Power Games

Aneil Kovvali

According to the traditional account, Congress has the "necessary constitutional means and personal motives to resist encroachments" by the president.[1] As commentators have recognized, however, the traditional account does not match reality. Individuals in Washington, D.C., are more interested in fighting for their political party than for their branch of government,[2] and the essentially reactive legislative branch lacks the capacity to respond to a rapidly changing policy environment.[3] But the traditional account suffers from a more basic flaw. The president can decide whether or not to cooperate with Congress on a situation-by-situation basis. By contrast, Congress's tools for disciplining the president, such as impeachment, typically preclude mutually beneficial cooperation between the branches across a broad set of situations. Since a prolonged collapse in cooperation would be costly to Congress, it will often not be worthwhile for Congress to respond to presidential provocations.

This Essay uses a simple game to show how a player with the ability to make situation-by-situation decisions can outperform a player with less flexibility. It then uses real world examples to map the game onto the reality of interactions between the president and Congress. Finally, the Essay explores the possible use of unusual institutional arrangements to address this power imbalance.

I. Introducing the Game

Consider a simple game played by two rational actors, Peter and Connie. Peter and Connie may each either "cooperate" with the other or "defect." Each must choose a move without knowing what move the other player will choose. If Peter and Connie both cooperate, each will get a payoff of 1. If Peter defects while Connie cooperates, Peter gets 2 while Connie gets 0. If Peter cooperates while Connie defects, Peter gets 0 while Connie gets 2. If Peter and Connie both defect, they each get an infinitesimal payoff, ε. The payoffs are summarized in Table 1 below:

Table 1


Connie Cooperates

Connie Defects

Peter Cooperates

Peter gets 1
Connie gets 1

Peter gets 0
Connie gets 2

Peter Defects

Peter gets 2
Connie gets 0

Peter gets ε
Connie gets ε

        If they play the game only once, both will defect. Connie will reason that if Peter is planning to cooperate, her best move is to defect, since 2 is better than 1. If Peter is planning to defect, Connie's best move is still to defect, since an infinitesimal payoff is better than 0. Peter will reason the same way about Connie. As a result, both Peter and Connie will defect. This is the familiar "Prisoner's Dilemma": Peter and Connie would both do better if they simply cooperated, but the dynamics of the situation make defection the only rational move.

The results become less bleak if Peter and Connie play the game repeatedly. In an iterated game, Peter and Connie have a wider range of strategies, including "Tit for Tat." If Connie plays a Tit for Tat strategy, she will cooperate as long as Peter cooperates, but if Peter defects without provocation one turn, she will punish him by defecting the next turn and the turn after.[4] As a result, Peter will reason that he can get 2 by defecting this turn only at the cost of getting essentially nothing for the next two turns. Therefore he will not defect this turn. If Peter selects a Tit for Tat strategy as well, Connie will reach a similar conclusion. They will both cooperate.

Now let's impose a limitation on Connie. If Connie defects this turn, she must defect in every subsequent turn. In other words, the only way that she can defect is by pulling a "Grim Trigger" that ends her ability to cooperate in this and every subsequent turn. With this change, the punishment that Connie can impose on Peter is more extreme than before. If Connie pulls the Grim Trigger, the most Peter will ever get in any later turn is an infinitesimal amount.

But the change actually degrades Connie's ability to retaliate. Suppose that Peter defects just once. Next turn, is it in Connie's best interest to pull the Grim Trigger? No. If she is locked in to defecting on every upcoming turn, Peter will defect on every upcoming turn because his infinitesimal payoff from the defect-defect scenario is better than his payoff of 0 from a cooperate-defect scenario. This means that Connie will also get only an infinitesimal amount every upcoming turn.

So if Peter defects once but plans to cooperate occasionally in the future, Connie would be better off refusing to pull the Grim Trigger. If she refuses, Peter may occasionally cooperate and let her pick up a payoff of 1. If she does pull the Grim Trigger, Peter will never cooperate and Connie will be limited to earning the infinitesimal amount forever. Recognizing that it is not in Connie's best interests to pull the Grim Trigger in response to every provocation, Peter will engage in occasional defections and will get away with it. Connie will only retaliate if she becomes convinced that Peter is the type of player who will defect continuously.

