Executive appointments: Duration,ideology,and hierarchy

Abstract

While contemporary scholars generally view the Senate’s nominee approval role as impacting bureaucratic capacities and the president’s ability to realize campaign pledges, empiricists and theorists focus on different elements of bargaining. Since empiricists typically study confirmation delays, and theorists normally analyze equilibrium nomination preferences, theory and data rarely inform one another.

We remedy this by specifying an executive appointment model jointly incorporating delays and appointee ideologies. Besides predicting appointees’ equilibrium ideologies, and contrary to past claims about the relationship between ideology and duration, this theory details how ideological differences between the president and his opposition do not straightforwardly induce longer delays; rather, effects are conditioned by factors such as the office’s policy importance and divided government. Additionally, different pathways for parties to impact appointee ideology and duration are flexibly incorporated. Empirically, theoretical hypotheses receive support and evidence of parties impacting Senate trade-offs between delay and policy outcomes and successfully pressuring key members over high-stakes appointments are uncovered.

Keywords

Appointments inter-branch bargaining legislature–executive relations

1. Introduction

A basic element of separation of powers systems is contestation over bureaucratic control. Scholars have focused on many of the resulting battles, including those over initial structural choices (e.g. Moe, 1989), rules dictating how bureaucratic decisions are made (e.g. McCubbins et al., 1989), and when legislatures delegate agencies’ discretion (e.g. Epstein and O’Halloran, 1999; Huber and Shipan, 2002). Recently, another inter-institutional conflict impacting control over appointments has received increased scrutiny. Although requiring senatorial approval, agency appointments have routinely been considered a principal presidential tool of influence in the face of formidable obstacles (e.g. Lewis, 2008, 2011; McCarty, 2004). Traditionally, it was assumed that presidential nominations would go through without incident; what Bond et al. (2009; see also Krutz et al., 1998) labelled the presumption of success. Related to this, it was asserted that the president was advantaged by a norm of senatorial deference (e.g. Gerhardt, 2000; Hammond and Hill, 1993; Ho, 2007). However, scholars increasingly acknowledge that this apparent flexing of presidential muscle cannot simply be assumed.¹

One reason, particularly voiced by empiricists, for this realization is that there has been a lengthening of time to confirmation. This is not a trivial change, given the president’s circumscribed period in office and that political appointees typically serve for short periods. Regarding the latter, for example, roughly half of all successful nominees depart by the end of their second year (Wood and Marchbanks, 2008). As such, even delays of several months are a significant percentage of many appointees’ tenures and, intuitively, a considerable barrier to the president implementing favorable policies. One recent manifestation of this problem occurred early in the Obama administration, where only 151 of the more than 1100 Senate confirmable nominations had been approved by the end of May 2009; around this time, there were 31 nominations that, having cleared all other hurdles, were simply waiting for Senate floor scheduling (Ornstein, 2009; see also O’Connell, 2010). The result of this failure to agree, it is frequently claimed, is a crisis in governance and yet another cause of gridlock, with the Senate’s blocking of high-level executive appointments coming in for most of the ensuing blame. For example, a Brookings Institution press release in July 2002 was blunt: ‘The Senate, not Bush administration, is to blame for current delay in confirmation process’ (Brookings Institution, 2002).

Those theoretically inclined have largely taken a different avenue. Such scholars have modeled the appointment process as a strategic game in which the chief executive uses his agenda-setting power to select a nominee and the Senate acts as a veto player over confirmation. Notably, despite confirmation duration’s rise and its substantive importance, such studies focus almost exclusively on how the president’s and the Senate’s ideological preferences impact the ideal points of political appointees chosen in equilibrium (e.g. Bertelli and Feldmann, 2006; McCarty, 2004; Nokken and Sala, 2000; Rohde and Shepsle, 2007; Snyder and Weingast, 2000; see Console-Battilana and Shepsle (2009) for an analysis also including interest groups) or, more recently, their competence (e.g. Jo and Rothenberg, 2012). Put differently, the major stylized fact of recent American politics regarding nominations, duration’s increasing relevance, is not theoretically well-integrated. Existing models neither offer predictions about duration per se nor consider the possibility that, in equilibrium, the president waits longer for approval to get an appointee more to his liking confirmed.

Conversely, as foreshadowed, with so few nominations rejected outright (and problems measuring nominee ideology in a common space with the president and the Senate), empirical scholars have largely focused on specifying factors impacting appointment duration, although they typically downplay the importance of presidential choices in impacting time to approval (e.g. Derouen et al., 2005; McCarty and Razaghian, 1999; Nixon, 2001; Nixon and Bentley, 2006; but see Nixon, 2004). Of these works, McCarty and Razaghian (1999) stands out, particularly given our desire to have a broad-gauged theory including appointee ideology and duration.

