How uncertainty about judicial nominees can distort the confirmation process

1. Introduction

For most of U.S. history, judicial nominees have refused to answer questions in public venues. They do so on the grounds that answering questions impugns their impartiality as jurists and undermines judicial independence. Robert Bork, the most famous failed Supreme Court nominee, wrote after he was rejected by the Senate that forcing judicial candidates to answer questions “effectively compel[s] nominees to make campaign promises or face the possibility of rejection” (Bork, 2009). Clarence Thomas, when asked during his hearings whether Roe v Wade was correctly decided, replied unconvincingly that he did not have an opinion “one way or the other” (Nomination of Clarence Thomas, 1991). And Ruth Bader Ginsburg started her hearings saying she would provide “no hints, no forecasts, no previews” (Nomination of Ruth Bader Ginsburg, 1993).

These responses are typical. Today, nearly all nominees shield their true beliefs before the Senate and in public hearings. However, why judicial nominees are allowed to keep private their views on issues likely to come before the courts remains an open puzzle. Why don’t the President and the Senate come together to ensure that candidates answer questions? We argue that one explanation behind this institutional opaqueness comes from the information environment surrounding judicial nominations. Using the Supreme Court as our primary example, we model how the Senate assesses the kind of nominee put forth by the President. We capture this by departing from the literature and formulating a game of complete but imperfect information between the President and Senate. More accurately reflecting the reality noted by legal scholars, we model the President as having comparably more information about a nominee’s exact ideological, policy, and legal positions than Senate or the public (Eisgruber, 2009; Lively, 1985), who rely on comparably weaker signals coming from public hearings and other indirect channels. The Senate must then decide whether to confirm or reject the nominee.

The model yields three critical insights. First, when the President and Senate are ideologically distant, the Senate cannot trust the President to select acceptable moderates in the absence of strong signals. To deter the President from consistently choosing extremist justices, the Senate sometimes rejects nominees even without direct evidence that the nominee is unacceptable. The President benefits occasionally from the lack of information, as he may sometimes sneak his most preferred (i.e., more extreme) choice. However, in expectation, both parties are worse off under these conditions than if they agreed on a moderate nominee. Second, and as a direct consequence of the first insight, having political capital hurts the President. When the President has great political capital, the chance that the signal will reveal an extremist fails to deter the President from trying to hoodwink the Senate. In turn, the Senate cannot trust the President to nominate a moderate and responds by sometimes rejecting unknown moderates. This leaves the President in worse shape than if the parties simply agreed on a mutually preferable nominee.

Given that the lack of transparency is costly, why aren’t nominees compelled to address how they would vote on important issues before the Court? Our third insight provides one answer. When the President and Senate are ideologically close, the President benefits from leaving the process opaque—that is, allowing his nominees to avoid answering tough questions. Under such conditions, he knows that the Senate will confirm his most preferred nominee. However, keeping the process nontransparent shields the President from suffering reputation costs from selecting more like-minded (and possibly extreme) judges. This is a key finding from our analysis and explains why a seemingly counter-intuitive practice—allowing nominees to avoid answering questions in public fora—has turned into an important, long-standing institutional feature of judicial nominations.

This paper proceeds as follows. Section 2 discusses the existing nominations literature, focusing specifically on the information environment. In Section 3, we describe the model and its equilibria before generalizing the key results. Section 4 discusses why allowing nominee reticence sometimes benefits both the President and the Senate, while Section 5 provides vignettes illustrating the application of the model. Finally, we summarize the results and consider extensions in Section 6.

2. Uncertainty between the Senate and the President in the confirmation process

Why are judicial nominees allows to keep their views private? Regarding uncertainty about judicial nominations, the literature falls into two camps. The first approach, taken by most theoretical literature (and some empirical papers as well), is that ideological positions are known with good certainty by the relevant political actors at the beginning of the process. Thus, a number of papers assume complete and perfect information in showing that the President can aim to manipulate the Supreme Court composition via the appointments process (Krehbiel, 2007; Moraski and Shipan, 1999; Rohde and Shepsle, 2007), or that Senators strategically evaluate the distance between them (or their constituents) and the nominee in deciding how to vote (Cameron et al., 1990; Johnson and Roberts, 2005; Segal et al., 1992). Other literature has extended the assumption of complete and perfect information to additional Senate voting configurations (Primo et al., 2008), to multiple time periods (Jo et al., 2012), and to possible obstruction via Senatorial courtesy (Jacobi, 2005). Whatever the focus, a thread running through much of this formal literature is the assumption that both the President and the Senate have complete and perfect information (Moraski and Shipan, 1999) and that they know with accuracy the ideological positioning of nominees (Cameron et al., 1990; Segal et al., 1992).

However, these substantive assumptions conflict with the legal literature and several empirical papers, which demonstrate that political actors often fail to predict nominees’ positions due to the little information they have. ¹ Most importantly for this analysis is that the Senate often has less direct ways of assessing candidates’ ideologies than the President. Indeed, although both the President and the Senate might have noisy signals about nominees’ ideologies, the President—as the one who who selects the nominee—can “assess judicial philosophies quite well by relying on two kinds of information: publicly available sources, such as judicial opinions, and private conversations” (Eisgruber, 2009). This is consistent with papers such as Shipan and Shannon (2003), which considers the role that the Senate’s uncertainty about the ideology of the nominee could have in incentivizing Senators to further delay confirmation.