II. The Relationship Between Congress and the President

Enough games. In certain respects, the relationship between Congress and the president is like the last version of the game played by Connie and Peter. In many areas, Congress and the president obtain mutually desirable outcomes through cooperation. In those areas, the president can sometimes seize additional benefits at Congress's expense by defecting in an individual case. By contrast, Congress lacks the flexibility to respond by defecting in a single case. Its tools for disciplining the president tend to make cooperation impossible across a large number of cases. For example, if Congress attempts to impeach the president, there will be no further cooperation between Congress and the president, and thus no joint benefits, for an extended period. As a result, even if Congress is unified and rational and sees itself as locked in a struggle with the president, it still will not punish the president for every defection.

These points hold true across a broad range of substantive areas. For example, to punish a person under a criminal law, Congress must cooperate by passing the law and the president must cooperate by prosecuting. However, the president can defect by refusing to prosecute particular cases for his personal political gain.[5] Congress can retaliate by impeaching the president or by refusing to enact reform laws that it and the president both want, but both options would entail the end of cooperation between the two players. An impeachment would consume both branches for months or even years. If Congress refuses to enact a reform bill that Congress itself wants, Congress denies itself the benefits of cooperation under its preferred, reformed regime. Given the costs of its retaliatory options, it could easily be irrational for Congress to retaliate.

Control over the military offers another example. Congress cooperates with the president by appropriating funds to maintain a powerful military. The president cooperates with Congress by using force only in circumstances where Congress would approve and defects by using force in circumstances where Congress would be skeptical or openly hostile.[6] Congress can discipline the president by defecting either before or after the president defects. Ex ante, Congress can discipline the president by refusing to fund the military at levels that would permit it to act without additional congressional appropriations. This would force the president to go to Congress to obtain the money required for each desired mission.[7] But in this scenario, Congress would be denied the benefit of a well-prepared military long into the future. In effect, Congress would have locked itself into a "defect" strategy for many years to come. Ex post, after the president has used force despite Congress's disapproval, Congress can defect by attempting to impeach the president. As already noted, however, impeachment is an imperfect mechanism. Congress could also attempt to cut off funds. But this would only work if the presidential action is ongoing (as opposed to a one-off measure like a punitive airstrike) and if Congress is prepared to pay the policy and political price of failing to support the troops. Full support is often the least-bad policy once forces have been committed, and Congress would bear substantial political costs if it declined to support forces deployed abroad. Historically, the potency of these considerations has compelled Congress to go along with various presidential adventures that it did not approve of.[8]

Of course, Congress could attempt to respond to a particular presidential misadventure with targeted legislation. But this approach also faces problems. If Congress were not relatively unified in its opposition, it would be unable to overcome a presidential veto. Even if Congress could overcome a presidential veto, all it would have produced is legislation. Congress would still have to enforce that legislation if it were defied.[9] Congress would face many of the same difficulties in enforcing its specific legislation as it faced in disciplining the president in the absence of legislation. For example, if it is not in Congress's best interest to impeach the president in the absence of specific legislation, it may not be in Congress's best interest to impeach the president even if specific legislation is in place.

That said, specific legislation can change the dynamics of the situation. A president who is happy to defy or manipulate the terms of a general law may be unwilling to violate legislation that speaks specifically to the situation, even if the specific legislation does not directly increase the likelihood of formal sanction.[10] This direct effect is reinforced by an indirect signaling effect. By violating specific legislation, the president might signal to Congress that he is the type of player who will defect constantly rather than just occasionally.[11] A president who is inclined to cooperate would obey, so by disobeying, the president would signal that he is not inclined to cooperate. If Congress concludes that the president will defect constantly, it might rationally discount the possibility of future cooperation and pull the Grim Trigger via impeachment.[12] Specific legislation can also alter incentives by shaping public opinion. If the public is more likely to be disturbed when the president violates clear legislation speaking to a particular international crisis, Congress may get a higher payoff from defecting via impeachment when specific legislation is in place.[13] But presidential nimbleness limits the impact of this mechanism. The president can read the political winds to decide when and how to defect. Consequently, the president will not be disciplined as a result of public opinion unless he badly miscalculates.