Specifically, McCarty and Razaghian assume that the president trades off nominating appointees who share his policy preferences with appointees who will receive quick confirmation. They also assume that the Senate majority party is the dominant force in the upper chamber; specifically, party polarization (opposition to the president’s views in the upper chamber, as measured by the ideological differences between the majority and minority medians), divided government, and their interaction, are keys for duration.² They hypothesize both that the hostility toward presidential appointments created by Senate party polarization and the ability of adversarial majority parties to employ obstructionist institutional prerogatives (e.g. scheduling) directly and interactively lengthen delay. Their empirical analysis of executive nominations from 1885 to 1996 supports their conjectures for so-called second- and third-tier appointments, but results for key variables pertaining to top-tier nominations at the apex of the bureaucratic hierarchy are not encouraging.

Despite its contributions, McCarty and Razaghian’s analysis leaves several questions unanswered. One is, when trading off between policy gains and delay costs in the face of party polarization, why does the president always prefer to endure greater duration rather than to trade away policy gains in equilibrium? Analogously, why, in equilibrium, does the Senate majority party, given divided government, always favor more duration rather than somewhat less preference alignment but quicker confirmation? Additionally, why are parties, whose importance is often debunked, or at least seen as modest relative to preferences (Krehbiel, 1998; McCarty et al., 2001), seemingly so crucial? Quite importantly given the considerable relevance of such appointments, what explains and differentiates high-level nominees from others? Finally, what are the linkages by which the equilibrium selection of appointees, divided government, polarization, and the hierarchical structure of appointments all fit together?

These conundrums point to the need for a theoretical model of strategic appointments incorporating duration as well as appointees’ equilibrium ideologies. Ideally, such a model should tell us under which conditions factors such as party polarization and divided government impact duration, what determines the president’s equilibrium selection of appointees, how party influence could work, and whether other factors, notably the level of bureaucratic office a nominee will fill, might matter.

Hence, we propose and analyze empirically an executive appointment model that sheds light on the interplay between confirmation delays and appointee ideology, permits examination of mechanisms by which party and divided government might influence this process, and highlights how an office’s importance impacts the findings generated by a formal analysis of the nomination process. Theoretically, besides predicting appointees’ equilibrium ideologies, and contrary to past claims and intuition, we demonstrate that ideological differences between the president and his opposition do not straightforwardly induce longer delays; rather, effects are conditioned by factors such as the office’s policy importance. Also, we find that the interactive effect between president–opposition ideological differences and divided government lengthens confirmation time. As for appointment level, we demonstrate that its direct impact is non-monotonic. Our empirical analysis largely confirms these theoretical predictions and demonstrates that parties influence confirmation delay by the weight that the Senate places on the costs of delay relative to the benefits of policy outcomes and, for top-tier appointees only, by moving the veto pivot’s location.

2. Model

As few nominees are rejected outright (four in our data set), we assume that players are concerned with the policy consequences of stalling a given nominee and with the costs of delay.³ Assume that a nominee has single-peaked, symmetric, preferences with ideal point x_a. Also, assume that the president, P, and the Senate filibuster pivot, E, are the only two relevant players in the appointment process. For now, E simply represents an extreme player. Conceivably, she could be the filibuster pivot defined by a pure preference model (Krehbiel, 1998), the opposing party’s median given a world of pure party pressure (McCarty and Razaghian, 1999), or a member subject to partial party pressure (Chiou and Rothenberg, 2009). Although we talk in terms of the filibuster pivot for expositional simplicity, this pivot could actually be any potential alternative legislative pivot, as long as she is the most extreme.

Without loss of generality, we assume a conservative president (in the American context, a Republican) and a liberal pivot (a Democrat), with the former’s utility function being

U_{p} (y, t | p) = - β t^{2} - | p - y |

(1)

where y, t, and p denote the policy outcome chosen by the nominee when successfully confirmed, the length of time taken in proceeding with the nomination, and the president’s ideal point, respectively.⁴ Presidential delay costs are a function of interrelated political impacts other than those directly associated with which policy is chosen in the ideological space, such as those costs that might be incurred if the delay of the nominee being considered impacts the speed of other nominees also in the appointment queue being acted on, in turn possibly resulting in less desirable policies because these other appointees are slowed in assuming office. Related to this, the president’s delay costs may be associated with his performance being evaluated more harshly.⁵ We assume that the president’s utility strictly decreases with more delays (i.e. higher t).

We employ β (assumed to be positive) to capture the weight that the president assigns to the costs of delay versus policy consequences, with smaller (higher) values associated with nominees who will have a greater (lesser) policy impact. Consistent with McCarty and Razaghian (1999), we assume that higher-level nominees influence policy outcomes more, and that the president, consequently, is willing to pay higher delay costs to ensure his ideal nominees relative to lower-level appointments.

