Negotiation in the Crisis of Democracy

The current crisis of democracy has many causes. The one I discuss here is the inability of our democratic institutions—and our understanding of those institutions—to keep up with our collective needs.1

I first argue that the primary purpose of government is to create legitimate coercion in order to solve the collective action problems caused by the need to create “free access goods.” I then argue that our needs for free access goods have increased exponentially in the last hundred years. If these two propositions are true, then the task of producing governing structures that can help us create these goods becomes one of the most urgent tasks of our time.

This task involves conceiving and building many different kinds of institutions that can be the source of legitimate coercion. Democracy is our most effective institutional means so far of creating legitimate coercion, but democracies are having trouble in almost every country on the globe. Our needs for legitimate coercion also now extend far beyond the borders of nation-states, which are the loci of democracy as we know it. To make both national democracies and international institutions work better, we need to understand the crucial role of negotiation in generating legitimate coercion. Despite the importance of negotiation both within democracies and outside them, however, political scientists have given scant attention to the character of successful and unsuccessful negotiations and to the institutions that facilitate successful negotiation. We need to know a lot more than we do now about how to make political negotiations work.

This comment will briefly explain the concept of free access goods and the collective action problem, explain the requirement for legitimate coercion, and then explain the importance of negotiation in generating legitimate coercion. Turning to the United States, it will then explain why the current polarization and close contestation between the parties make it almost inevitable that negotiations between the parties will break down. In a system like the U.S., characterized by a strong separation of powers and thus an institutionalized requirement for negotiation, the breakdown of negotiation produces deadlock and the incapacity to solve collective action problems. Failures of negotiation at the global level produce a comparable incapacity to solve collective action problems such as global warming. Faced with these problems, we need to revise current popular and academic concepts of democracy to take into consideration the conditions for successful negotiation.

Free access goods and the collective action problems they create

Until the middle of the twentieth century, human beings did not have access to the logic of free access goods and the collective action problems they create. As is often the case with new discoveries, at this one historical moment, researchers in several different fields independently converged upon the insights that would become codified as the “collective action” (or “free rider”) problem. Around 1950, in the new field of game theory, researchers at RAND discovered the logical structure of the “prisoner’s dilemma.” In the field of public finance, William Baumol and Paul Samuelson began to work out the structure of what was later called a “public good.” In the field of fisheries management, two economists demonstrated the dynamics of depletion of common supply. By 1968, after the seminal conceptualizations of the economist Mancur Olson and the microbiologist Garrett Hardin, the results of this discovery began to spread across the social sciences.

The key to this newly discovered logic is the character of a free access good. A free access good is a good open to all potential users. Once such a good is produced, anyone can consume it—even if they did not contribute to producing it. If some people provide common defense against enemies, anyone can benefit even if they did not contribute; so too with law and order, or a toll-free road. When a good has the characteristic of being “free access,” or open to all potential users, everyone is tempted to free ride and use the good without helping to create it by contributing effort or money. That good thus tends to be underproduced relative to demand and willingness to pay.2

To fix ideas, imagine that in a given community of three to three hundred billion individuals, each person can contribute either zero dollars or one hundred dollars to a common pot, and then a “doubling machine” will double all the money in the pot and distribute it back to everyone in the community equally. This is the “common pool” version of the collective action problem. In such a setup, it is to each individual’s interest to contribute zero dollars, thus keeping one’s own stake and also getting an equal share of what others have contributed. But if everyone does this, the doubling machine will be completely wasted. One standard way of solving this problem is for all affected to agree that they will be fined two hundred dollars (or given another punishment more costly than one hundred dollars) if they do not contribute. Under the threat of this sanction, everyone will contribute, and the society will prosper more than one that could not get as many members to contribute.

Societies through the ages have solved collective action problems like this (such as common defense and law and order) through informal social sanctions in small groups, or, in larger polities of strangers where informal sanctions would not work, through coercion legitimated by hereditary monarchy or religion. Democracy is the mechanism that human beings have devised for creating legitimate coercion to solve collective action problems when, for various reasons, monarchical and religious sanctions no longer have sufficient legitimacy within a community. The legitimacy of democracy derives from the approximation of democratic procedures to a situation in which each individual affected has an equal vote.

