Sage Journals: Discover world-class research

Abstract

Congress’s power of the purse is effective enough to block the implementation of a policy Congress disagrees with, especially in the case of foreign policies initiated by presidents. However, it is puzzling that congressional deference to presidents, instead of defiance, has been common. Conventionally, presidents’ informational advantages over Congress have been presented as the main account for congressional deference. This account connotes that congressional deference is Congress’s surrender to presidents because Congress wants a successful outcome and presidents’ policies are more likely to succeed. However, I present a model demonstrating that congressional deference occurs even if there is no such asymmetric information. The result is that the deference can be Congress’s gambling on presidents’ failure. Congress may defer to presidents not because Congress wants the success of presidents’ policies but because Congress wants to show presidents’ failure to convince voters that Congress’s policy is better than those of presidents.

Keywords

separation of powers congressional deference presidency gambling on failure bargaining

In many policy areas, Congressional deference is essential for presidents to implement their policies. For example, Madison claimed that Congress’s power of the purse is “the most complete and effectual weapon” (Federalist Paper No. 58). The basic intuition is clear: Money is indispensable to the implementation of most policies. Congress can, therefore, block policy implementation by cutting off funds. Particularly, the power of the purse might be the last resort for Congress to check presidents in policy areas such as foreign policy, where presidents have de facto power to initiate a policy unilaterally (Wildavsky 1966).

It is puzzling that congressional deference to presidents’ budget requests seems common despite Congress’s ideological disagreement with presidents’ policies. In 2007, the Democrat-led Congress agreed to provide nearly $100 billion of supplemental funding for the Iraq war despite its anti-war position they successfully highlighted in the 2006 elections. In 2015, Republican-led Congress threatened to use the power of the purse to thwart Obama’s opening of diplomatic relationships with Cuba, but the threat has never been realized.

Presidents’ informational advantage over Congress has been recognized as what leads to this puzzling congressional deference (Canes-Wrone et al., 2008; Dahl, 1950; Howell et al., 2013; Kriner, 2010; Schlesinger, 2004). Presidents have access to superior intelligence resources that are not available to Congress (Gailmard and Patty 2012), which leads Congress to defer to presidents because Congress believes that presidents’ policies are more likely to succeed even if Congress ideologically disagrees with presidents’ policies. Thus, congressional deference connotes Congress’s statesmanship in the sense that Congress prioritizes a successful outcome for a nation over its ideological policy preferences. For example, Steny Hoyer (D-MD) explained that Democrats passed the supplemental funding bill for the Iraq war in 2007 because “none of us want to see Iraq as a failure.” Nancy Pelosi (D-CA) also announced that they would not use the power of the purse to end the war “as long as our troops are in harm’s way” (Curry 2008).

I present an alternative rationale behind congressional deference: Congress is gambling on the failure of presidents’ policies. Congress may defer to presidents to convince voters that Congress’s policy is better than those of presidents by showing the failure of presidents’ policies. To do so, Congress must agree to implement the presidents’ policies in the first place. Furthermore, Congress needs to provide sufficiently large funds to prevent presidents from blaming insufficient funds for the policy failure. Going back to our Iraq war example, “Democrats will still need to take care not to give Bush and his team any easy pretext for shifting responsibility for the outcome in Iraq …[by] pushing for funding cut-offs that could be cast as undermining the troops” (Curry 2008). This may facilitate congressional deference to presidents even if Congress disagrees with presidents’ policies. This rationale also may be helpful to understand congressional swing between deference and defiance, which is observed occasionally (Lindsay 2003) but left blank by the account based on presidents’ informational advantages. Congress defies presidents when voters already consider Congress’s policy better than those of presidents because Congress does not need to convince voters by showing the failure of presidents’ policies.

My model also provides an interesting welfare implication. Assuming that implementing a policy is better for the voter’s welfare than non-implementation, whether Congress and presidents agree or disagree to gamble (i.e., Congress bets on failure, whereas presidents bet on success) affects the voter’s welfare, which, in turn, depends on how much presidents and Congress care about political benefits. Specifically, presidents who receive moderate political benefits from successful policy outcomes are willing to agree to gamble, whereas, paradoxically, Congress is more willing to fund presidents’ policies as it receives higher political benefits from the failure of presidents’ policies.