As for the filibuster pivot, her utility function is given by

U_{E} (y, t | x_{e}) = - γ t^{2} + t | x_{a} - x_{e} | - | x_{e} - y |

(2)

where x_e denotes the ideal point of the Senate filibuster pivot. The pivot is assumed to be concerned with the cost of delay (for a similar assumption see Koger (2010)), her preference divergence with the nominee, and with policy outcomes; note that the pivot’s delay costs are mainly related to audience costs when obstructing or stalling a nominee. The second term of the right-hand side of equation (2) captures the benefit of delaying a nominee with different preferences—the pivot will not obstruct if the nominee’s ideal point coincides with hers and will increasingly delay as this preference divergence grows—while the third term represents the pivot’s concerns over policy outcomes, although this term plays no role in her delaying decision.⁶

We use γ, which is assumed to be strictly positive, to capture the weight that the pivot assigns the costs versus the benefits of delaying.⁷ As part of this weighting process, we assume that, everything else being equal, E hesitates more in obstructing higher-level nominees because of potential audience costs (be they emanating from the mass public or organized interests). In other words, for the pivot, while stalling a low-level appointment will likely not create a ripple of negative attention outside of the Washington Beltway (and little inside), stonewalling a high-end appointment (e.g. a cabinet secretary) may potentially make her life quite difficult by attracting attention from various sources (for evidence that negative perceptions of congressional performance hurt incumbent senators, see e.g. Jones and McDermott (2009).⁸ Hence, the pivot suffers more from obstructing higher- than lower-level appointees, and the corresponding per unit of time marginal effects of delay on her utility are higher. Therefore, we assume that γ is larger for higher-level nominees. Also, assuming that partisan politics operate such that a pivot with a different party affiliation from the president cares less about the administration’s performance (or even has an incentive to sabotage it), such a pivot will weigh the costs of delay less and, hence, have a lower γ.⁹

The game sequence is (1) the president proposes a nominee, $x_{a} \in ℜ$ (2) the pivot chooses a delay time, $t \in ℜ^{+}$ ; (3) Nature randomly selects a status quo, q, from a uniform distribution $[\underline{q}, \bar{q}]$ and reveals it to all players;¹⁰ (4) the confirmed political appointee selects a policy, y; and (5) the president chooses whether to veto y, sustaining q, or to accept it. Subgame perfect equilibrium is used as a solution concept.¹¹ To simplify, we let q = 0 and $\bar{q} = 1$ and, without loss of generality, we assume that 0 < x_e < p < 1. Since we assume that the pivot is more liberal than the president, to ensure that the status quo distribution is not greatly lopsided in favor of the president or the pivot, and that the support of the distribution covers the president and the pivot well, we assume that p > 1/2 and p − x_e ≤ 1/2, with the case of p − x_e > 1/2 being addressed in Appendix 2.¹²

Before solving our executive appointment game, it is helpful to point out three prominent features. First, integrating the notion that the chief executive has more influence than Congress in controlling the appointee’s policy implementation choices, we incorporate a presidential veto of the political appointee’s policy; while the president and the Senate lack formal veto rights,¹³ the chief executive’s facilitating institutions, such as the Office of Management and Budget, and his ability to remove political appointees from office, should give him the upper hand. Given the uncertainties that the Senate and the president face about what policies the nominee will deal with upon confirmation, we assume that the political appointee has a choice over which policy to adopt. Finally, consistent with the upper chamber holding nominees up via a variety of tactics—committee intransigence, manipulation of the floor’s schedule, threat or use of the filibuster—rather than rejecting them, and with presidential anticipation of such actions in selecting a nominee, we assume that Senate choices are over delay and not rejection.¹⁴

Second, our model allows us to examine how hierarchical level shapes the appointment process, and how this level interacts with other factors. Specifically, we can think of an appointment’s hierarchical level as captured by β, the president’s weighing of policy gains versus the cost of delay, and by γ, the pivot’s weighting.

Finally, our model is flexible enough to examine varied means of party influence, shedding light on the scholarly debate about whether party matters. More precisely, it allows for two mechanisms of party influence: (a) shifting E’s location; and (b) conditioning E’s ability to obstruct or, equivalently, her weighting of delay’s costs, regardless of the pivot’s ideological location. By the first mechanism, partisan influence would help determine the filibuster pivot’s location. With no partisan impact, E is the filibuster pivot defined by Krehbiel (1998) and, if parties strengthen, E locates increasingly close to the opposition party’s median.¹⁵ In the spirit of McCarty and Razaghian (1999), the second mechanism operates if being in the majority party allows E to delay more effectively; this is associated with divided rather than unified government and is captured by γ. Equivalently, this mechanism can work if E, when a member of the opposition party, is less vested in the electoral prospects of the president or his party than a majority party pivot. Note that, by distinguishing these two possible party influence mechanisms, we also explicate why divided government might matter, rather than merely ‘black boxing’ its importance.