In line with these observations, we do not assume that Senators have no information; instead, we explore how the Senate has invariably less information than the President. Indeed, it is significantly more likely for the Senate, but not the President, to be surprised by candidates (e.g., Harriet Miers); however, to the authors’ knowledge, no instances exist where the Senate was aware of key information while the President was caught by surprise. Underlying this information environment is the notion that the Senate usually tries to decipher what type of candidate the President has nominated—an extremist, a moderate, etc. As such, it is unsurprising that the Senate rejects, or refuses to vote on, certain nominees about whom it has insufficient information. ² True, the President simply misjudging what the Senate will accept might explain some rejections; however, such a simple explanation overlooks key examples wherein the President clearly had more information about a nominee than the Senate. The 2005 nomination of Harriet Miers illustrates this idea. As Eisgruber (2009: 161) notes “[T]he president probably had more information than did the Senate about Harriet Miers, who was a personal friend of his, but who had a relatively low profile in Washington and the national legal community.” Our approach to the information asymmetry therefore reflects that “Senators also have access to informal, private information about a nominee,” but it is “not necessarily the same information that the president has” (Eisgruber, 2009). ³

Reinforcing the difference in information is the fact that the Senate’s primary information-gathering device, the confirmation hearing, provides little insight into candidates’ judicial philosophies (Eisgruber, 2009). Indeed, nominees to the courts routinely invoke judicial prerogatives to keep views before the Senate private (Lively, 1985; Ringhand, 2008; Strauss and Sunstein, 1991; Totenberg, 1987), a practice that has engendered significant criticism (Eisgruber, 2009; Kagan, 1995; Post and Siegel, 2006; Turley, 2005, 2009; Watkins, 2010). Ruth Bader Ginsburg, who began her hearings by vowing to provide “no hints, no forecasts, no previews,” set the modern standard of judicial evasiveness before the Senate and thus lends her name to the so-called “Ginsburg Rule.” All subsequent nominees have invoked some form of the Ginsburg Rule; John Roberts, for example, explicitly cited the Ginsburg Rule 10 times in his confirmation hearings (Sarat, 2008). Nominees are also highly selective of what views they share, being more likely to invoke the Ginsburg Rule on controversial issues such as reproductive rights or executive power (Ringhand, 2008). Even Elena Kagan—who wrote in her pre-nomination days as a law professor that the Senate had ceased “to engage nominees in meaningful discussion of legal issues” and that the confirmation process was little more than “a vapid and hollow charade” (Kagan, 1995)—refused to answer questions at her hearings on the grounds that a nominee simply “has to be protective of certain kinds of interests” (Nomination of Elena Kagan, 2010).

In part because of nominees’ evasiveness, critics have argued that the Senate is fundamentally hampered in its ability to acquire information, which further widens the information gap with the President. As Ringhand (2008) notes, a nominee’s ability to avoid discussion makes “the senators tentative and hesitant in their questioning, undermining their ability to confidently play a meaningful constitutional role in the confirmation process” (Ringhand, 2008). Nominees have effectively drawn lines to limit “the capacity of the Senate to acquire useful information about a nominee’s constitutional commitments” (Post and Siegel, 2006), while the hearings themselves “evolved into a stylized ritual of moves and countermoves” in which “nominees respond carefully and avoid revealing any more than is absolutely necessary” (Eisgruber, 2009). Strauss and Sunstein (1991) summarize these sentiments:

there is now practically a script: the nominee is open-ended, has ‘no agenda,’ enthusiastically accepts Brown v Board of Education and Griswold v Connecticut, is humbled by the difficulty of being a Justice, and admires Justice Harlan. The nominees commit themselves to liberal-sounding principles of privacy and racial and gender justice; but the commitments are at such a high level of platitudinous abstraction that they reveal nothing about the nominees’ views on controversial issues.

To sum, the empirical literature suggests that predicting a nominee’s true ideological position is difficult, while legal observers note that the Senate is usually in a disadvantaged position compared to the President. In addition, as confirmation hearings have increasingly become irrelevant for information gathering, Senators must seek other pathways to obtain information; when it comes to candidates who are virtually unknown to the Senate (and more known to the President), this becomes extremely difficult. Despite this, most of the modeling literature tends to assume both the President and the Senate know the nominee’s ideology with precision. The discrepancy leads us to the core of our inquiry, which is (1) how varying levels of information between the President and the Senate could affect the selection and confirmation process and (2) why nominees are not forced to be more forthcoming about their policy views.

3. A model of nominee uncertainty

To illustrate how uncertainty creates inefficiency, we begin with a stylized model of judicial nominees, using the Supreme Court as an example. ⁴ Although simple, the model establishes the existence of a commitment problem in judicial nominations. Afterward, we show that this commitment problem persists in richer environments.

3.1. Players and structure

The game has two players, the President and the Senate (or the median senator, Moraski and Shipan (1999)). To establish our baseline result, suppose the President can choose between only two nominees: the President’s most preferred nominee (what we call an extremist) or a nominee that both parties prefer to the status quo (what we call a moderate). (Later, we show our key results extend to situations in which the President selects a nominee from a larger pool, in line with the standard spatial model assumptions.) The choice is private information to the President. ⁵

After the President moves, Nature chooses whether to send a signal. If the nominee is moderate, Nature sends no signal; if the nominee is extremist, Nature sends a signal with probability p ∈ (0,1). ⁶ This probability is exogenously given, though we later show that the commitment problem holds if the signal is partially a function of costly effort from the Senate. In any case, if the Senate receives a signal, it knows that the nominee is an extremist. But if the Senate receives no signal, it cannot directly infer the nominee’s type. ⁷ With or without the signal, the Senate chooses whether to confirm or reject the nominee. The game then ends.