To demonstrate the above intuitions clearly, I develop a stylized model in which an incumbent president and Congress play a single-period bargaining game before a voter. There are two novel features in this model. First, Congress will defer to the president even if there is no asymmetric information between the president and Congress. Second, congressional deference is observed even in a very strong partisan environment in which Congress and the president do not share a common goal to gain a successful policy outcome.1 These features clearly demonstrate that Congress’s gambling on failure leads to congressional deference even in the absence of the president’s informational advantage and Congress’s statesmanship.

Although my model focuses on congressional funding decisions, the intuition of my model is generally applicable to various forms of congressional deference or endorsement that are required for presidents to gain successful policy outcomes in a broader context of the inter-branch relations. Particularly, this model would be a useful framework to identify an important but understudied component of partisan politics in a wide range of policy areas in which presidents and Congress have conflicting ideological preferences: Congress’s attempts to make presidents look bad (see Lee 2009, 2013), and gambling on failure of presidents might be one way to do so.

Finally, I do not claim that Congress’s statesmanship and presidents’ informational advantage never lead Congress to defer to presidents. A more proper understanding of my model is that this gambling on failure aspect of policy implementation can make it easier for Congress to defer to presidents. The separation of the sword and the purse might be, whether intended or not by the Framers, a more robust institutional design against the worst scenario than we understand: Even when politicians are so ideologically driven and lack statesmanship that they do not prioritize national interests, the separation of the sword and the purse prevent political gridlock and leads Congress to cooperate with presidents when doing something is better than doing nothing to further national interests.

Model

An incumbent president (P) and a single-member Congress (C) play the game before a representative voter (V).2 Both P and C have their own ideal policy. Since I am interested in situations of divided government, I assume that P’s ideological orientation of policy, denoted by p, is different from that of C, denoted by c (e.g., dovish v. hawkish). For the convenience of exposition, I also use p and c to denote ideal policies of P and C based on their ideological orientation, respectively (e.g., waging war v. diplomatic negotiation). These ideal policies of P and C are common knowledge among all players.

An outcome of policy x ∈ {p, c} must be either success or failure if implemented. Success (resp., failure) means policy x succeeds (resp., fails) to solve a problem. Two factors jointly affect the probability of the success of policy x. First, the type of policy matters. A policy x must be either a good type (G) or a bad type (B). A good-type policy is more appropriate to solve a problem than a bad-type policy in a given situation. Hence, a good-type policy is probabilistically more likely to succeed than a bad-type policy. Isolating the effect of a funding level, the probability that a good-type policy produces a successful outcome is q_G, while the probability for a bad-type policy is q_B, where 0 < q_B < q_G < 1, and both probabilities are known to all players.3 However, all players do not know which policy is a good type. Formally, let θ_x ∈ (0, 1) denote the probability that policy x is a good type. This probability is also known to all players.

Second, independent of policy types, insufficient funding would cause policy failure. Let $b (m_{x}) : ℝ_{+} \to [0,1]$ be a weakly increasing continuous function that maps the level of funding provided to policy x, denoted by $m_{x} \in ℝ_{+}$ , to the probability that funding level does not cause the failure of policy x, where there exist ${\underline{m}}_{x}$ and ${\bar{m}}_{x}$ such that ${\underline{m}}_{x} < {\bar{m}}_{x}$ , b (m_x) = 0 for all $m_{x} \leq {\underline{m}}_{x}$ , b (m_x) = 1 for all $m_{x} \geq {\bar{m}}_{x}$ , and b^′ ≥ 0 for all $m_{x} \in ({\underline{m}}_{x}, {\bar{m}}_{x})$ . In other words, any $m_{x} \geq {\bar{m}}_{x}$ is large enough to prevent the failure of x caused by insufficient funds. Any $m_{x} \leq {\underline{m}}_{x}$ is so small that policy x always fails. For any $m_{x} \in ({\underline{m}}_{x}, {\bar{m}}_{x})$ , the probability increases as m_x increases. Consequently, the ex-ante probability that policy x produces a successful outcome is (θ_xq_G + (1 − θ_x)q_B)b (m_x).