With these distinguishing features in mind, we return to solving our executive appointment game. To do so, we derive possible policy outcomes for any status quo for a given x_a, then determine the president’s and the pivot’s expected payoffs, and finally derive the president’s equilibrium nominee and the pivot’s equilibrium delay time.

At the final stage of the game, the president will not veto the agency’s decision if and only if the policy choice of the successfully confirmed political appointee provides the president utility at least as high as what he can obtain from the status quo. Given this and any x_a confirmed in the Senate, this appointee’s equilibrium policy choices are given by

y^{*} = {\begin{array}{l} q & if q is between x_{a} and \\ 2 p - q & if q is between 2 p - x_{a} and p \\ x_{a} & otherwise \end{array}

(3)

With this, the president’s and the Senate pivot’s expected utilities from policy outcomes (i.e. the second term of equation (1)’s right-hand side and the third term of equation (2)’s right-hand side, respectively) for appointing an appointee with x_a can be obtained once we integrate equation (3) over the status quo distribution. In general, the president (the pivot) is better (worse) off with a political appointee whose ideal point is closer.

The president’s and the filibuster pivot’s utility functions imply that the president prefers no delays, while the pivot’s induced ideal delay time is

t^{*} = \frac{| x_{a} - x_{e} |}{2 γ}

(4)

As discussed above, this means that E will delay more when the preference divergence between herself and the nominee is greater and when her ability to obstruct is higher (or, alternatively, when her sensitivity to delay’s costs is lower). One important insight offered by our model, distinct from previous work, is that the president anticipates the pivot’s strategy in equilibrium and may counter these efforts.

Given these results, and after solving via backward induction, we can now characterize the president’s nomination decision and the pivot’s optimal delaying decision in equilibrium.

Proposition 1. Assume that p − x_e < 1/2. Let α ≡ 2γ²/β.

When p − x_e <α, $x_{a}^{*} = p$ and t* = (p − x_e)/2γ;

When $1 - \frac{x_{e}}{2 p - 1} < α \leq p - x_{e}$ , $x_{a}^{*} = (β x_{e} + 2 γ^{2} - 4 γ^{2} p) / (β - 4 γ^{2})$ and t* = γ(1 − 2p + 2x_e)/(β − 4γ²);

When $α \leq 1 - \frac{x_{e}}{2 p - 1}$ , $x_{a}^{*} = β x_{e} / (β - 2 γ^{2})$ and t* = γ x_e/(β − 2γ²).

Proof. See Appendix 1.

Note that α represents the relative value of the pivot’s weighting over the president’s weighting. As such, we can think of it as capturing where a nominee falls in the bureaucratic hierarchy, with larger values of α corresponding to higher nominee levels.

The intuition behind this proposition’s result can be summarized as follows. For high-profile nominations, whose policy consequences have great impact on the president’s success in his office, the president is more willing to incur high delay costs in exchange for nominees with less preference divergence. Moreover, the Senate will suffer more from obstructing high-profile nominees. In equilibrium, for very high-level nominees, the president will appoint someone sharing his preferences in equilibrium, with bearable delay costs. When a position’s importance decreases, the president will trade-off between someone whose preferences are closer to his own with incurring less delay to get his policy programs rolling; to strike a balance, the president will compromise by appointing someone with moderate preference divergence to avoid potentially long delays. For very low-profile nominees, whose policy impact on the success of the president’s program is low and the Senate is rather unconcerned about the political costs of nomination delay, the president will be more willing to defer to the Senate by appointing nominees sympathetic to the Senate’s preferences, which may, in turn, lead to speedier appointment.

3. Comparative statics and hypotheses

Proposition 1 allows us to derive several important comparative statics and related hypotheses. We do so with a series of corollaries, which offers a number of novel predictions, many of which we can examine empirically (we will emphasize predictions regarding duration, for which we have data).¹⁶ To begin, Corollary 1 considers the effects of the appointee’s place in the bureaucratic hierarchy and its joint impact with divided government.

Corollary 1.

When α ≥ (p − x_e), $\partial x_{a}^{*} / \partial β = 0$ , $\partial x_{a}^{*} / \partial γ = 0$ , ∂t*/∂β = 0 and ∂t*/∂γ < 0;

When α < (p − x_e), $\partial x_{a}^{*} / \partial β < 0$ , $\partial x_{a}^{*} / \partial γ > 0$ , ∂t*/∂β < 0 and ∂t*/∂γ > 0.