We adopt standard spatial preferences to derive payoffs. The President has an ideal point P ∈ R and the Senate has an ideal point S ∈ R . The Court as currently constructed generates a status quo policy Q ∈ R . Without loss of generality, suppose Q < P. Confirming a moderate repositions the court’s ideology to M ∈ R and confirming an extremist yields E ∈ R , with M < E. Players’ policy payoffs are therefore the negative of the distance between their ideal point and the implemented policy after the nomination game ends. In addition, if the President nominates an extremist and Nature reveals his choice, the President pays a cost c > 0 to reflect that the President is “caught” trying to maneuver an extremist onto the Court. Note that this cost occurs regardless of whether the Senate confirms the nominee. ⁸ Though c could reflect many things, political capital is a compelling consideration. Large amounts of capital correspond to small values of c since the President does not care much about being discovered, while less capital corresponds to larger values.

To rule out trivial cases, we impose two restrictions on the parameter space: −|M − S| > −|Q − S| and −|E − P| > −|M − P| > −|Q − P|. Combined, these restrictions ensure that the President and Senate prefer a moderate to maintaining the status quo. If these did not hold, the agreement set would be empty and the parties would be deadlocked. Meanwhile, the second restriction also ensures that the President prefers extremist nominees to moderates. If this was not the case, the parties would trivially agree on moderate nominees. Figure 1 displays the game in extensive form.

OPEN IN VIEWER

Figure 1.

The game’s extensive form.

3.2. Equilibrium

Since the game has complete but imperfect information, perfect Bayesian equilibrium (PBE) is the appropriate solution concept. ⁹ Our first result concerns the game’s equilibrium generally:

Proposition 1. Equilibrium exists and is unique for all non-knife-edge parameters.

The uniqueness result may come as a surprise, as PBE often yields multiple equilibria depending on off-the-equilibrium-path beliefs, especially when the informed actor moves first. Our model lacks this issue because only two things can be true about the Senate’s updated belief: (1) the President nominates an extremist with positive probability so all information sets are on the equilibrium path or (2) the President nominates a moderate as a pure strategy but the Senate knows the President must have nominated an extremist if it receives a signal off the equilibrium path. In either case, the Senate has a unique consistent set of beliefs.

We now turn to the outcome of the interaction, breaking up the parameter space into four regions as Figure 2 illustrates.

OPEN IN VIEWER

Figure 2.

The equilibrium outcome as a function of the Senate’s ideological position and the strength of Nature’s signal. The strategies and reasoning for both Proposition 2 and Proposition 4 apply to the top right parameter space.

Proposition 2. (Risk Deterrence) If the expected reputation cost of nominating an extremist outweighs the possible benefit (i.e., if pc > |M − P| − |E − P|), the President nominates a moderate and the Senate confirms if and only if it does not learn that the nominee is an extremist.

Intuitively, if the risk of being “exposed” trying to nominate an extremist and the punishment for doing so are sufficiently large, the President settles for a moderate (Figure 2, upper part of the diagram). Note that if the President nominates an extremist, his best case scenario is that the Senate confirms regardless of its signal. The President earns p(−|E − P| − c) + (1 − p)(−|E − P|) = −|E − p| − pc here. Alternatively, the President could appoint a moderate and receive −|M − P|. Thus, the President prefers appointing the moderate if pc > |M − P| − |E − P|. The Appendix verifies that the Senate would always confirm upon receiving no signal and that the equilibrium is unique.

Proposition 3. (Tolerance) If the expected reputation cost of an extremist is worth the possible benefit and the Senate finds extremists acceptable (i.e., if pc < |M − P| − |E − P| and −|E − S| > −|Q − S|), the President nominates an extremist and the Senate confirms in all circumstances.

Proof: If the Senate tolerates an extremist, confirming strictly dominates rejecting for the Senate at each of its information sets. By iterated elimination of strictly dominated strategies, the President safely nominates the extremist and reaches his optimal outcome.    □

Consequently, for the parties to be in conflict, −|E − S| < −|Q − S| must hold. The remainder of the proof covers the situation where the Senate prefers vacancy to a confirmed extremist. Our next proposition shows that the appointment process still runs smoothly as long as Nature’s signal is sufficiently strong.

Proposition 4. (Senate Deterrence) If signals are moderately strong relative to the President’s reputation cost and the Senate finds extremists unacceptable (i.e., if p > p * ≡ | M − P | − | E − P | | Q − P | − | E − P | + c and−|E − S| < −|Q − S|), the President nominates a moderate. The Senate confirms but would reject if it learned that the nominee was an extremist.

For intuition, consider the game if p = 1, meaning the signal is perfectly informative. Backward induction yields the solution, as the game has perfect information. The Senate rejects if the President offers an extremist and confirms if the President offers a moderate. Since the President prefers confirming a moderate to the status quo, the President nominates a moderate.

This logic holds with sufficiently little noise. Strong signals deter the President from nominating an extremist. After all, Nature will reveal the nominee’s type with a high degree of probability, and the Senate will in turn give the President his punishment payoff of |Q − P| − c. The Senate confirms the candidate upon not receiving a signal because it trusts that the risk of exposure deters the President from picking an extremist.

The Appendix formally derives the equilibrium. Most of the work is finding the critical probability below which Nature deters the President from nominating an extremist. This equals | M − P | − | E − P | | Q − P | − | E − P | + c , which appears in Proposition 4.

Proposition 5. (Commitment Problem) If signals are weak relative to the President’s reputation cost and the Senate is unwilling to confirm an extremist (i.e., p < p * and−|E − S| < −|Q − S|), the President nominates an extremist with probability σ M * = ( 1 − p ) ( | E − S | − | Q − S | ) ( 1 − p ) | E − S | + p | Q − S | − | M − S | . If the Senate learns the nominee is an extremist, it rejects; otherwise, its posterior belief that the nominee is moderate equals | E − S | − | Q − S | | E − S | − | M − S | , and it confirms with probability σ C * = pc | M − P | − p | Q − P | − ( 1 − p ) | E − P | .

