Contemporary theories of executive appointments use the framework of the spatial model, in which two or more actors strive to minimize the distance between their ideal point and the appointee's, settling on a nominee located somewhere in the middle. Yet appointees make thousands of decisions with some degree of autonomy from their appointing president and confirming Senate. Spatial models miss the mark by failing to recognize that the confirmation process is both a political negotiation and a principal-agent problem.
Although the selection of a new Supreme Court justice poses an informational problem for presidents, prior research presenting spatial models of judicial appointments (Johnson and Roberts 2005; Krehbiel 2007; Moraski and Shipan 1999; Rohde and Shepsle 2007) assumes that presidents have perfect information about future behavior. These models offer a useful heuristic in illustrating the strategic interaction between the actors involved, and this article will not suggest that the implications of these models should be discarded. However, a shortcoming of these models is their failure to predict or explain the notable instances wherein a president expressed disappointment in his choice. Further, the institutional alignments that are the heart of these models exhibit sensitive statistical relationships with the relative level of congruence between justice and appointing president.
Conventional wisdom suggests that the behavior of Earl Warren, David Souter, and others deviated considerably from the ideological expectations of their appointing presidents (Nemacheck 2007, 44). Indeed, "no more Souters" is a well-worn rallying cry of conservative activists. Yet the theoretical depiction of the appointment process as a strategic game between the Senate and president to move the Court's median implicitly predicts that disappointments should never occur. While journalistic accounts of presidential frustration do not prove that spatial models are incorrect or that presidents are routinely disappointed with their choices, these accounts do provide motivation for systematically investigating the factors associated with presidential success in all Court appointments. Probing this relationship requires analyzing all appointments for which we have reliable data, including, but not limited to, the selected anecdotes favored by journalists. If disappointments and instances of poor justice-president congruence occur not randomly, but in identifiable patterns, then the assumptions of spatial theories should be reassessed, since these assumptions lead to avoidable systematic errors. Specifically, the principal-agent problem implicit in the selection process must be taken into account. Because information is not symmetric between the principal and possible agents, the greater the president's ability to ascertain a potential nominee's ideology, the greater will be the congruence between the president's preferences and the justice's behavior.
In this article, I will demonstrate that the proximity of the president's ideal point to the subsequent behavior of his appointee depends substantially on the amount of information available to the president in the form of the relevant professional experience. Further, the president's ability to appoint predictable, experienced justices is conditional on the level of constraint imposed by the positions of the Senate and Court medians. Whenever all potential gains by the president (changes to the Court's median in the president's favor) would be realized as losses by the Senate, presidents appear to accept the uncertainty of a less-experienced nominee in order to secure confirmation. The findings suggest that scholars should consider synthesizing the principal-agent problem with the institutional alignment of the president and Senate for a richer and more accurate depiction of the process.
Presidents face an unusually difficult principal-agent problem when appointing a new Supreme Court justice. The principal-agent relationship is primarily one of asymmetry between the authority of the principal and the information of the agent. In the canonical model, the principal's authority derives from control over the incentives of the agent (Miller 2005). The principal delegates decision-making responsibility to the agent, but uses incentives, monitoring, and contracts to ensure that the agent's efforts maximize the principal's utility. But the incentives of a Supreme Court justice cannot be altered by presidential flat; justices serve life terms without any formal constraints on their behavior. Therefore, while the president must delegate the decision-making responsibility to the nominated justice, he cannot induce desired behavior by rewarding or punishing the justice's decisions on the Court. The president can only improve his fortunes by reducing the comparative advantage of the agent.
The agent's advantage in the principal-agent relationship is informational: the agent knows things that the principal does not, which affect the payoffs of both the principal and the agent. In the case of Supreme Court nominations, the nominees almost certainly have more information about how they will rule on future cases than does the president. As such, presidents have a powerful incentive to avoid disappointment by accurately forecasting the future behavior of their nominee. Any information that reduces the president's uncertainty about how the nominee might rule on future cases should therefore improve the congruence between the president and the appointee.
I argue that nominees for whom the president has more relevant information are more likely to be congruent with presidential preferences. Conceptually, congruent behavior is the extent to which a justice's rulings are as close to the president's ideal point as possible on average, as modeled in formal theories of the process. To operationalize justice-president congruence, I utilize data (Bailey 2007) that demonstrate that presidents have specific policy preferences over Supreme Court decisions and that measure how the president would have ruled if he himself, rather than his nominated justice, were on the Court--I then compare these preferences with the rulings of the justice herself to obtain an aggregate measurement of congruence. With this definition in mind, I analyze each successful Supreme Court nomination from Earl Warren to Samuel Alito. I find that nominees whose backgrounds offered the president more relevant information, in the form of executive branch experience and federal judicial experience, are significantly more congruent with their appointing president. The relationship between justice-president congruence and the ideological distance between the president and Senate at the time of confirmation is weaker and sensitive to the test chosen. Further, the average amount of nominee experience varies systematically by the level of constraint on the president at the time of nomination (Moraski and Shipan 1999). The results suggest that while strategic spatial interaction is important, so too are informational concerns; synthesizing these two concerns would be a worthy goal for theoretical progress on this problem.
Imperfect Information in Court Appointments
The polarized atmosphere surrounding modern appointments reduces the quantity and quality of information available to the president. Wittes (2009, 1502) remarks: "supporters [of judicial nominees] find themselves unable to acknowledge truths that might justify no-votes by senators, while opponents feel obliged to allege far more than they can prove ... about the substance of the nominee's views." The nominees themselves are reticent to disclose their preferences. Recall Justice Ginsburg's remark at her confirmation hearing: "I cannot say one word on that subject that would not violate what I had said to be my rule about no hints, no forecasts, no previews" (Meese and Gaziano 2005).
Since potential nominees may not divulge information candidly throughout the process, presidents must look elsewhere for clues about a potential justice's future behavior. As the amount of relevant information about a nominee increases, the president's estimate of the nominee's ideal point necessarily improves. Thus, nominees whose backgrounds provide more relevant information to the president should be more congruent, ceteris paribus. Despite the fact that nominees vary considerably in the amount of information their backgrounds provide to the president, four prominent theories of the process (Johnson and Roberts 2005; Krehbiel 2007; Moraski and Shipan 1999; Rohde and Shepsle 2007) assume that the president can perfectly forecast at which point on the real line a nominee will be at on the Court's ideological scale. (1) But Cameron and Park (2009), in their work devising a new scaling metric for Supreme Court nominees, demonstrate that predicting a future justice's behavior once on the Court is far from a straightforward task. Informational variance in nominees helps to explain why some nominees are more congruent than others. Given that the Senate is surely aware of the nominee's background as well, the president must act strategically by selecting less informative nominees who stand a better chance of confirmation under certain institutional configurations. Refining move-the-median theories to incorporate the reality of imperfect information would better predict and explain presidential success. (2)
Professional Experience as Information
Prior work on Supreme Court appointments both posits and finds that professional experience relates to subsequent behavior by justices. Szmer and Songer (2005) find that nominee experience in various arenas has positive and negative effects on the variance of judicial voting behavior. However, Szmer and Songer only consider the effect of experience on volatility. This article directly models justice-president congruence as a function of professional...