Corollary 1 generates a key hypothesis about duration and an office’s policy importance. As mentioned, when high-level nominees (e.g. agency heads) are considered, β is low and γ is high, meaning that α is very high. Therefore, in a world with three position levels (such as that which we will examine empirically)—high, middle, and low—when considering positions from middle to top, duration likely falls in the first scenario (i.e. when α ≥ (p − x_e)) and is lower for higher-level nominees, as ∂t*/∂γ < 0.¹⁷ Given the president’s great stake in higher-level nominees and the pivot’s possibility of facing negative public reaction given delay, while the president prefers those with similar or identical preferences in equilibrium, the pivot’s potential audience costs lead her to nonetheless pass the nomination more swiftly relative to lower-level appointments. This is in the same spirit as empirical analyses finding that cabinet members face the least problem of confirmation (Aberbach and Rockman 2009; Krutz et al., 1998).

Conversely, when comparing middle positions to lower appointments, duration likely falls in the second scenario (α < (p − x_e)), so that lower-level nominees receive swifter approval. For such nominees, in equilibrium, the president chooses an appointee whose ideal point lies between him and the pivot, providing one reason why delay should be modest. This tendency for a lower-level nominee to have short duration is reinforced by the small weight, relative to middle-level appointees, that the president will place on policy implications on the one hand, and the credibility that the pivot will have in threatening delay given low audience costs on the other hand. In other words, the president will be more willing to compromise and nominate someone with preferences closer to the pivot for lower-level positions, invoking lesser delays. Thus, as reflected in Hypothesis 1, we should witness a non-monotonic relationship between position level and confirmation time.¹⁸

Hypothesis 1. High-level and low-level positions will be confirmed more quickly than middle-level positions.

Note that this non-monotonic hypothesis differs from the conventional monotonic prediction. This discrepancy results primarily from our accounting for the strategic interaction between the president and the Senate and the interplay between duration and the president’s nominee choices. Also note that Corollary 1 generates an indeterminate prediction for a direct divided government effect, as any impact depends on which scenario divided government falls into, and thus we have no directional prediction without specifying further conditions.

We now turn to how ideological differences between the president and the filibuster pivot effect nominations. Contrary to intuition, greater distance does not always lengthen duration. Rather, as Corollary 2 shows, an ideological difference’s effect depends crucially on the value of the position—where it falls in the hierarchy.

Corollary 2. Denote d = (p − x_e) > 0 (i.e. the ideological distance between the president and the pivot).

When α ≥ (p − x_e), $\partial x_{a}^{*} / \partial p > 0$ , ∂t*/∂d > 0;

When $1 - \frac{x_{e}}{2 p - 1} \leq α < (p - x_{e})$ , $\partial x_{a}^{*} / \partial p < 0$ , $\partial x_{a}^{*} / \partial x_{e} > 0$ , ∂t*/∂d < 0;

When $α < 1 - \frac{x_{e}}{2 p - 1}$ , $\partial x_{a}^{*} / \partial x_{e} > 0$ , ∂t*/∂x_e > 0.¹⁹

The intuition behind Corollary 2 is that the president, given that he greatly values policy impacts associated with high-level positions, will emphasize nominating candidates close to his ideal point. Thus, in turn, as the distance between the president and the pivot increases, that between the pivot and high-level nominees goes up. Conversely, the president more willingly nominates candidates closer to the pivot’s ideal point given delay’s costs and the lower policy consequences associated with middle- and, especially, with low-level positions.²⁰ As a result, greater distances between the president and the pivot will induce the president to offer more moderate nominees, lessening delays. Put differently, since high-level positions will increase α, the first scenario in Corollary 2 is more likely, while the second and third scenarios will more likely capture the situation involving middle- or low-level positions. This gives us Hypothesis 2.²¹

Hypothesis 2. Ideological distance between the president and the pivot has a positive effect on confirmation duration for high-level positions, but a negative impact for middle- or low-level ones.

Note that Hypothesis 2 substantially differs from earlier work in two respects. The first involves the impact of the ideological gap between the president and the filibuster pivot on duration. Rather than assuming that the president always prefers trading away policy gains rather than enduring greater duration, our model suggests that the president’s trade-off between policy gains and duration is conditioned by a position’s hierarchical status and associated policy importance. The second is more subtle and involves the impact of party pressure on E. Instead of assuming that the party median is the pivot, which is equivalent to assuming complete party pressure, or that the filibuster pivot faces no pressure (corresponding to Krehbiel (1998)), our model allows for any level of party influence. Hence, stronger (weaker) partisan pressure shifts E toward (away) from the median of the president’s opposition party.

Finally, we consider the interaction between the pivot’s effectiveness in obstructing a nomination and the ideological difference between the president and the pivot. As Corollary 3 details and, interestingly, in contrast to the non-monotonic relationship of the ideological difference between the president and the pivot on duration, the interaction of this difference with the ability to obstruct is always negative.

Corollary 3. Denote d = (p − x_e) > 0.