Thus, the President sometimes nominates a moderate and sometimes nominates an extremist; the Senate sometimes confirms and sometimes rejects the nominee when it fails to receive a signal. That is, the President sometimes bluffs and sometimes does not, while the Senate sometimes calls potential bluffs and sometimes does not. As a result, many outcomes occur with positive probability. Sometimes a moderate fills the seat, and sometimes the President successfully tricks the Senate with an extremist. Sometimes the seat remains vacant, and sometimes the signal catches the President and he pays the reputation cost.

Figure 3 provides the intuition. Given the choice between appointing an extremist and nominating a moderate, the President prefers an extremist. However, the ideological positioning of the court with a confirmed extremist is not within the set of policies the Senate prefers to the status quo. As a result, if the President nominates an extremist as a pure strategy, the Senate knows that any nominee it receives is an extremist will reject regardless of the signal.

OPEN IN VIEWER

Figure 3.

Parameters that lead to a commitment problem. Letters correspond to the ideal points and policy positions of the various actors. Peaks show the range of respective policies the Senate and President prefer to the status quo.

The President cannot credibly commit to nominating a moderate as a pure strategy either. The Senate would confirm with certainty even without receiving the signal in that case since it infers the nominee is a moderate and the moderate shifts policy in a way the Senate prefers to the status quo. However, the temptation to nominate an extremist is irresistible—because p is low, the President prefers gambling on the extremist (hoping to trick the Senate) to sticking with the safe-bet moderate. The game is finite and thus has an equilibrium, meaning the President must mix.

The same holds for the Senate when it does not receive a signal. If it always rejects, the President would nominate a moderate to save on the reputation cost. But this causes the Senate to want to confirm because it prefers the moderate to the status quo. If it always accepts, the President would nominate an extremist because the gamble is worthwhile in that case. Yet this leads the Senate to want to reject with certainty. So the Senate must mix as well. The Appendix gives a full proof and derives the cutpoints and mixing probabilities Proposition 5 describes. ¹⁰

3.3. Loss of welfare under weak signals and great political capital

When the Senate is tolerant of the President’s decision, obfuscating institutions assist in confirming ideological allies. However, consider the parties’ welfare when the Senate prefers rejecting extremists. Our main theorem summarizes the result:

Theorem. If the Senate finds extremists unacceptable (i.e., −|E−S| < −|Q−S|), both actors’ payoffs are weakly increasing in the strength of the signal.

Proving the theorem requires investigating how each player’s payoff fluctuates as p increases. This requires some work because the players’ equilibrium strategies change as a function of it.

Figure 2 provides a helpful guide. The theorem concerns the right half, in which the Senate prefers rejecting known extremists. Deterrence prevails in the top two-thirds, sometimes because of the risks of exposure and sometimes because of the Senate’s punishment if the President is caught nominating an extremist. Either way, the President plays it safe and selects a moderate. In the bottom half, these risks are minimal because of the weak signal, leading the commitment problem and inefficiency. The theorem says that moving up the Figure weakly increases both players’ payoffs.

To begin the proof, note that the top two-thirds of the right side lead to a deterministic outcome: the President nominates a moderate and the Senate confirms with or without a signal. The Senate receives its best possible outcome under such conditions. Meanwhile, the President receives −|M − P|.

In contrast, when the signal is weak, or p < p^*, the actors encounter the commitment problem, leading to smaller payoffs for both parties than when p > p^*. To see this, consider each player’s equilibrium payoff. Since the players are mixing and therefore indifferent between their pure strategies, the President’s payoff equals his expected utility for nominating a moderate:

U ( moderate ) = σ C * ( − | M − P | ) + ( 1 − σ C * ) ( − | Q − P | )

Since −|M − P| > −|Q − P|, the President receives a strictly worse payoff than if the Senate confirmed a moderate. That said, σ C * = pc | M − P | − p | Q − P | − ( 1 − p ) | E − P | is strictly increasing in p in Proposition 5’s parameter space. This means that increasing the signal gives the President a greater share of the good moderate outcome and less of the bad status quo outcome. The President’s payoff then jumps to −|M − P| when the signal crosses the p^* threshold.

Similarly, the Senate’s welfare equals its payoff for rejecting. Here, the Senate earns −|Q − S| regardless of the nominee, so the Senate receives −|Q − S| in equilibrium for all values of p < p^*. But this is clearly worse than confirming a moderate, as doing so generates the greatest possible payoff for the Senate.□

More succinctly, these results highlight how the outcome in which the Senate confirms a moderate Pareto dominates the equilibrium outcome under weak signals. Strong signals prevent the President from nominating ideologically extreme justices. Thus, it may appear that the President would benefit from obscuring the nominee’s true positioning. But while the President sometimes tricks the Senate when signals are weak, the hidden information is ultimately detrimental to everyone.

Before moving on, it is worth noting that we can restate the theorem as the following corollary:

Corollary. (Paradox of Political Capital) If the Senate finds extremists unacceptable (i.e., −|E−S| < −|Q−S|), both actors’ payoffs are weakly increasing in the reputation cost.

While it is not surprising that the Senate’s payoff increases here, it is strange that the President’s does. After all, the reputation cost (in part) represents how much political capital the President is willing to spend to obtain his more preferred outcomes. Yet the corollary says that this political capital ultimately hurts the President—he is better off having no capital to spend than being willing to pay great reputation costs.

The intuition once again comes down to the commitment problem. ¹¹ When the reputation cost is great, even weak signals deter the President from risking nominating an extremist. This builds trust with the Senate, allowing the parties to confirm a moderate. However, when the President is willing to gamble, the Senate sometimes rejects to protect itself from extremists. This leads to an increase in the President’s welfare as the reputation cost increases.