Both P and C care about policy and political benefits. First, the policy preferences of P and C are purely ideological. Both P and C gain policy benefits only if their ideal policies are successful because policy failure is equivalent to non-implementation in the sense that their ideal policies fail to yield the meaningful effects they value. In addition, P and C’s ideological preferences of policy are so strong that any positive benefits from the success of a policy with which they ideologically disagree are outweighed by their ideological discomfort (for a similar approach, see Kartik and McAfee, 2007). Both P and C receive policy benefits of 1 if their own ideal policy is successfully implemented, 0 if no policy is implemented or an implemented policy fails, and −1 if the other payer’s ideal policy is successfully implemented. Second, P and C receive political benefits of

π_{P}, π_{C} \in ℝ_{+ +}

, respectively, if they are approved by V (e.g., Groseclose and McCarty 2001). Naturally, π_i can be interpreted as the extent to which politician i cares about political benefits. For now, I assume π_P = π_C = π to focus on the main intuition, but this assumption will be relaxed later. Table 1 summarizes the payoffs for P and C.

Table 1.

The payoffs for P and C.

The outcome of policy p	President’s payoff		Congress’s payoff
	V approves P	V approves C	V approves P	V approves C
success	1+πP	1	-1	-1+πC
failure/non-implementation	πP	0	0	πC

Meanwhile, V is a pragmatist. That is, V only cares about whether a policy is more appropriate to solve a problem regardless of its ideological orientation. In this sense, V’s ideal policy is defined as a policy that is more likely to be a good type, and this ideal policy can change after V’s updating. Naturally, V wants to approve a politician whose ideal policy is the same as V’s ideal policy. Based on this pragmatic preference, V receives policy benefits 1 if either policy succeeds, 0 if either policy fails or none of the policies are implemented. In addition, V receives electoral benefits 1 if V approves a politician whose ideal policy is the same as V’s ideal policy, and 0 otherwise. The tie-breaking rules are specified in Supplemental Material.

The game begins with C’s choice of m_p, a funding level for implementing policy p. All players observe C’s choice. If C proposed $m_{p} \leq {\underline{m}}_{p}$ , then the implementation of policy p is blocked by C, so P’s choice is skipped and the game ends. If C proposes $m_{p} > {\underline{m}}_{p}$ , then P must decide whether to implement policy p with m_p or not. If P accepts to implement p with m_p, then the outcome of p is realized and observed by all players. If P rejects, policy p is not implemented. Finally, V chooses whether to approve P or C, and the game ends.⁴ The solution concept is perfect Bayesian Nash equilibrium.

Results

V’s choice does not affect V’s policy benefits because the outcome of policy p is determined before V makes a choice, but it affects V’s electoral benefits. Thus, V’s optimal choice is to approve the player whose policy is more likely to be a good type. Since V makes a decision after observing the outcome of policy p, V’s ideal policy may change according to V’s posterior belief. Let $θ_{p}^{e}$ denote V’s posterior belief that policy p is a good type after an event e ∈ {N, S, F}, where N, S, F, respectively denote non-implementation, success, and failure. Then, we see that $θ_{p}^{N} = θ_{p}$ , $θ_{p}^{S} = \frac{θ_{p} q_{G}}{θ_{p} (q_{G} - q_{B}) + q_{B}}$ , and $θ_{p}^{F} = \frac{θ_{p} (1 - b (m_{p}) q_{G})}{1 + θ_{p} b (m_{p}) (q_{B} - q_{G}) - b (m_{p}) q_{B}}$ . Meanwhile, V’s posterior belief that policy c is a good type is the same as the prior belief because policy c is not implemented, so V does not have an opportunity to observe the outcome of policy c to update her belief. Then, V’s optimal decision is to approve P if $θ_{p}^{e} > θ_{c}$ , while approving C is the optimal decision if $θ_{p}^{e} \leq θ_{c}$ .