When α ≥ (p − x_e), ∂²t*/∂d∂γ < 0;

When $1 - \frac{x_{e}}{2 p - 1} \leq α < (p - x_{e})$ , ∂²t*/∂d∂γ < 0;

When $α < 1 - \frac{x_{e}}{2 p - 1}$ , ∂²t*/∂x_e∂γ > 0.²²

In the first scenario (i.e. α ≥ (p − x_e)), the president will appoint someone very close to his ideal point and a larger distance is associated with greater duration. Furthermore, given the partisan politics assumption that divided government lowers the pivot’s concerns about delay costs, the interaction of divided government and ideological distance further induces duration. Under the last two scenarios (α < (p − x_e)), while the effect of ideological distance on appointee moderation is positive, it is weaker under divided than under unified government. Moreover, as divided government makes the pivot worry less about delay costs, its interaction with ideological distance positively impacts duration. In short, the interaction of ideological difference and the pivot’s concern about the president’s administration performance (or, conversely, her willingness to obstruct) is not conditioned by α. Hence, there is a monotonic interaction effect regardless of appointment level, leading to Hypothesis 3.

Hypothesis 3. The effect of the ideological difference between the pivot and the president is higher under divided government than under unified government.

Interestingly, this hypothesis is the only one in our analysis coinciding with McCarty and Razaghian, albeit with a different rationale. Of course, whether this hypothesis is confirmed after including different variables than they do, and measuring the same variables differently (e.g. with different assumptions about party influence built in), remains to be seen.

To summarize, our model is specific enough to generate distinct hypotheses but is sufficiently flexible to investigate the mechanisms under which party effects might operate. Regarding the former, we generate three hypotheses, two differing from what an intuitive verbal theory might predict, and which have not been proposed. As for party effects, we can distinguish and examine two possible partisan roles—impacting the pivot’s position or changing how policy and delay are weighted, as well as taking divided government out of the black box by detailing how it might impact choices. Thus, in considering Hypotheses 2 and 3, we can see if a model assuming that party pressure sets E at the median of the president’s opposition party beats the pure preference alternative that specifies E as the senator at the percentile dictated by Senate rules at the time (for the specific rules, see Binder and Smith (1997)). Additionally, in analyzing Hypothesis 3 we can determine if the majority party’s claimed institutional prerogatives, or hostility toward the president, impacts obstruction.

4. Empirical analysis

We investigate our hypotheses by combining McCarty and Razaghian’s (1999) 1885–1996 data with preference data, using Poole’s (1998) common space scores, corresponding to our theoretical model. Thus, our dependent variable is the number of days between duration and nomination, excluding recesses exceeding 30 days.²³ With respect to independent variables, we improve on past work by directly measuring the ideological distance between the president and the pivot. Note that common space scores allow us to measure Senate and presidential preferences comparably (and that we cannot use DW-NOMINATE scores because available presidential scores are only comparable to House members, but not to Senators, although common space and DW-NOMINATE scores are heavily correlated). As implied, we adopt two variants of this ideological distance measure, one measuring the distance between the president and the relevant filibuster pivot,²⁴ the other the distance between the president and the relevant party median.²⁵ Based on our theoretical results, and consistent with McCarty and Razaghian (1999), we define three dummy variables to measure the first, second, and third tiers of agency positions, which correspond to high-, middle-, and low-level appointments, respectively.²⁶ We also define a dummy variable for divided government. However, whether we incorporate the relevant filibuster pivot or the party median, the correlation coefficient between divided government and its interaction with ideological distance is.99. As such, our analysis possibly faces multi-collinearity problems. In addition, although our theory says nothing about these variables, we follow McCarty and Razaghian by including dummy variables for whether the nominee had been previously confirmed within the last four years, for whether the nomination took place in the first three months of an administration (the honeymoon period), and for main departments (results are unaltered if we drop these measures).²⁷

Jointly, these data allow us to test our three hypotheses and to investigate mechanisms of party influence. We examine the non-monotonic relationship predicted by Hypothesis 1 by omitting the second-tier dummy variable and seeing if the coefficients for both the first and third tiers are negative. We investigate Hypothesis 2’s prediction that greater ideological distance between the president and the pivot will be positive for high-level positions and negative otherwise by estimating separate models for each appointee tier. Finally, we consider Hypothesis 3’s prediction that ideological distance has a stronger effect under divided government by interacting our distance and divided government measures, with a positive coefficient providing supportive evidence.

To reiterate, we can also distinguish between two party influence mechanisms—the willingness to use the perquisites of majority Senate status to delay confirmation (or sharing the president’s concerns about government performance), and the impacting of the relevant pivot through party pressure. Per the former, if the interaction term for divided government used for Hypothesis 3 operates as expected, we then have evidence that divided government matters. Concerning the latter, as discussed, if the model assuming that E is the party median fits the data and theoretical expectations better than one using a preference-based filibuster pivot, we have support for the contention that party pressure shifts the key veto player’s location for each hierarchical level.