3.4. Robustness checks

As with any stylized model, ours makes tradeoffs between simplifying assumptions and empirical plausibility. Cognizant that our results might be sensitive to particular modeling choices, we address some major concerns. In each case, the commitment problem persists in the alternate specification. Thus, the stylized model captures a consistent barrier to reaching efficient agreements.

Costly information acquisition. To begin, note that the time between nomination and confirmation or rejection is implicitly exogenous because p is fixed. Consistent with empirical results from Shipan and Shannon (2003), one might believe that the informational discrepancy between the President and the Senate results from the President’s ability to research potential nominees for years before a Supreme Court vacancy. In turn, the Senate might narrow the gap by engaging in costly information acquisition. ¹² The Senate could then pass damaging revelations to the public, triggering the reputation cost from the model. Consequently, we may wonder whether inefficiency persists if an ideologically-opposed Senate could instead delay confirmation, learn about the nominee, and vote later.

However, even under these conditions, it is straightforward to verify that the actors would still mix under such conditions and thus inefficiency still results. For the Senate to pay for costly information acquisition, the President must nominate an extremist with positive probability. But the President cannot appoint an extremist with certainty since the Senate would never confirm and the President would suffer the reputation cost with positive probability.

Whether the Senate would then engage in information acquisition depends on how costly such an action is relative to the precision of the signal it generates. But regardless, inefficiency persists. The Senate must sometimes confirm and sometimes reject or delay confirmation, otherwise the President would be unwilling to mix. If the Senate does not acquire information, the inefficiency is the same as in the model. If the Senate acquires information, then inefficiency results purely from the delay. ¹³ Either way, our central claim still holds: uncertainty causes mutually undesirable outcomes when the President and the Senate are ideologically unaligned.

To some degree, the limited Senate hearings mimic such costly information acquisition—opposition Senators spend days fishing for red flags in an environment capable of publicly shaming the President. However, with Robert Bork’s damaging hearing a distant memory, the question still remains why the President and Senate do not alter the rules to make hearings more productive.

Defections by moderate nominees. The model as described ignores potential actions by nominees and whether they may volunteer information in a manner inconsistent with the role of Nature. That is, while the President has incentives to obscure when he nominates an extremist, one might wonder why moderate nominees would keep mum under these conditions. After all, without information transmission, the Senate will reject them with positive probability (believing that they might be extremist). On the other hand, they could answer committee questions in full to signal their moderate status.

The most straightforward counterargument is that every candidate—be they moderate or extreme—has an incentive to appear moderate. Even extreme nominees would answer questions in the same fashion as more moderate candidates, resulting in any signals (i.e., answers to any questions) being tantamount to non-credible cheap talk. This is borne out by the confirmation hearings of most nominees (with some exceptions—e.g., Bork); what limited responses are given tend to be middle-of-the-road and, to the extent that nominees voice opinions about legal issues, they do so regarding non-controversial settled issues of law, for example segregation or civil rights. ¹⁴ In addition, because there may be little benefit for sharing information but many potential downsides, especially for extremists, nominees by and large stay quiet or talk only about “safe” subjects.

A second response concerns the potential pool of candidates. The pool of candidates—that is, lawyers more generally—have an interest in maintaining the independence and legitimacy of the judiciary. If statements made in public venues are a commitment toward a certain kind of decision making, then invested nominees may later on incur costs associated with deviating from those stated intentions; in turn, this would undermine the rewards of having a judicial career (Bork, 2009). Along these lines, a long-standing norm to appear “above politics” exists among judicial candidates. Out of concern for institutional legitimacy, we would expect strategic candidates to shy away from taking positions that could be viewed as overly “political” and thus damage the Court’s institutional and popular standing (Eisgruber, 2009; Kagan, 1995). Indeed, the Justices themselves appear highly sensitive to attacks that the Court has become too “political” over time (e.g., Bazelon, 2013). Pooling on silence is a safe way to avoid politicking.

In addition, having significant and stringent inquiry into policy and moral positions may discourage qualified candidates from coming forward (Carter, 1994). For these reasons, lawyers’ organizations like the American Bar Association have strong institutional incentives to protect the pool of potential nominees, both moderate and extremist. Indeed, as evidence of this, the American Bar Association has formally instituted rules coercing individuals into pooling on silence under a formal “Code of Judicial Conduct” (American Bar Association, 2010). Under this and similar sorts of code, judges who speak publicly about issues that might arise would have to recuse themselves later on. This is a stiff punishment for those who defect.

Larger candidate pool. Lastly, our model restricts the President to nominating two types of justices. However, our key result—Pareto inefficiency when the Senate is intolerant of the President’s most preferred choice—holds if the President instead selects from n potential nominees.

Three assumptions generate this result. First, a potential nominee exists who is mutually preferable to the payoffs for maintaining the status quo ideology of the Court; this rules out the uninteresting cases of gridlock. Second, let the reputation cost associated with each nominee be increasing as the ideological positioning of the nominee drifts away from the status quo. And third, the Senate prefers the status quo to confirming the President’s most preferred nominee.