As different versions of democracy have evolved in different countries over the years, they have differed in the extent to which they have stressed the importance of solving collective action problems through legitimate coercion over the importance of preventing the tyranny that some parts of a society could exercise over others in the course of applying that coercion. The United States Constitution emerged at a historical moment when concern about potential monarchical tyranny was great and the number of collective action problems needing solution was relatively small. The strong separation of powers that ensued was explicitly designed to prevent tyranny. In practice, that separation of powers requires that in order to get anything done, members of Congress have to negotiate first within the House of Representatives or the Senate, then between houses of Congress, and finally, on some occasions, between Congress and the President. Recognizing this ubiquity, Tom Edsall concluded spontaneously at the end of a workshop on polarization in the U.S. Congress: “Politics is negotiation!”3

Politics as negotiation: A program for the reform of democratic institutions and democratic theory

Not all politics is negotiation, and Edsall would be the first to admit this. Yet even if politics is only in some relatively large part negotiation, it is striking how little attention the discipline of political science has given the subject. Outside political science, scholars of negotiation in business schools, law schools, and policy schools have developed a body of theory and practical recommendations that help individuals become better negotiators. They do not, however, address the question of what larger institutions might facilitate good negotiation. Such institutions are the domain of political science. The discipline of political science has not taken up this challenge in part because most negotiations take place behind closed doors, produce little quantifiable data, and are therefore not under the light of the lamppost. It has also not done so because popular and academic theories of democracy so far make little place for negotiation.

This short comment asks how we would see representation differently if we took negotiation seriously. If we did, it suggests, we would take a different approach to such institutional norms as transparency, contested elections, and opposing “Washingtonitis.” We would aim for procedures that made the job of the legislator as negotiator easier and we would rethink our popular and academic theories of democracy in this light. We would begin, for example, to detail the circumstances in which publicity, or transparency, did more harm than good. We would begin to value procedures and institutions that created ongoing interactions among legislators. We would see more reason for a “selection model” of representation and for “deliberative accountability” rather than accountability based solely on monitoring and sanctions.

The job of the legislator as negotiator is particularly difficult today in the United States because, in a “perfect storm,” two largely separate dynamics have produced parties that are each more polarized and elections that are more closely contested. This polarization and contestation have effectively blocked the negotiation required for successful democracy in a system characterized by a strong separation of powers.

In the first dynamic, several influences have contributed to polarization. Since President Johnson signed the Civil Rights Act in 1964, white Democrats in the South have become largely Republican. This process has made both the Democratic and the Republican parties more ideologically coherent and therefore more easily polarized, even outside the South. The same process has also strengthened the radical wing in the Republican Party and made it more populist. At the same time, institutional reforms, such as the party reforms replacing smoke-filled rooms with primaries and the congressional reforms strengthening party discipline, have contributed to the polarization. Furthermore, since the end of the nineteenth century, increases and decreases in economic inequality have closely tracked increases and decreases in polarization.4

In the second dynamic, contestation has increased as Republican gains among the electorate have led the parties to become closer in electoral strength. Whenever a minority party in a two-party system has a good chance of defeating the majority party in the next election, it is strategically rational for that party to oppose all actions by the majority party in order to deny it any claims to success in the next election. There is considerable asymmetry in this dynamic, as the Democratic Party is more committed to change and the Republican Party more committed to the status quo. There is a parallel asymmetry in the intensity with which partisans in the two parties embrace obstruction. Nevertheless, the incentives apply to both parties.5

The U.S. legislative system is consciously designed to block all measures that cannot survive the many veto points embedded in an extensive separation of powers. It thus requires negotiation. Yet the two current dynamics of party polarization and contestation have produced two parties with little incentive to negotiate. The parties have become “Westminster-like” parties, governed by party discipline. Embed such parties in a separation-of-powers system and the result is deadlock. That deadlock has generated the greatest failure to produce legislation since the late nineteenth century, when the parties in the U.S. were equally polarized.6