III. Other Congressional Tools

Admittedly, Congress does have a few tools other than impeaching the president, passing legislation, and exerting control over appropriations. Congress can also grill executive officials in committee hearings and delay confirmation of the president's nominees to various positions. But these tools have limits. A congressional hearing over a presidential defection would be most effective if the president's chosen policy is unpopular. But the president has presumably chosen the policy because it is in his best interest. Since many of the president's defections will be popular, at least with his political base, congressional hearings would merely offer the president another venue for touting his successes.[14] Hearings on embarrassing presidential slip-ups can cause him real damage, but Congress's capacity to punish the president for accidents should not be mistaken for a capacity to punish him for deliberately defecting when he rightly concludes that it is in his best interest to do so.

Holding up executive appointments can also frustrate the president, but in many ways, it is simply the equivalent of denying an appropriation. Congress denies itself the benefits of cooperation-of a functional agency fulfilling its statutory mission-long into the future.[15] Congress might employ these strategies when it is happy to find itself in a defect-defect equilibrium, as when the current Congress disagrees with the policies enacted by previous Congresses.[16] But it is not a workable approach for encouraging the president to cooperate. Besides, Congress often hurts itself as much as it hurts the president when it holds up a nominee. For example, Congress is not going to let the position of director of the Central Intelligence Agency ("CIA") sit empty for a long period; allowing the nation's intelligence apparatus to spiral into chaos is bad for everyone.

These last few points may seem inconsistent with recent experience. In February 2010, Senator Richard Shelby placed a "blanket hold" on dozens of President Obama's nominees in order to pressure the Obama Administration on a Defense Department contract that could have proven lucrative for his state.[17] In March 2013, Senator Rand Paul filibustered President Obama's nominee for CIA director to draw attention to the Obama Administration's positions on the targeted killing of Americans.[18] But individual legislators made these decisions. The decisions do not appear to have been optimal for Congress as a whole, or even for the legislators' political parties.[19]

However, behavior like Shelby's and Paul's suggests a potentially winning strategy for Congress, and for Connie. Connie's essential problem is that the threat to pull the Grim Trigger isn't credible. Connie can threaten to pull the Grim Trigger if Peter defects, but one turn after Peter's defection, it will not be in Connie's best interest to pull the Grim Trigger and end the possibility of any further cooperation. Because it would not be in Connie's best interest to pull the Grim Trigger, Connie will simply let Peter's defection go. Foreseeing this, Peter will engage in occasional defections. But Connie could make her threats credible by delegating the power to pull the Grim Trigger to a third party whose incentives are different from hers.

Imagine a third player, Cato, who lacks the power to affect Peter's payoffs but derives a large payoff when Peter is punished for Peter's defections. Connie might delegate her power to pull the Grim Trigger to Cato. If Peter were to defect, it would still not be in Connie's best interest to pull the trigger on the next turn, but the decision would no longer be hers. It would be in Cato's best interest to pull the Grim Trigger; Cato would exercise his delegated authority and pull it. Foreseeing this, Peter would not defect. This tool improves incentives for Peter not to defect by heightening the risk of a defect-defect scenario.[20]

The rules of the Senate may reflect a Cato approach. Congress delegates enormous power to individual senators, at least some of whom are likely to derive satisfaction from opposing the president in any given situation. Through such delegations, Congress can ensure that its power will be used to punish the president for defections, even in situations where Congress as a whole would be better served by declining to retaliate. However, there are significant problems with this approach. First, it only works if Congress cannot override the decisions of the individuals to whom it has granted authority. If Congress can override Senator Paul, Senator Paul cannot make Congress's threats of retaliation more credible. Second, rules that empower individuals will end up empowering individuals on both sides of the ideological spectrum. If a liberal senator draws one "red line" and a conservative senator draws another, there is no guarantee that the president will be able to avoid a highly destructive confrontation. Third, rules that empower individuals will end up empowering the president's allies in Congress. Instead of helping to block the president, such individuals could cause Congress to support the president's choice or at least block Congress from taking action against the president.[21] Fourth, the empowerment could spill over into substantive areas, which could cause perverse consequences. If Congress granted a veto to too many individuals and committees, it would be unable to pass legislation that adapts policy to a changing environment. As a result, Congress would likely be forced to make broad delegations of policymaking authority to the executive branch. This, of course, would have the perverse effect of empowering the president at Congress's expense.