To prove that inefficiency still occurs with multiple nominees, consider the following proof by contradiction. If an equilibrium outcome is efficient, then confirmation must occur with probability 1; if there is positive probability of rejection, both sides would be better off if they confirmed one of the mutually preferable nominees rather than suffer bargaining breakdown that percentage of the time. So suppose that the Senate confirms with probability 1 upon not receiving a signal. The above assumptions state that a potential nominee exists that the President prefers to the most favorable nominee that the Senate would be willing to confirm with perfect information. Let c be the reputation cost associated with the President’s most favorable mutually acceptable nominee and c′ be the reputation cost associated with the President’s more preferred nominee. (By the above assumptions, c′ > c.) Similarly, let the Court’s ideal point under the most favorable nominee be x and the ideal point under the President’s more preferred nominee be x′. Then President would prefer nominating the more extreme candidate if

− | x ′ − P | ( 1 − p ) + p ( − | Q − P | − c ′ ) > − | x − P | − pc

Since −|x′ − P| > −|x − P|, the left side of the inequality is greater for sufficiently small values of p. However, this means that the President would want to deviate to a nominee that the Senate would prefer to reject. In turn, no equilibrium exists in which the Senate confirms with probability 1. Nevertheless, the game is finite and has an equilibrium. Therefore, the equilibrium must be inefficient.

Figure 4 further illustrates the logic. The values x_i reflect the ideological positioning of the Court following the confirmation of nominee i. Because nominees exist that both the President and Senate prefer to maintaining the status quo, the parties would like to reach an agreement. However, if the Senate confirms with certainty, the President has incentive to cheat and nominate someone more extreme. As long as the information environment is sufficiently noisy (to reduce the risk of cheating) and the reputation costs incurred are sufficiently low, the Senate cannot confirm with certainty. Thus, the main intuition—that uncertainty damages the bargaining process—holds for richer environments.

OPEN IN VIEWER

Figure 4.

Illustration of why bargaining still fails with n nominees, with x_i representing a potential nominee’s ideal point. Because nominees exist that the President would want to sneak in, the Senate must continue mixing to deter the nomination of extremists.

4. Why is uncertainty tolerated?

At present, the actors tolerate informal institutions like the Ginsburg Rule, which obscure pertinent information. As shown here, however, both sides could improve their welfare by increasing the strength of signal. So why don’t the President and Senate create a more transparent system?

One trivial explanation is that no institutional features can compel judicial nominees to divulge more information. We find this unconvincing. Undoubtedly, no institutional feature could lead to total information revelation. For starters, simple cheap talk from the President to the Senate is insufficient to reveal information due to the incentives to misrepresent. Consequently, any effective reform could not be made on an ad hoc basis but rather would require meaningful institutional reform. That said, ABA guidelines instruct nominees to keep their views private. Moreover, Supreme Court nominees are frequently experienced appellate attorneys who can deftly evade direct answers to the simplest of questions.

Fortunately, however, our results show that complete transparency is unnecessary to reach better outcomes. Indeed, the theorem states that any increase in information weakly increases welfare when the Senate and President are in disagreement. As such, the question is not whether different institutions can compel nominees to divulge all information but rather whether different institutions could lead to more information.

If nominees are not the obstacle to reform, what is? The tolerance outcome provides an explanation. One critical component of tolerance is ideological convergence between the President and Senate. In these cases, the utility of allowing judicial candidates to keep information private is clear. The President recognizes that he can induce the Senate to confirm an extremist; allowing the nominee to keep information private reduces the costs associated with being exposed. However, in instances where p is close to 1 (i.e., compelled disclosure by nominees, perhaps under a rule change), extremist candidates will almost certainly be unmasked—if not by the Senate majority, then almost certainly by members of the minority party. Thus, the President would incur costs for nominating an extremist and would be deterred. The President therefore benefits from obfuscation since it allows him to sidestep the costs from having selected an extremist and moves the parameters into his more preferred territory, the bottom left of Figure 2. Nominee evasiveness also benefits extremist Senators, who would be more abundant under ideological convergence.

We thus have our primary justification for why nontransparent institutions remain in place: they allow Presidents and like-minded Senators the possibility of appointing ideological allies. We further note that instances of ideological alignment represent an overwhelming proportion of Supreme Court observations (see Table 1). In light of these considerations, few reasons exist for Democrats and Republicans to come together to engage in meaningful institutional reform.

OPEN IN VIEWER

Table 1.

Supreme Court nominees, Ford through Obama administrations.

Name	President	Year	Hearing Length (days)	Vote	Outcome	Senate
Elena Kagan	Obama	2010	3	63-37	Confirmed	D
Sonia Sotomayor	Obama	2009	4	68-31	Confirmed	D
Samuel Alito, Jr	Bush II	2005	4	58-42	Confirmed	R
Harriet Miers	Bush II	2005	NA	NA	Withdrawn	R
John Roberts	Bush II	2005	4	78-22	Confirmed	R
Stephen Breyer	Clinton	1994	4	87-9	Confirmed	D
Ruth Bader Ginsburg	Clinton	1993	4	96-3	Confirmed	D
Clarence Thomas	Bush I	1991	11	52-48	Confirmed	D
David Souter	Bush I	1990	5	90-9	Confirmed	D
Anthony Kennedy	Reagan	1987	3	97-0	Confirmed	D
Douglas Ginsburg	Reagan	1987	NA	NA	Withdrawn	D
Robert Bork	Reagan	1987	12	42-58	Rejected	D
Antonin Scalia	Reagan	1986	2	98-0	Confirmed	R
William Rehnquist *	Reagan	1986	4	65-33	Confirmed	R
Sandra Day O’Connor	Reagan	1981	3	99-0	Confirmed	R
John Paul Stevens	Ford	1975	3	98-0	Confirmed	D

*

Nomination from Associate Justice to Chief Justice

In addition, we expect that continuing pressure on nominees to be more forthcoming over the course of confirmation hearings will come not from the majority party (when the majority party is the same as the President’s), but from the minority party. This is exactly what we see, with recent opposition to John Roberts’ and Samuel Alito’s evasiveness coming from Democrats and opposition to Elena Kagan’s and Sonia Sotomayor’s evasiveness coming from Republicans.