The structural changes that have produced more polarized and equally contesting parties are unlikely to disappear. The Southern Realignment has had spillover effects throughout the United States, creating more ideologically coherent and therefore (in a two-party system) more polarized parties. The two parties may also continue to be relatively equally popular among the electorate, producing continuing contestation-based incentives for obstruction. Other reforms are unlikely to produce much change in this picture. Redistricting reform, a commonly suggested solution, is likely to have only a small effect. The Senate, where the districts have not changed over the years, has a historical trajectory of polarization which parallels that of the House. Open primaries, another suggested reform, may have some small effect, and the new, relatively untested “top two” primary system of California, combined with its redistricting reforms, may have a somewhat bigger effect; but so far the empirical effects on reducing polarization of the current open primaries have been small. Significant campaign finance reform, a third suggested solution, is unlikely in the near future, not least because of the current composition of the Supreme Court. We can thus safely predict a discouraging climate for negotiation in the near future.7

Given this unpromising structural situation, what can be done? On the practical side, filibuster reform would help, as would conscious efforts to facilitate interaction among members of opposing parties.8 But in addition to the deep structural constraints produced by a polarized and highly competitive party system overlaid on a strong constitutional separation of powers, we also face a set of popular and academic theories of how democracy should work that stands in the way of good negotiation. On the academic side, we can at least begin to rethink the work of the legislature to put negotiation at its core.

Putting negotiation at the core

Putting negotiation conceptually at the core of the legislative process would require building on the past fifty years of work on negotiation at the individual level. That work has produced many insights, including defining negotiation success and identifying the human errors that most frequently produce failure in negotiation.

Negotiation success takes two forms:

a) When a zone of possible agreement exists on various issues within the negotiation, the parties discover these possibilities and make those agreements.

b) When the various parties to the agreement value different outcomes differently and differential trade-offs can be made in which parties gain on items of relatively high value to them and lose on items of relatively low value to them, the parties discover such differences and exploit them appropriately in the trade-offs.

Many human errors produce a failure to find and exploit zones of possible agreement and potential differential trade-offs. The cognitive errors include fixed-pie bias and self-serving bias:

a) Fixed-pie bias: The preconscious assumption of zero-sum conflict even when more issues can be brought into a negotiation to produce outcomes that are better for all.

b) Self-serving bias: The preconscious cognitive bias that makes all human beings process facts and interpretations differentially, favoring those that benefit them in contrast to those that benefit others. Self-serving bias applies to concepts of justice and the common good as well as to facts. It includes optimism bias, the tendency to see things more favorably to oneself or one’s party than is justified by the reality. (Think of Karl Rove’s cognitive and emotional certainty regarding the poll results in the 2010 election.)

Many laboratory studies have demonstrated the ways in which these pan-human preconscious biases impede successful negotiation.9 Making negotiation more central to the legislative process would mean designing institutions to make these sources of human error less, rather than more, likely.

With such institutions in mind, making negotiation more central would transform the norms with which we approach the representative process in several areas, including transparency, contested elections, and accountability.

Possible reforms

  1. A nuanced approach to transparency

Between 1980 and 2004, the frequency of the word transparency more than tripled in books in English.10 As post-World War II unity faded and citizens became more distrusting of government, the demands for greater transparency skyrocketed. Yet privacy is critical to negotiation. It allows opposing parties to explore possible zones of agreement delicately and gradually, beyond the reach of the public media, where words can be taken out of context, “feelers” misinterpreted as “giveaways,” and parts of an eventual package rejected by constituents before the package itself is complete. Privacy also allows for the recognition in an informal setting that one’s own interpretation of the facts and the justice of a situation are not universal and that different interpretations can be held by persons of goodwill. As early as 1982, senators singled out the new rule opening Senate committee meetings to the public as the “most often” reason for the decline in negotiation and political self-sacrifice.11

Recognizing the importance of negotiation requires making privacy in legislation more legitimate. For privacy in negotiation to be democratically legitimate, the following conditions must hold:

  • The level of corruption in the system must be low.
  • Citizens must agree to negotiation privacy.
  • The relevant interests must be represented fairly in the negotiation.
  • After the negotiation, the negotiators must make the larger rationale for the outcome public.