Econometrically, we use hazard models, estimating both parametric Weibull and semiparametric Cox models. As results are substantively identical, we only present the Weibull estimates.²⁸

Tables 1 –3 present these findings. Table 1 examines Hypotheses 1 and 3; Tables 2 and 3 cover Hypothesis 2; and the results in all three Tables, considered together, tease out potential party effects.

Table 1.

Confirmation duration—Hypotheses 1 and 3.

	Preference-based distance measure	Party-based distance measure
Divided government	−1.401^c	−.565^a
	(.274)	(.298)
Ideological distance of president and pivot	−.663^c	1.332^c
	(.201)	(.256)
Ideological distance of president and	3.307^c	1.694^c
pivot*divided government	(.462)	(.435)
Tier one appointment	−1.055^c	−1.056^c
	(.067)	(.067)
Tier three appointment	−.042	−.030
	(.038)	(.038)
Previously confirmed	−.129^a	−.143^a
	(.065)	(.065)
Honeymoon period	−.701^c	−.657^c
	(.043)	(.043)
Agriculture	.097	.002
	(.085)	(.085)
Commerce	.225^c	.197^c
	(.062)	(.062)
Justice	.341^c	.304^c
	(.063)	(.062)
Labor	.349^c	.274^c
	(.079)	(.079)
Infrastructure	.283^c	.232^c
	(.056)	(.057)
Social welfare	.471^c	.419^c
	(.063)	(.064)
Non-departmental agency	.532^c	.483^a
	(.077)	(.077)
Constant	3.595^c	2.355^c
	(.111)	(.182)
Duration dependence^d	1.010	1.011
	(.013)	(.013)
Log-likelihood	−5276.20	−5252.98

Weibull estimates with standard errors in parentheses.

Two-tailed tests:

p < 10, ^b .01 < p < .05, ^c p < .01.

H₀:σ = 1.

Number of observations = 3558.

Table 2.

Confirmation duration, no party influence on veto player—Hypothesis 2.

	Tier one	Tier two	Tier three
Divided government	−.319	−1.629^c	−2.103^c
	(1.009)	(.352)	(.443)
Ideological distance of president and	.619	−1.048^c	−.801^c
pivot	(.671)	(.276)	(.302)
Ideological distance of president and	2.158	3.497^c	4.607^c
pivot*divided government	(1.717)	(.591)	(.749)
Previously confirmed	−.288^a	.023	−.294^c
	(.172)	(.092)	(.104)
Honeymoon period	−1.707^b	−.667^c	−.541^c
	(.153)	(.058)	(.068)
Agriculture	−.085	.124	.337^a
	(.254)	(.102)	(.175)
Commerce	.540^c	.319^c	.090^c
	(.238)	(.085)	(.090)
Justice	1.024^c	.128	.597^c
	(.223)	(.079)	(.110)
Labor	.349c	.346^c	.472^c
	(.267)	(.102)	(.128)
Infrastructure	.512^c	.389^c	.196^c
	(.212)	(.080)	(.080)
Social welfare	.924^c	.451^c	.507^c
	(.235)	(.081)	(.106)
Non-departmental agency	1.828^c	–	.520^c
	(.394)		(.090)
Constant	1.674^c	3.861	3.565^c
	(.380)	(.152)	(.161)
Duration dependence^d	1.024	1.112	.983
	(.045)	(.021)	(.019)
Log-likelihood	−395.11	−2310.60	−2470.52

Weibull estimates with standard errors in parentheses.

Two-tailed tests:

p < .10, ^b.01 < p < .05, ^cp < .01.

H₀:σ = 1.

Number of observations = 269 for tier one, 1661 for tier two, and 1628 for tier three.

Table 3.

Confirmation duration, party influence on veto player—Hypothesis 2.

	Tier one	Tier two	Tier three
Divided government	.102	−.803^b	−1.628^c
	(1.067)	(.390)	(.442)
Ideological distance of president and	1.346^a	1.263^c	.799^c
pivot	(.780)	(.365)	(.384)
Ideological distance of president and	1.466	1.792^c	3.372^c
pivot*divided government	(1.578)	(.562)	(.727)
Previously confirmed	−.261	−.007	−.314^c
	(.170)	(.092)	(.104)
Honeymoon period	−1.675^c	−.611^c	−.501^c
	(.152)	(.057)	(.069)
Agriculture	−.086	.045	.248^a
	(.250)	(.102)	(.177)
Commerce	.582^b	.316^c	.048^c
	(.237)	(.085)	(.091)
Justice	1.001^c	.130^a	.566^c
	(.222)	(.079)	(.110)
Labor	.346^a	.298^c	.380^c
	(.264)	(.102)	(.130)
Infrastructure	.517^c	.371^c	.134^c
	(.210)	(.080)	(.082)
Social welfare	.932^c	.419^c	.458^c
	(.233)	(.081)	(.107)
Non-departmental agency	1.828^c	–	.470^c
	(.395)		(.091)
Constant	1.030^c	2.458	2.646^c
	(.573)	(.263)	(.266)
Duration dependence^d	1.028	1.113	.984
	(.045)	(.021)	(.019)
Log-likelihood	−393.17	−2298.37	−2464.65

Weibull estimates with standard errors in parentheses.