Notice that a successful outcome of policy p makes p more likely to be a good type (i.e., $θ_{p}^{S} > θ_{p}$ ), while the failure of policy p leads p less likely to be a good type (i.e., $θ_{p}^{F} < θ_{p}$ ). Furthermore, the higher b (m_p) is, the lower $θ_{p}^{F}$ is. This is because the failure of policy p is more likely based on it being a bad type when sufficient funds are provided. Therefore, if C needs to change V’s policy from p to c, C must show V the failure of policy p, which requires C to provide funds to implement policy p in the first place. Furthermore, C must provide sufficient funds to lead the failure of policy p to make policy p look worse than policy c.5

Otherwise, V would conclude that the failure of policy p mostly comes from insufficient funds and would maintain p as V’s ideal policy despite the realization of the failure.6 However, providing such sufficient funds increases the probability of the success of policy p, which will not only lead to negative policy benefits for C but also a loss of V’s approval. This is the crucial trade-off, turning C’s decision into gambling on the failure of policy p. Proposition 1 presents a main finding:

Proposition 1

In an equilibrium, the following holds:

(1) If θ_p ≤ θ_c, then C blocks the implementation of policy p by choosing $m_{p}^{*}$ such that $m_{p}^{*} < {\underline{m}}_{p}$ .

(2) If θ_p > θ_c and $\frac{(θ_{p} - θ_{c}) (θ_{p} (q_{G} - q_{B}) + q_{B})}{θ_{p} (1 - θ_{p}) (q_{G} - q_{B})} \leq π \leq \frac{θ_{p} (q_{G} - q_{B}) + q_{B}}{1 + θ_{p} (q_{B} - q_{G}) - q_{B}}$ , then C chooses $m_{p}^{*}$ such that $b (m_{p}^{*}) = \frac{π}{(1 + π) (θ_{p} (q_{G} - q_{B}) + q_{B})}$ and P implements policy p.

(3) Otherwise, policy p is never implemented.

All proofs are in Supplemental Material. The first part of Proposition one states that C blocks the implementation of policy p when V’s ex-ante ideal policy is policy c. The intuition is straightforward: There is no reason for C to change V’s ideal policy by implementing policy p. If policy p is not implemented, then V cannot learn anything about policy p, so V’s posterior belief about policy p is the same as the prior belief. Then, c still remains as V’s ideal policy, which leads V to approve C. Furthermore, blocking policy p guarantees the prevention of the policy p’s success that yields negative policy benefits to C. Therefore, C can receive the highest payoff π by blocking policy p.

It seems natural to expect then that C would provide funds to implement policy p when C needs to change V’s ideal policy from p to c by showing the failure of policy p. However, the second and third parts of Proposition 1 state that this is not always the case. The implementation of policy p is possible only when the political benefits for P and C from V’s approval, π, are moderate. There are two reasons for this. First, the implementation of policy p is basically C’s gambling on the failure of policy p to acquire V’s approval. To make the failure of policy p convincing evidence that policy p is worse than c, C must provide a sufficiently high level of funding. However, this high level of funding increases the risk that policy p succeeds. Therefore, if C’s aspiration for political benefits is not sufficiently strong (i.e., $π < \frac{(θ_{p} - θ_{c}) (θ_{p} (q_{G} - q_{B}) + q_{B})}{θ_{p} (1 - θ_{p}) (q_{G} - q_{B})}$ ), then it is better for C to block the implementation of policy p to avoid negative policy benefits for sure than to take the risk of the success of policy p to receive relatively less important political benefits.

Second, when π is too high (i.e., $π > \frac{θ_{p} (q_{G} - q_{B}) + q_{B}}{1 + θ_{p} (q_{B} - q_{G}) - q_{B}}$ ), then C is willing to gamble; but now P is not willing to implement policy p. Given that V’s ideal policy is already p, P does not need to show the success of policy p to maintain it as V’s ideal policy. The only reason for P to implement policy p is to gain policy benefits. So, P is willing to implement policy p only if the risk of failure is sufficiently low given that the failure of policy p will change V’s ideal policy from p to c. This risk-aversion of P becomes stronger as P cares about political benefits more seriously. Although C can reduce the risk of failure by providing a sufficiently high level of funding in an attempt to induce P to implement policy p, C cannot reduce the risk enough to induce P to implement policy p if P cares about π too much as in this case.