5. Illustrations of the model

Figure 5 includes some key Supreme Court nominations from the last 25 years. Note that the President and Senate majority have usually been ideologically aligned at the time of vacancies. The model predicts that we should generally see agreement, with rejection only occurring mostly when the President and Senate are ideologically divided.

OPEN IN VIEWER

Figure 5.

Model predictions based on ex ante information.

However, applying the logic of the model requires more detailed examinations of specific examples. We therefore illustrate the Propositions with individual cases:

Proposition 2: Risk Deterrence. For instances of strong signals or high reputation costs, we look to the nominations of Robert Bork and Anthony Kennedy during the Reagan Administration, both in 1987. The Bork nomination was notable because Bork repeatedly violated the “Ginsburg Rule,” stating his thoughts on subjects like abortion, civil rights, and freedom of speech—to the surprise of everyone, including the Reagan White House (Crawford Greenburg, 2007). ¹⁵ Bork was therefore revealed over the course of his confirmation hearings as being (in our terminology) an extremist (Totenberg, 1987). Thus, it is no surprise to us that the ideologically distant Senate rejected Bork’s nomination, as we would expect the Senate do so when an extremist is unmasked. We also note that Reagan was subsequently punished in political discourse and in the press for attempting to nominate someone who had been unmasked as a clear extremist.

Recall that the cutpoint for the strength of signal dimension depends on c; the higher the potential reputation cost, the more likely the President is to appoint a moderate. In part due to the increased risks associated with nominating yet another extremist, Reagan’s next two picks for the Democratic-controlled Senate were moderates. Douglas Ginsburg, the first, was withdrawn after allegations of marijuana use, revealed after his nomination in a surprise report from NPR’s Nina Totenberg. (In this sense, Ginsburg’s failed nomination provides an excellent illustration of the capricious reevaluations engendered by Nature, which is p in our model.) However, the model fits Kennedy’s confirmation. At the time of his nomination, Kennedy came from the 9th Circuit with a fairly detailed history of decision making. This, combined with the high costs to Reagan after Bork’s failed nomination, meant that Kennedy was confirmed by a unanimous Senate.

Proposition 3: Tolerance. Most nominations fall under this proposition, for which we have a like-minded President and Senate (see Figure 5). Two examples illustrate this. The first is the nomination of Antonin Scalia, who was nominated during the first Reagan term under a Republican-controlled Senate. At the time, Scalia had spent most of his time in academia and later as a judge on the DC circuit appeals court. However, “Scalia was nine years younger than Bork and Scalia did not have the controversial paper trial that Bork did” (Staab, 2006). Moreover, Scalia (again, unlike Bork) flawlessly performed at his confirmation hearings, consistently relying on his privilege to be non-responsive. Throughout, Scalia “refused to bite at the bait, refusing to testify about any legal issue that could possibly come before him as a Supreme Court justice” (Staab, 2006). With the press unable to find any skeletons in his closet, the Senate confirmed him, 98-0.

The second example is George W Bush’s nomination of John Roberts to Chief Justice under a Republican-controlled Senate. The worry for Bush was not nominating an extremist, but nominating the right kind of extremist (Toobin, 2008). Under this proposition, Bush’s interests would be to nominate someone who was fairly conservative. Roberts was a relatively straightforward choice. Although he had served as a lower-court judge for three years before being nominated, Roberts was exceedingly careful to limit his writings on plausibly ideological topics (Toobin, 2008). Accordingly, Roberts had no problem in securing a nomination. Neither did he express any views in his confirmation hearing that could potentially be cast as extreme, invoking the “Ginsburg Rule” explicitly in over 10 instances (Sarat, 2008).

Proposition 4: Deterrence. Delineating the cut-off between Proposition 4 and Proposition 5 is difficult ex post, as it relies on the relative ratios of p and c. However, several instances appear to fit Proposition 4, in which moderates are nominated and confirmed under (1) Senate opposition and (2) moderate information/reputation costs. The best example on point is David Souter, who was nominated during the Bush I administration with a Democratic Senate. Souter, then a federal appeals judge, had authored several opinions, but not enough to have a clearly delineated “paper trail.” Indeed, commentary at the time emphasized that Souter had “not given a speech, written a law review article or, as far as anyone knows, taken a position on the correctness of the Supreme Court precedents on abortion or any other issue” (Greenhouse, 1990). Given these conditions, why Bush did not take the opportunity to nominate someone more conservative is surprising; we believe that the reason is because Bush himself misestimated Souter’s own ideological positioning as a New England conservative. Thus, ex ante, it is possible that Bush was seeking to achieve Proposition 5. This leads to the Thomas nomination, discussed below.

Proposition 5: Commitment Problem Documenting instances of the commitment problem has obvious obstacles. The model predicts mixing behavior—the President sometimes appoints a moderate and sometimes appoints an extremist, while the Senate sometimes confirms and sometimes rejects in the absence of a signal. Thus, the prediction supports virtually all observable outcomes. ¹⁶ While we are cognizant of the observable indeterminacy this implies, comparative statics of c and p still reveal the relative likelihood of particular outcomes occurring.

Our first example is Harriet Miers, George W Bush’s personal attorney, who has perhaps been the most low-information candidate in modern times. She had no intellectual writings, no clerkship or judicial experience, and little exposure to appellate litigation. However, the Senate was in Republican hands, and, thus, the expectation under Proposition 3 is that Bush should have appointed a staunch conservative (bottom left of Figure 2). However, recent evidence suggests that Miers was more of a centrist, and that Bush was making the nomination largely on personal—rather than ideological—grounds (Toobin, 2008). Republican senators consequently led the charge against Miers, fearful of her unknown judicial identity.