For all their excesses, the independent counsel provisions of the Ethics in Government Act may have represented a more effective implementation of the Cato approach. Under the Act, an independent counsel could be appointed to investigate and prosecute high-ranking executive officials for violations of federal law.[22] An independent counsel could play the role of Cato by focusing enormous attention and disproportionate resources on a presidential defection, even when Congress would have an incentive to just move on. Empowering an independent counsel does not require empowering other figures or gumming up the lawmaking process.

But there is one final problem with the Cato approach, and it applies to independent counsels as well. In the real world, a person who derives a benefit from punishing the president for defecting is likely to derive a benefit from hurting the president regardless of whether the president cooperates or defects. When such a person is armed with the power to punish the president in response to defections, she is unlikely to be cautious in declaring that the president has defected and using her delegated power to punish him. The result is a trembling hand that often pulls the Grim Trigger in circumstances where it is not warranted. The Grim Trigger is expensive for both the president and Congress-indeed, it is so damaging to Congress that Congress would not pull it even in the face of a presidential defection if left to its own devices, and so it must resort to Cato figures to make its threats of retaliation credible. As a result, the cost associated with such unwarranted use of the Grim Trigger may be too great for the Cato approach to be a viable strategy.


This discussion has been descriptive, not normative. In the game between Peter and Connie, we have the intuition that the two of them ought to share by cooperating with each other. In the case of the president and Congress, intuitions are far less clear. The ability to make tailored one-off decisions in particular cases may be inherent in executive power, and the president may thus be justified in making that type of call as he sees fit. Conversely, perhaps legislative power is not properly exercised on a case-by-case basis.[23] But while the normative implications are contestable, the descriptive claim appears secure. Even if Congress were unified and sought to fight the president to a draw, it would lack the tools to do so.



 [1]. The Federalist No. 51, at 257 (James Madison) (Oxford University Press ed., 2008).

 [2]. See Curtis A. Bradley & Trevor W. Morrison, Historical Gloss and the Separation of Powers, 126 Harv. L. Rev. 411, 443 (2012) ("[I]ndividual members of Congress tend overwhelmingly to act in accord with the preferences of their party."). See generally Daryl J. Levinson & Richard H. Pildes, Separation of Parties, Not Powers, 119 Harv. L. Rev. 2311 (2006).

 [3]. See Eric A. Posner & Adrian Vermeule, The Executive Unbound: After the Madisonian Republic 31 (2010).

 [4]. This is actually a slight variant on the traditional "Tit for Tat" strategy. See Paul G. Mahoney & Chris William Sanchirico, Norms, Repeated Games, and the Role of Law, 91 Calif. L. Rev. 1281, 1285-86 (2003).

 [5]. The Obama Administration's decision not to enforce certain immigration laws against certain especially sympathetic individuals may be one example of such behavior. Memorandum from Janet Napolitano, Sec'y, U.S. Dep't of Homeland Sec., to David V. Aguilar, Acting Comm'r, U.S. Customs & Border Prot., Alejandro Mayorkas, Dir., U.S. Citizenship & Immigration Servs. & John Morton, Dir., U.S. Immigration & Customs Enforcement (June 15, 2012), available at (outlining the policy of exercising "prosecutorial discretion" to limit enforcement of "immigration laws against certain young people who were brought to this country as children and know only this country as home").