When these conditions hold and negotiation is important, we should demand transparency in rationale, not in process. Transparency in rationale means making the reasons for decisions public, not opening to the public the process of reaching those decisions.

  1. A nuanced approach to contested elections

It is almost an article of faith in political science that elections should be contested. Critics often consider long incumbencies an unquestioned indicator of lack of democracy. Some quantitative indexes of democracy specifically use lack of turnover in legislative seats to denote low levels of democracy. Yet long incumbencies generate the longstanding, ongoing relationships that are critical to negotiation. The “great negotiators” in Congress have all been members of the institution for years. Human beings build relationships over time, learning whom they can trust and for what, what parts of what issues are most important to others and why, and how others see both facts and conceptions of the common good.

Recognizing the importance of negotiation requires relegitimating long incumbencies. For long incumbencies to be democratically legitimate, the following conditions must hold:

  • The level of corruption in the system must be low.
  • The representative by and large must promote policies and a larger political direction that the majority of constituents approve.
  • The majority (and even the minority) of constituents must be relatively satisfied with their representative.
  • Satisfaction with the representative must not be the result of ignorance or manipulation.
  • The existing media and interest group systems must be healthy, present alternative policies, and able to publicize departures from citizen preferences or interests.
  • The internal party system must be vital, self-policing, and continually infused with new activists and new ideas.
  • The citizens must be active in other forms of politics and therefore able to inform themselves easily and take action to remove the incumbent when he or she no longer represents their interests adequately.

3. A nuanced approach to representation.

The current reigning model of representation and accountability, both among the public and among political scientists, is based on sanctions. Most political science courses today teach that politicians are motivated primarily by the desire to win election and reelection, and that their accountability to the electorate results from the voters being able to exercise the sanction of voting them out of office. The representatives do what the voters want them to do because they fear that if they act otherwise, the voters will not reelect them. This “sanction model” of political representation has dominated political science since 1973. Before that time, the sanction model was typically complemented by what I have called “a selection model”—as in, for example, Warren Miller and Donald Stokes’s famous 1963 “diamond” figure, depicting political representation as having these two paths.

In a “selection model” of representation, voters chose “gyroscopic” representatives with intrinsic motivations whose objectives are most closely aligned with their own. This model conforms better than the sanction model to the way many representatives see their jobs and also to the way many voters see their own representatives (as opposed to the way they see the system of representation in general). Whereas in a “sanction model,” the voters exercise power over the representative through the threat of non-reelection, in a selection model, the voters instead exercise power over the legislature in which they place the representative. That is, they get the legislature, rather than the representative, to do what it would otherwise have not done. The selection model allows voters to choose representatives “like” themselves, thus avoiding the many connotations, including elitism, in the Burkean “trustee.” In a selection model, the voters are not picking someone who they think knows better than they do, although of course they want someone with experience and knowledge; they are picking an individual whose reputation gives a reliable signal that he or she will continue to have the same long-run goals as in the past—goals that are similar to the voter’s own.12

Although no pure form of either the sanction or the selection model exists (instead, all individual legislators have a greater or lesser “selection core” and “sanction periphery” on specific policies), negotiation behind closed doors tends to work best with legislators primarily of the selection-model gyroscope type. For a sanction model to work well, the voters or their proxies in the media must be able to monitor the representative at all times. In a selection model, the voter expects the commitments that the voter selected for originally to continue throughout the representative’s life, even behind closed doors. That established reputation also makes it easier for the representative to explain unexpected outcomes of a negotiation (think Nixon on China).