In relating the Miers case to the model, note that the “moderate” and “extremist” terminology from the model are more precisely “acceptable” and “unacceptable” nominees from the Senate’s perspective. Given the strategic uncertainty, the Senate could not pin down Miers’ type. Nevertheless, recall that the Senate confirms with probability pc | M − P | − p | Q − P | − ( 1 − p ) | E − P | for these parameters. At the time of Miers’ appointment, there were many reasons to believe Bush’s reputation cost c for appointing the wrong candidate was extremely low. First, Bush had just won reelection and had publicly declared that he had “political capital” and he “intended to spend it” (Knowlton, 2004). Second, Bush, as a second-term President, had no electoral motives. And third, Miers’ presumed centrist positioning would have meant the reputation cost would have come from his own party. When c is so low, the confirmation probability of pc | M − P | − p | Q − P | − ( 1 − p ) | E − P | goes toward 0. Indeed, Senate Republicans’ disapproval of Miers led Bush to withdraw her nomination without a vote. ¹⁷

The second example, Clarence Thomas, is more straightforward. Thomas, a former Equal Employment Opportunity Commission chairman, was a relative unknown at the time of his confirmation. Although some suspected that Thomas had conservative views, the full extent of his conservatism was shielded from public view at the time of his nomination. One reason was that Thomas had relatively few intellectual writings. Another reason was that, over the course of the confirmation hearings before a Democratic Senate, Thomas repeatedly denied requests to explain his political and legal philosophies. (Asked about his thoughts on Roe v Wade, Thomas responded that he hadn’t given it much thought.) This, combined, with the allegations of sexual harassment against Anita Hill, made the hearings particularly unproductive from an information-gathering perspective. Thus, we group Thomas into the category of extremists who were never unmasked (low p), with the opposition-controlled Senate being unable to make a determination with accuracy. Thomas may also be considered an instance where the President also misinterpreted the nominee’s ideology; a moderate conservative like Bush may have been unlikely to nominate Thomas had he known the nature of Thomas’ beliefs on originalism or civil rights. In addition, the revelations about Anita Hill really came to light after the nomination, suggesting another instance (like Douglas Ginsburg’s) in which the press exogenously provided some, though clearly not all, information.

Cases orthogonal to strategic uncertainty. Our model does not predict a handful of nominations, primarily because they were nominated for reasons orthogonal to ideology. One illustrative anomaly is Sandra Day O’Connor, who, as a former Arizona state senator and state appeals judge, was a national unknown. With Republicans controlling the Senate, our model would predict that Reagan would have capitalized on nominating someone so unknown by nominating an extremist; however, he did not. Reagan had earlier announced his intention of nominating the first female Justice; thus, he was constrained to the pool of female candidates, which likely contained fewer conservative “extremists.” O’Connor went on to be a moderate on the Court.

6. Conclusion

Voting whether to confirm a nominee is tricky for the Senate, which must determine what type of candidate the President has chosen. On one hand, Senators want to confirm nominees that reflect a fair compromise between the branches; on the other, they wish to reject nominees that who are ideologically extreme. With a clear understanding of a nominee’s preferences, the decision is straightforward. However, the information gap between the Senate and the President potentially leaves the Senate unsure of what to do.

Whereas prior models have simplified bargaining environments to remove this hurdle, our model addresses it. When the Senate and the President have aligned ideologies, the Senate trusts the President to nominate an acceptable candidate and subsequently confirms the nominees. However, when their ideal points are distant, trust breaks down. To avoid exploitation, the Senate rejects nominees despite not receiving clear red flags. Inefficiency results and has become institutionalized over time to the point where, today, the system appears unlikely to change. After all, in instances where the Senate and the President are ideologically aligned (which represents a majority of circumstances in modern times), the President benefits from having a low information environment, as it decreases the probability that an ideologically like-minded candidate is exposed. Our model provides a rationale for why nominees are sometimes rejected, which complete information models can only do in a ad hoc way. In addition, our model produces the counter-intuitive insight that political capital can actually hurt the P resident by making the Senate even more skeptical of his nominee’s extremity.

We conclude with two thoughts. First, although we focused on Supreme Court nominations, the model’s intuition maps onto other judicial nominations, including lower-court nominations. Not only are trial or appeals courts nominees more obscure, but media also pay less attention to these nominations. To this extent, the (lack of) information conveyed in confirmation hearings may be significantly more important, particularly when the norm of Senatorial courtesy is not followed (e.g., when the home-state Senators and the President are of opposing parties, or for appeals court nominations). Restricting the flow of information makes the process more unpredictable, allowing for the confirmation of candidates that would never otherwise be confirmed. In this sense, Supreme Court nominees, as already established intellectuals or former judges, present the most conservative application of our model; we should be as, if not more, worried about information uncertainty for lower-court nominations. Indeed, even under Senatorial courtesy (for example, for certain district court nominations), we might think that the information transmitted might vary according to the ideological distance between the Senators and the President. In both instances, the framework we develop here applies, and future research should take into account these different kinds of informational environments.

The logic also applies to non-judicial nominations. The President appoints not only federal judges, but ambassadors, members of his Cabinet, and other administrative officials; these are instances where the President nearly always has more information than the Senate, which is left to figure out what type of candidate the President has named. For example, the President will know with more certainty that his nominee to head the EPA favors the development of natural gas or that his candidate to head the Department of Education opposes standardized testing. In instances where such candidates easily shield their policy positions, we might see similar inefficiency. We do note, however, that uncertainty is perhaps most salient for judicial nominees, owing to the high degree of protection they have in refusing to answer questions. As the judiciary is a coequal branch of government and as judges have lifetime tenure, the uncertainty for judicial nominations is perhaps the most important and also the most problematic.