Therefore, if the political benefits π is moderate, both P and C are willing to implement policy p if both players see that the funding level does not produce too much risk of success (from C’s perspective) or failure (from P’s perspective).⁷ P is willing to take the risk of failure only when the funding level is at least $m_{p}^{*}$ such that $b (m_{p}^{*}) = \frac{π}{(1 + π) (θ_{p} (q_{G} - q_{B}) + q_{B})}$ , which produces the highest risk of the success that C is willing to take. Thus, we see that both P and C agree to implement policy p at the above funding level, although they hope for different policy outcomes.

There are two takeaways from the result. First, what moves the pendulum between congressional deference and defiance is whether voters support the president’s preferred policy or congress’s preferred policy. This finding may explain the puzzling shifts from congressional deference to congressional defiance of the president’s policy in policy areas where congressional deference is expected due to the president’s informational advantages. Second, what seems to be Congress surrendering to the president may be Congress gambling on failure to convince voters that the president’s policy is worse than that of Congress. They do not defer to the president in hope of the success of the president’s policy but simply let the president fail to garner voters’ approvals.

Voter’s welfare

Recall that V receives 1 if a policy produces a successful outcome. It is always better for V to at least try to implement a policy rather than non-implementation in terms of V’s ex-ante expected payoff. Moreover, the higher funding level is better for V’s welfare because the probability of policy success increases as the funding level increases. Although all decisions that directly affect the policy outcome are made by P and C before V makes the approval decision, V still can affect their decision through their concerns about political benefits. Then, is it good for V’s welfare if we have politicians with strong aspirations for political benefits?

If policy c is believed to be better than policy p, then the answer is no because C does not implement policy p regardless of how much C cares about the political benefits. If policy p is in advantage with V’s belief, however, then the answer might be yes because C would not have provided funds to implement policy p had it not been for C’s aspiration for political benefits. However, as Proposition 1 states, the non-implementation of policy is still possible. To find a clear answer, we now focus on this case. I relax the assumption that the political benefits for P and C are the same. Proposition 2 provides the answer:

Proposition 2

Suppose θ_p > θ_c. Policy p is implemented with $m_{p}^{*}$ such that $b (m_{p}^{*}) = \max \{\frac{π_{P}}{(1 + π_{P}) (θ_{p} (q_{G} - q_{B}) + q_{B})}, \frac{θ_{p} - θ_{c}}{q_{G} θ_{p} (1 - θ_{c}) - q_{B} θ_{c} (1 - θ_{p})}\}$ if and only if (i) π_C ≥ π_P, (ii) $π_{C} \geq \frac{(θ_{p} - θ_{c}) (θ_{p} (q_{G} - q_{B}) + q_{B})}{θ_{p} (1 - θ_{p}) (q_{G} - q_{B})}$ , and (iii) $π_{P} \leq \frac{θ_{P} (q_{G} - q_{B}) + q_{B}}{1 + θ_{P} (q_{B} - q_{G}) - q_{B}}$ .

The second and third conditions of Proposition 2 are simple modifications according to the new assumption, and the intuition behind them is the same as that of Proposition 1. However, the first condition now clearly demonstrates that the implementation of policy p is possible only when C cares about political benefits more seriously than P. To implement policy p, C must reduce the risk of failure (from P’s perspective) enough to gain P’s acceptance by providing sufficiently large funds. But recall that providing sufficiently large funds increases the risk of success (from C’s perspective) at the same time. If C does not care about the political benefits at least as seriously as P, then C never gambles on failure, resulting in non-implementation. Therefore, a Congress with strong aspirations for political benefits (i.e., large π_C) is desirable for V’s welfare because it motivates C to implement policy p as gambling on failure, which is close to Madison’s view that public opinion should play a dominant role in politics.

In contrast, the president with a moderate aspiration for political benefits is desirable for V’s welfare, which is somewhat in line with Hamilton’s view that political elites should be insulated from public opinion (see Sheehan 2004). To see why, notice that the equilibrium funding level is defined π_P. This means that the equilibrium funding level weakly increases as π_P increases up to some point. However, if π_P becomes too large, then P chooses the safe option of non-implementation to maintain V’s approval.