 [6]. It is also possible for the president to defect by refusing to take military action despite Congress's preference that he use force. But there are good reasons to believe that this type of defection will be relatively rare, since the president is more likely to support the use of military force than Congress. "James Madison said as much when he wrote, ‘The constitution supposes, what the History of all Gov[ernments] demonstrates, that the Ex[ecutive] is the branch of power most interested in war, and most prone to it.' " Saikrishna Bangalore Prakash, The Separation and Overlap of War and Military Powers, 87 Tex. L. Rev. 299, 313 (2008) (quoting Letter from James Madison to Thomas Jefferson (Apr. 2, 1798), in The Republic of Letters: The Correspondence Between Thomas Jefferson and James Madison 1776-1826, at 1031-32 (James Morton Smith ed., 1995)) (alterations in original). While history does offer counterexamples, these instances appear disconnected from the modern pattern of presidents that are "more hawkish than Congress." J. Gregory Sidak, To Declare War, 41 Duke L.J. 27, 85 (1991). In any event, Congress's ability to force the president into a war he does not want is probably even more limited than its ability to keep the president out of a war he supports. There are few obvious mechanisms tailored to compelling action, and the cost of attempting impeachment would only go up in the context of a foreign policy crisis. But see Charles Tiefer, Can Congress Make a President Step up a War?, 71 La. L. Rev. 391 (2011) (arguing that mechanisms discussed below, including congressional hearings and specific legislation, can be used to goad the president into being more aggressive in his conduct of a war).

 [7]. The Constitution itself imposes similar discipline by requiring that spending for the armed forces must be reauthorized by Congress every two years. U.S. Const. art. I, § 8 cl. 12.

 [8]. See Terry M. Moe & William G. Howell, The Presidential Power of Unilateral Action, 15 J. Law, Econ. & Org. 132, 162 (1999) (discussing historical examples); Jennifer Steinhauer, House Rebuffs Libya Mission; No Funds Cut, N.Y. Times June 25, 2011, at A1, available at (explaining that Congress refused to pass a bill that would limit financing for a Libya mission the House voted not to authorize).

 [9]. See Posner & Vermeule, supra note 3, at 87 (noting that a president may ignore a statute if he calculates that Congress will not retaliate when a statute is violated). Specific legislation may make more of a difference in situations in which Congress does not have to enforce its legislation. For example, if the presidential action is subject to judicial review, the courts may enforce the legislation on Congress's behalf. The existence of specific prohibitory language in a statute can affect the outcome of such cases. See Youngstown Sheet & Tube Co. v. Sawyer, 343 U.S. 579, 637-38 (1952) (Jackson, J., concurring) ("When the President takes measures incompatible with the expressed or implied will of Congress, his power is at its lowest ebb . . . . Courts can sustain exclusive Presidential control in such a case only by disabling the Congress from acting upon the subject."). However, such situations are relatively rare "[b]ecause disputes about the outer limits of presidential power to keep the nation safe and to manage international affairs seldom ripen into justiciable controversies" that the courts can adjudicate. Richard H. Fallon, Interpreting Presidential Powers, 63 Duke L.J. 347, 349 (2013).

 [10]. Put differently, Holmes's "bad man" model of the law may not be a useful model for the constraints that guide presidential decisionmaking. See Richard H. Pildes, Law and the President, 125 Harv. L. Rev. 1381, 1404 & n.69 (2012) (book review).

 [11]. See id. at 1407-08 (noting that the president has an incentive to follow the law because it is a signal of his "credibility as a well-motivated user of discretionary power," which in turn is crucial to his capacity to govern). There is good reason to believe that Congress and the public are attuned to signals of the president's type, even apart from the practical consequences of the signaling action. President George W. Bush's use of presidential signing statements provoked substantial outcry in Congress, despite the fact that signing statements do little other than reveal the president's attitudes about the legislation and about presidential prerogatives. See Developments in the Law-Presidential Authority, 125 Harv. L. Rev. 2057, 2071, 2086-87 (2012).

 [12]. These effects can be heightened by a process of adverse selection. An increased risk of congressional retaliation will lead a president who is slightly inclined to be uncooperative to feign a cooperative nature. This means that when a president defects, he is signaling that he is strongly inclined to be uncooperative. This stronger signal further increases the risk that Congress will retaliate if the president defects, and this increased risk will lead a president who is strongly inclined to be uncooperative to feign a cooperative nature.

 [13]. See Posner & Vermeule, supra note 3, at 88 (observing that there can be a "political cost" to violating specific legislation).

 [14]. See id. (noting that the president is unlikely to suffer political damage from illegal action if the action is popular).