For a selection model of political representation to be democratically legitimate, the following conditions must hold:

  • The level of corruption in the system must be low.
  • The representative must in fact have inner commitments congruent with those of a majority of the constituents.
  • The voters must be able not only to select but also to deselect the representative if either the representative or the constituency changes, or if another candidate comes along with an equally congruent constellation of commitments plus greater competence. The criteria for democratically legitimate long incumbencies (see above) must therefore be in place.
  1. A more nuanced approach to accountability

Demands for accountability as monitoring and sanctioning have increased in tandem with demands for transparency. At the same time, the traditional deliberative concept of accountability based on “giving an account” has been replaced by a newer threat-based concept, based on extensive monitoring and the threat of punishment. As noted earlier, accountability as monitoring and sanctioning is not easily compatible with negotiation behind closed doors. For traditional deliberative accountability to be democratically legitimate, the same conditions must hold as for the selection model of representation.

Conclusion: The role of negotiation in institutions that create legitimate coercion

Our lives today rest on millions, indeed billions, of deep, broad, and subtle global interdependencies. Those interdependencies in turn rest on free access goods that require coercion—optimally legitimate coercion—to maintain. At one extreme looms the planet-threatening catastrophe of global warming. Many still do not recognize that reduction in global warming is a free access good that we will have to coerce ourselves to produce. At the other extreme, take the trivial case of blueberries in the winter. At every stage, state coercion generates the free access goods necessary to get those berries on our tables. In January, 60 percent of U.S. blueberries come from Chile. Even before they are planted, the Chilean Agricultural Ministry gives farmers information about the crop; the Chilean Plant and Animal Health Policy, one of the strictest in the world, helps keep dangerous organisms out of the agricultural system; thereafter, many other free access goods (such as highways, ports and airports, emissions and pollution regulations, safe seas, law and order, and property rights protected by the courts) are each provided through some sort of state coercion, including the coercion needed to collect taxes. Those free access goods allow us to walk into a supermarket and buy those berries, accurately labeled and safe to eat. The market depends on the coercion that makes those free access goods possible.

Many studies have shown that coercion perceived as legitimate is far more efficient and effective than coercion perceived as illegitimate and therefore avoided. In a good world, coercion perceived as legitimate would actually be legitimate in ways that could be defended to those affected and that, ideally, those affected themselves would agree to.13 Democracies are to date our best way of generating genuinely legitimate coercion. Two sorts of democracies depend heavily on negotiation. Democracies with a strong separation of powers depend on negotiation throughout the system to surmount the many veto points. Democracies with many political parties—the vast majority of democracies across the globe—depend on negotiation continually to maintain their governing (and opposing) coalitions. In structures like the international system, which cannot rely on institutionalized democracy, negotiation provides the only possibility for creating legitimate coercion.

The point of this comment has been to show how crucial free access goods are for producing most of the things we want in life, how crucial legitimate coercion is for producing those free access goods, and how crucial negotiation is for producing legitimate coercion. When we look at the stalemate in legislation in the United States and the inability of nations to agree on reductions in carbon emissions, we are looking at failures in negotiation. We know that human beings constantly make predictable cognitive mistakes that undermine our capacities to negotiate outcomes that will benefit us. We need to design our institutions to correct for predictable human error, not exacerbate it. To do so, we need democratic theories that reinterpret traditional democratic ideals such as transparency, contested elections, representation, and accountability in ways that are compatible with successful negotiation. The next stages in human theorizing about democracy and international institutions must focus on improving our collective capacities to negotiate.


Barber, Michael, and Nolan McCarty. 2013. “Causes and Consequences of Polarization.” In Negotiating Agreement in Politics: Report of the American Political Science Association Presidential Task Force 2012-13: 14–53.

Buchanan, Allen. 2002. “Political Legitimacy and Democracy.” Ethics 112 (4): 689–719.

Foster, Chase, Jane Mansbridge, and Cathie Jo Martin. 2013. “Negotiation Myopia.” In Negotiating Agreement in Politics: Report of the American Political Science Association Presidential Task Force 2012-13: 73–85.

