of fairly regulated bargaining processes and of various forms of argumentation, including pragmatic, ethical, and moral discourses, each of which relies on different communicative presuppositions and politics."

COMPROMISE AND NEGOTIATION
As we have seen, Habermas tends to treat "bargaining" (Verhandlungen) and "compromise formation" (Kompromißbildung) as interchangeable terms, uniformly reflecting the circumstances of conflicting interests, the unavailability of rationally motivated consensus, and the exercise of "non-neutralizable" but equally distributed power. The literature in contemporary political theory has usually focused on compromise, rather than bargaining, though theorists differ in how they characterize the relationship between the two practices. For instance, Gutmann & Thompson (2014) follow Habermas in treating the two as basically fungible. Compromises often result from "unprincipled bargaining and reinforcements of the prevailing balance of power," though bargaining can both promote and undermine mutual respect (Gutmann & Thompson 2014, p. 31). In their words, "Mutual respect is consistent with a wide range of nonviolent means of reaching agreement, which includes bargaining, provided it is in good faith. But mutual respect excludes means-including some other kinds of bargaining-that are intended to degrade, humiliate, or otherwise demean opponents who themselves demonstrate a willingness to negotiate in good faith" (Gutmann & Thompson 2014, pp. 34-35). Such disrespectful bargaining-again, synonymous with "compromising" or "negotiating"-takes the form of entering in bad faith, misrepresenting others' positions, and refusing to cooperate on matters on which agreement is readily available. Their argument, in part, is that a bad bargaining process, one that fails to demonstrate mutual respect, yields a suspect compromise.
As we will discuss in a moment, we are broadly sympathetic to this view. Yet Gutmann & Thompson's focus on actors who operate in bad faith or seek to humiliate, to our mind, is misplaced, because fair bargaining relies on altering the motivations of agents, rather than on their institutional incentives. To be sure, the authors identify the pressures of the permanent campaign as a primary cause of the uncompromising mindset; the uncompromising mindset enables candidates to differentiate themselves from rivals and to affirm their commitment to key principles necessary to mobilize the electorate. Yet they insist that institutional reform, while necessary, is bound to be insufficient; citing the Beatles, they suggest: "'You tell me it's the institution. Well, you know. You'd better free your mind instead'" (Gutmann & Thompson 2014, p. 203).
The negotiation literature similarly tends to rely on the disposition of actors. Much of this work rests on a distinction between "integrative" and "distributive" agreements, coined by Walton & McKersie (1965). Integrative processes aim to realize joint gains or create some surplus value, whereas the latter merely distribute the existing value in such a fashion that one party benefits at the expense of another. In a chapter titled "Deliberative Negotiation," Warren et al. (2016) draw on this distinction; they place deliberative negotiation between "pure deliberation" and "pure bargaining," and then distinguish between its integrative and distributive forms. In the distributive form of deliberative negotiation-one step away from bargaining, on their schema-parties look for a fair compromise. The deliberative setting is marked by parties' transparency and mutual commitment to fair dealings and outcomes, their understanding of which they freely share as a means of producing a legitimate outcome, in which each party sacrifices something of value. Warren et al. (2016, p. 160) specifically connect this view to Gutmann & Thompson's (2014, pp. 16-24) insistence that parties should adopt a "compromising mindset" to mitigate the effects of polarization. Politicians ought to adopt "principled prudence" (the recognition that an outcome, even if far from ideal, is superior to the status quo) and an attitude of mutual respect, rather than cynicism and distrust (Warren et al. 2016, p. 160).