 [15]. The observations in the text may not extend to judicial nominees. Certain executive agencies are not able to take action, let alone successfully complete their statutory missions, if they have unfilled vacancies in key positions. See, e.g., New Process Steel, L.P. v. Nat'l Labor Relations Bd., 130 S. Ct. 2635, 2640-42 (2010) (the National Labor Relations Board cannot function without a quorum of three members). By contrast, courts will generally continue to hear and resolve cases even when they are severely understaffed. As a result, there may be less cost to Congress in delaying or rejecting judicial nominees. On the other hand, the costs that Congress can impose on the president by opposing judicial nominees may be lower as well. When Congress delays or rejects executive nominees, it denies the president the ability to staff the government and may even make it impossible for the president to satisfy his constitutional obligation to "take care that the Laws be faithfully executed." U.S. Const. art II § 3; see Matthew C. Stephenson, Can the President Appoint Principal Executive Officers Without a Senate Confirmation Vote?, 122 Yale L.J. 940, 955 (2013) (suggesting that appointment of executive officers is more central to the president's core responsibilities than appointment of judges). Executive appointees are also more amenable to control by the president and less constrained in their ability to adopt positions that advance the president's agenda. See Matthew C. Stephenson, Legislative Allocation of Delegated Power: Uncertainty, Risk, and the Choice Between Agencies and Courts, 119 Harv. L. Rev. 1035, 1047-48 (2006). Executive appointees can thus be more valuable to the president than judicial appointees. While it may be less costly to Congress to refuse to confirm a nominee to a judicial post than to an executive post, it is also less of a punishment to the president.

 [16]. This point explains recent stretches of congressional inaction over presidential nominees to the Consumer Financial Protection Bureau, the National Labor Relations Board, and the Bureau of Alcohol, Tobacco, Firearms and Explosives.

 [17]. E.g., Scott Wilson & Shailagh Murray, GOP's Shelby Blocking Nominees for Home-State Pork, Wash. Post, Feb. 6, 2010, at A3, available at

 [18]. E.g., Ed O'Keefe & Aaron Blake, Rand Paul Launches Talking Filibuster Against John Brennan, Wash. Post Politics Blog (Mar. 6, 2013 7:19 PM),

 [19]. Wilson & Murray, supra note 18 (describing Senator Shelby's move as a "rare political gift" to President Obama).

 [20]. The concept of heightening risk to improve incentives has been explored in other contexts. See, e.g., Robert Cooter & Ariel Porat, Anti-Insurance, 31 J. Legal Stud. 203 (2002). Commentators have also made the related point that Congress can improve its bargaining position relative to the president by introducing internal vetogates, as it does when it gives committees the power to veto legislation. Robert Cooter, The Strategic Constitution 338 (1996).

 [21]. See Terry M. Moe & William G. Howell, The Presidential Power of Unilateral Action, 15 J. Law, Econ. & Org. 132, 146 (1999).

 [22]. Although the provisions theoretically allowed the attorney general to decline to seek appointment of an independent counsel, as a practical matter, the attorney general would likely have found it difficult to refuse a request from members of Congress. See Morrison v. Olson, 487 U.S. 654, 701-02 (1988) (Scalia, J., dissenting).

 [23]. See U.S. Const. art. I, § 9, cl. 3 (barring Congress from enacting a bill of attainder).

   //  VIEW PDF
Volume Archive
Monthly Archive
Tagged Posts

& Other Current Events

Crawford v. Washington: A Ten Year Retrospective

No one disputes the significance of Crawford v. Washington, 541 U.S. 36 (2004), which fundamentally transformed Confrontation...

Come Back to the Boat, Justice Breyer!

I want to get Justice Breyer back on the right side of Confrontation Clause issues. In 1999, in Lilly...

Crawford v. Washington: The Next Ten Years

Imagine a world . . . in which the Supreme Court got it right the first time. That is,...

The Crawford Debacle

First a toast-to my colleague Jeff Fisher and his Crawford compatriot, Richard Friedman, on the...

Confrontation and the Re-Privatization of Domestic Violence

When the Supreme Court transformed the right of confrontation in Crawford v. Washington, the prosecution...