Lee, Frances E. 2009. Beyond Ideology: Politics, Principles and Partisanship in the U.S. Senate. Chicago: University of Chicago Press.

Mann, Thomas E., and Norman J. Ornstein. 2012. It’s Even Worse than it Looks: How the American Constitutional System Collided with the New Politics of Extremism. New York: Basic Books.

Mansbridge, Jane. 2003. “Rethinking Representation.” American Political Science Review 97 (4): 515–527.

-----. 2009. “A ‘Selection Model’ of Political Representation.” The Journal of Political Philosophy 17 (4): 369–398.

-----. 2011. “Clarifying Political Representation.” American Political Science Review 105 (3): 621–630.

-----. 2012. “On the Importance of Getting Things Done (The 2011 American Political Science Association James Madison Lecture)." P.S.: Political Science and Politics 45 (1): 1–8.

-----. 2013. “What is Political Science For?” (American Political Science Association Presidential Address). Perspectives on Politics 12 (1): 8–17.

Mansbridge, Jane, and Cathie Jo Martin. 2013. “Introduction.” In Negotiating Agreement in Politics: Report of the American Political Science Association Presidential Task Force 2012-13: 1–18.

McCarty, Nolan M., Keith T. Poole, and Howard Rosenthal. 2006. Polarized America: The Dance of Ideology and Unequal Riches. Cambridge, MA: MIT Press.

“Special Report: The Individualist Senate.” 1982. Congressional Quarterly Weekly 40 (Sept 4): 175–182.

Tyler, Tom R. 2006. Why People Obey the Law. Princeton: Princeton University Press.

Miller, Warren E., and Donald E. Stokes. 1963.Constituency Influence in Congress.” American Political Science Review 57 (1): 45–56.

  1. The following pages draw from my APSA Madison Lecture, “On the Importance of Getting Things Done” (Mansbridge 2012) and my APSA Presidential Address, “What is Political Science For?” (Mansbridge 2013), as well as from the work of the APSA Presidential Task Force on Negotiating Agreement in Politics: particularly the U.S. working group chaired by Nolan McCarty, the EU working group chaired by Cathie Jo Martin, and the normative working group chaired by Mark Warren. Citations for particular empirical findings can be found in the draft papers from those groups, particularly Michael Barber and Nolan McCarty’s “Causes and Consequences of Polarization.” The opinions advanced here and the weight given to different explanations are my own. Please send comments to
  2. I do not use the economists’ terms “public good” and “nonexcludable good” for technical reasons. For these reasons and for the history of the discovery of the collective action problem, see Mansbridge (2013).
  3. Thomas Edsall, December 2, 2012, as reported in Mansbridge and Martin (2013).
  4. See McCarty, Poole, and Rosenfeld (2006) and the relevant sections in Barber and McCarty (2013) for the relation between inequality and polarization.
  5. See Frances Lee (2009) and the relevant sections in Barber and McCarty (2013) for the dynamic of contestation.
  6. See Mann and Ornstein (2012) for the Westminster analogy and constitutional issues, and Barber and McCarty (2013) for the legislative record.
  7. See McCarty and Barber (2013) for more on these suggested reforms.
  8. See, e.g., the set of reforms advocated on the “No Labels” website (
  9. See studies cited in Foster, Mansbridge, and Martin (2013).
  10. See Google Ngram calculation of “transparency” as a percentage of words in a sample of books in English from 1800 to the present (
  11. "Special Report” (1982: 177).
  12. For gyroscopic representation, see Mansbridge (2003); for the selection model, Mansbridge (2009) and (2011).
  13. See Tyler (2006) for studies of empirical legitimacy; see, e.g., Buchanan (2002) for normative legitimacy.
About the Author
Jane Mansbridge is the Charles F. Adams Professor of Political Leadership and Democratic Values at Harvard University, and chairs the Selection Committee of the Anxieties of Democracy program’s Negotiating Agreement in Congress Research Grants.