Conclusion

This article demonstrates that Congress defers to presidents even in the absence of asymmetric information and a common interest in success, which provides an alternative interpretation of congressional deference: Congress defers to presidents not because they consider presidents’ policies are more likely to succeed but because they want to show the failure of presidents’ policies. Therefore, bi-partisan support for presidents may not necessarily be the consequence of Congress’s statesmanship but Congress’s gamble for a partisan goal.

There is no doubt that the presidential informational advantage or Congress’s statesmanship may result in congressional deference. However, it is possible that congressional deference may be the joint product of these two different dynamics: statesmanship and gambling on failure. Recognizing these dynamics would allow us to have a better understanding of how public opinion influences the relationship between the executive and Congress.

Supplemental Material

sj-pdf-1-rap-10.1177_20531680221093435 – Supplemental Material for Let presidents fail: Congressional deference to presidents as gambling on failure

Supplemental Material, sj-pdf-1-rap-10.1177_20531680221093435 for Let presidents fail: Congressional deference to presidents as gambling on failure by Myunghoon Kang in Research and Politics

Footnotes

Acknowlegment

The author thanks Randall Calvert, Justin Fox, Andrew Reeves, Keith Schnakenberg, Jon Rogowski, Alexander Hirsch, Hye Young You, Jinhee Jo, the seminar participants at Sogang University, and the anonymous reviewers for their constructive and helpful comments and suggestions. This work was supported by 2021 Research Fund of Myongji University.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was supported by 2021 Research Fund of Myongji University.

ORCID iD

Myunghoon Kang

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Canes-Wrone

Howell

Lewis

(2008) Toward a broader understanding of presidential power: a reevaluation of the two presidencies thesis. The Journal of Politics 70(1): 1–16.

Curry

(2008) Pelosi: ’We Will Not Cut Off Funding’ for Iraq. NBC News. December 16 https://www.nbcnews.com/id/wbna16057734 (Accessed on 2 February 2021).

Dahl

(1950) Congress and Foreign Policy. Westport, CT: Greenwood.

Dewan

Hortala-Vallve

(2019) Electoral competition, control and learning. British Journal of Political Science 49(3): 923–939.

Gailmard

Patty

(2012) Learning while Governing: Expertise and Accountability in the Executive Branch. Chicago, IL: University of Chicago Press.

Groseclose

McCarty

(2001) The politics of blame: bargaining before an audience. American Journal of Political Science 45(1): 100–119.

Hirsch

(2016) Experimentation and persuasion in political organizations. American Political Science Review 110(1): 68–84.

Hirsch

Kastellec

(2022) A theory of policy sabotage. Journal of Theoretical Politics. doi: 10.1177/09516298221085974.

Howell

Jackman

Rogowski

(2013) The Wartime President. Chicago, IL: University of Chicago Press.

10.

Kartik

McAfee

(2007) Signaling character in electoral competition. American Economic Review 97(3): 852–870.

11.

Kriner

(2010) After the Rubicon: Congress, Presidents, and the Politics of Waging War. Chicago, IL: University of Chicago Press.

12.

Lee

(2009) Beyond Ideology: Politics, Principles, and Partisanship in the U.S. Senate. Chicago, IL: University of Chicago Press.

13.

Lee

(2013) Presidents and party teams: the politics of debt limits and executive oversight, 2001-2013. Presidential Studies Quarterly 43(4): 775–791.

14.

Lindsay

(2003) Deference and defiance: the shifting rhythms of executive-legislative relations in foreign policy. Presidential Studies Quarterly 33(3): 530–546.

15.

Park

(2021) When Do Politicians Grandstand? Measuring Message Politics in Committee Hearings. The Journal of Politics 83(1): 214–228.

16.

Schlesinger

(2004) War and the American Presidency. New York, NY: W.W.Norton.

17.

Sheehan

(2004) Madison v. hamilton: the battle over republicanism and the role of public opinion. American Political Science Review 98(3): 405–424.

18.

Wildavsky

(1966) The two presidencies. Trans-action 4(2): 7–14.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.10 MB