Some scholars have responded skeptically to these arguments, holding (as we do) that the creation of fair political bargains cannot depend on actors' motives but must be produced through institutions. For instance, in a critique of Warren et al.'s (2016, p. 166) view that "not revealing one's reservation price. . .border[s] on the unethical," Beerbohm (2018, p. 18) argues that bluffing and threats may be essential to the practice of compromise, and rightly suggests that "in any legislature, information will be unevenly distributed." He recognizes that virtually any legislative setting will be competitive, and each side may seek a "better deal," marked by "some inequalities in bargaining power." Yet in seeking to justify outcomes generated by threats and inequality, Beerbohm (2018, p. 19) identifies the goal of ensuring that legislation constitutes a "jointly performed" activity, holding that we can realize democracy's value through "equal sharing in authority over our political institutions," despite unequal power (Beerbohm 2018, p. 30). Although this account seems, attractively, to emphasize institutions over motivations, one might worry that a joint-intention account, like Beerbohm's, simply pushes the motivational question back a stage. How can we ensure that any bargained-for outcome realizes a jointly intended plan for legislation, since joint intentions constitute an especially thick account of shared motivations?
In sum, we are skeptical that changes in mindset would be sufficient to overcome the institutional and personal incentives to hold out when such strategies may be instrumentally beneficial. Even if in certain cases such changes could suffice, motivations are inescapably opaque: We can never know for certain what motivates actions, and people have strategic incentives to misrepresent their motives, even to themselves. A focus on motivational changes may make actors who are properly motivated prey to those who would seek to misrepresent their motivations, and reliably discerning motivational types-particularly when institutions are the parties-is likely impossible. The focus cannot be on the unconditioned incentives of actors to change their attitudes or find themselves motivated through principles; rather, their incentives to promote fair outcomesthrough the binding force of institutional rules-must change (Schwartzberg 2018). If we accept that political actors always have mixed motives that are inscrutable to others, then even were we to impose the civilizing hypocrisy of reason giving (Elster 1986), how could we ever hope to realize fair bargains? Whatever strategy democratic theorists adopt for the justification of bargains must rest on institutional design rather than motivation.
Appeals to bargaining or other "nondeliberative mechanisms" respond to the challenge that legislatures and other political institutions fall far short of meeting the conditions of ideal deliberation. But the challenge of justifying these mechanisms has proven quite difficult. Theorists of negotiation and compromise tend to find themselves between the Scylla of descriptive accuracy, providing empirically sensitive models of bargaining power and the use of threats and promises, and the Charybdis of normative justifiability. Indeed, in other work, Mansbridge et al. (2010) recognize that although ideally, democratic decision making would take the form of deliberative negotiations-rather than what Raiffa (1982) terms "cooperative antagonist negotiations," characterized by the presence of threats (and of which bargaining is a subcategory)-in reality, contemporary legislatures are marked by the latter. Although equal bargaining power would be a "regulative ideal," in practice it is probably unachievable because of the innumerable factors affecting parties' reservation wages. Mansbridge et al. (2010) argue, invoking James Madison, that insofar as the use of power is inevitable, democratic institutions ought to incorporate it rather than ignore it; they also suggest that deliberative procedures might justify the use of cooperative antagonist negotiations. But they set aside for future work the scope and limits of these nondeliberative mechanisms.
In a footnote to a section called "The Ideals of Just Negotiation," Warren et al. (2016, p. 147) concede: "This section must be taken as a placeholder for a more thorough discussion that we hope to have in the future and that if we as a collective do not have, we hope that the larger community of normative theorists will." We too can only offer a promissory note, but we hope to distinguish more sharply our ambition from that of theorists who place their hopes in deliberative or integrative solutions.

FAIR POLITICAL BARGAINS
The challenge, as we see it, is as follows. Legislatures and other political institutions are invariably marked by bargaining and by the use of resources through threats and inducements to secure advantage and outcomes. However, insofar as political outcomes derive from unequal bargaining power, it is difficult to see how they could be justified. Why am I obliged to accept an outcome that derives from sublimated force rather than reason? We believe the exercise of unequal bargaining power may be justifiable, and indeed even beneficial, subject to certain constraints. Within legislatures, bargaining power, at least optimally, derives from the distribution of support across the electorate, and in a democratic system it is prima facie justifiable. Moreover, the ability of political parties to attain their ends depends on leadership, i.e., the unequal distribution of power among individual legislators. If we believe that legislatures and other political institutions depend for their functioning on certain inequalities, it is crucial to identify the bounds of these inequalities. Our aim is to identify these parameters-to provide an account of a fair bargaining process that could underlie both legislative decision making and politics more generally.
A fair bargaining process in a democracy needs to serve democratic aims, and we as democratic theorists (predictably) take egalitarian considerations to be paramount (Mansbridge et al. 2010, p. 90). However, insofar as democratic institutions are representative or rely on other forms of expertise (administrative or judicial), they immediately depart from the commitment to equality. One strategy may be to justify such institutions-and bargaining within and among them-insofar as they do not undermine citizens' claims to equal treatment.
Take, for instance, the exercise of party leadership within a legislature; as just mentioned, this is a key domain in which both inter-and intraparty bargaining occurs. Optimally, parties derive their power from the share of support they receive from the electorate. They may accrue resources over time, which party leaders will seek to distribute to their membership. These resources include campaign funds, committee seats, appointees in different branches, and so forth. Such leadership is crucial; it constitutes the means by which parties coordinate to realize aims for which their members were elected. However, the exercise of party discipline and the distribution of these resources are sensitive to normative considerations. A fair bargaining process within a legislature has to identify the conditions under which these unequal resources might be justifiably deployed. For instance, one key consideration governing the use of threats and inducements, either across the aisle or among copartisans, is whether they immediately affect the interests of citizens, redounding to their benefit or harm-particularly, if not exclusively, from the standpoint of political equality. One might argue that the use of resources to reward individual legislators for loyalty or to punish them for defection could be permissible, but perhaps only within certain constraints given by the equal standing of citizens and the functioning of democratic procedures. If leaders use the tools of party discipline to ensure cohesion, threatening individual legislators' private ambitions rather than harming constituents by stripping them of resources to which they would otherwise have a claim, the use of threats may not raise serious concerns. Yet if the unequal campaign resources of parties help to secure incumbents against challenge, undermining the accountability of representatives, they may well oppose the egalitarian interests of citizens. Suitably constrained, parties may use their resources as leverage to promote candidates aligned with their aims and to threaten those who defect, but they may not seek to reduce the competitiveness of elections through gerrymandering, for instance.
Furthermore, our concerns about fair bargaining must extend beyond the confines of legislative decision making itself; we cannot merely evaluate the legislative procedures under which laws are created, nor look to the substantive justice of the laws as written. This point often seems to be lost in debates among democratic theorists. Rather, the justifiability of democratically enacted laws depends on their implementation, interpretation, and enforcement. This highlights the importance of interbranch interactions in the process of democratic governance. As the research shows, institutional bargaining plays an important role in how the legislature interacts with the executive and judicial branches of government. Legislative-executive bargaining is a common feature of the implementation and enforcement of laws, sometimes as executives use threats of vetoes to influence the content of laws, and other times as administrators seek to imprint their own preferences onto the laws in the implementation process. Similarly, legislative-judicial bargaining constitutes a central feature of the process of interpretation and enforcement, as courts use their authority to shape how laws are employed in everyday life. Each of these forms of institutional bargaining can alter the initial content of the law and thus call into question its normative legitimacy. An account of fair bargaining must demonstrate that these influences can be reconciled with fundamental democratic commitments.

CONCLUSION
Many theorists responded to the explanatory ambitions of pluralism skeptically, accusing it of a complacent, conventionalist approach to the study of politics that failed to challenge the distribution of power. Even Dahl-who did more to unite the study of political theory with empirical political science than virtually any other scholar-was vulnerable to this criticism. In a foreword to the expanded edition of the Preface, Dahl [2006Dahl [ (1956, p. xx] acknowledges that such critics may have had a point, as the concluding paragraphs of the Preface "sound more complacent" to his ears now than they must have in 1956. In those paragraphs, he had argued that the American systemmarked by robust contestation among interest groups-constituted a "relatively efficient system for reinforcing agreement, encouraging moderation, and maintaining social peace in a restless and immoderate people operating a gigantic, powerful, diversified, and incredibly complex society" [Dahl 2006[Dahl (1956, p. 151]. He acknowledges, as well, that he did not address in the Preface (though he did in later works) inequalities deriving from race, education, information, and other socioeconomic institutions.