Two-tailed tests:

p < .10, ^b.01 < p < .05, ^cp < .01.

H₀:σ = 1.

Number of observations = 269 for tier one, 1661 for tier two, and 1628 for tier three.

Overall, our hypotheses garner considerable support. Table 1 demonstrates that Hypothesis 1 is largely supported regardless of whether we specify the pivot as the preference-based filibuster pivot or the median of the party opposing the president in the Senate. Durations for both the first- and third-tier nominations are less than for the second-tier nominations, although the p-value of the third-tier coefficient is only 13% and 21% for one-tailed tests with preference-based and party-based distance measures, respectively.

Table 1 also provides strong evidence for Hypothesis 3. This supports the notion that majority parties under divided government, other things being equal, prefer obstructing nominees more as they care less about government performance.

In examining Hypothesis 2, we find more support for the variant of our model assuming that party pressure changes the filibuster pivot’s position only for first-tier appointees. In Table 2, where the filibuster pivot is assumed to not be subject to party pressure, the coefficient for ideological polarization is statistically insignificant while, in Table 3, where party influence is assumed, the coefficient is positive and significant.²⁹ As such, we infer that the partisan assumption better explains first-tier nominations.³⁰ By contrast, the nonpartisan assumption works better for lower-tier nominees, in the sense that it corresponds to our model’s predictions. In Table 2, the coefficients for ideological polarization are statistically significant and negative for the second- and third-tier positions, with or without including control variables.³¹ In Table 3, the coefficients are always positive, i.e. if our theory is right, there is no evidence for partisan ideological polarization effects for either of these lower tiers. These contrasting results between top- and lower-tier appointments are intuitive: Parties presumably concentrate their resources on key appointment battles, i.e. on first-tier nominations.

These results have two important implications. One is that, in contrast to past analysis, first-tier nominations can be explained rather well by including partisan effects. Related to this, in contrast to the previously discussed puzzle that the president always prefers enduring more delays to installing individuals with closer preference alignments to the legislative pivot, in fact the president only makes this trade-off for high-level appointees. For less policy-relevant offices, where legislative obstruction is more feasible, he does not.

In short, our overall results from examining Hypotheses 2 and 3 imply that parties play significant roles in shifting the filibuster pivot to their own medians (because Hypothesis 3 is supported, regardless of the assumption about party pressure) for high-level nominees. However, there is relatively more evidence that party members can vote their nonpartisan preferences in assessing lower-tier nominees. Moreover, the importance of hierarchical structure in shaping nomination duration highlighted in our model is corroborated.

As for the control variables, they tend to be statistically significant, although not always, and the effects are not necessarily directionally consistent. Divided government tends to have a positive, significant, impact when the ideological distance between the president and the pivot is large but has a negative effect when distance is small or moderate.³² Those previously confirmed take longer, and appointments in the honeymoon period are quicker. Nominees for the Treasury Department tend to be approved faster than the norm, while those for the Social welfare and non-departmental agencies typically require more time.

5. Conclusions

Recent years have seen heightened sensitivity to confirmation dynamics and the ability of such strategizing to not only influence who gets nominated but to condition when appointments actually go through. However, we have lacked a formal model carefully elaborating this process, so theory and data have rarely informed one another. We offer a simple but powerful theory that fills the gap between previous theoretical and empirical works by jointly considering duration and nominee selection, incorporating possible mechanisms of partisan impacts, and generating testable, often non-intuitive, hypotheses. Stated differently, our model demonstrates how, once we take into account strategic behavior between the executive and pivots and the impacts of hierarchy and party, the appointment process is more subtle and nuanced than we might initially believe. Furthermore, our empirical analysis shows that the data conform to the implications of the executive appointment model rather well.

To recap, our theory focuses on the importance of the position under consideration, the weight that the Senate puts on delay’s costs relative to the policy influence of nominees, and the trade-offs made by strategic actors with respect to ideological compatibility and the costs of time. All of these features seem to matter, often in subtle ways not previously appreciated, which depend on their interaction with one another.

Our empirical results, based principally on preference measures that correspond to our theory and which are now widely available, differ from past findings and show that the appointment process is more nuanced than much political science research or popular commentary suggests. For example, consistent with our theoretical rationale, we provide empirical evidence that presidents will sometimes accept greater duration in order to procure a more ideologically compatible nomination. We also follow our theoretical results by highlighting how features such as the policy value of an office and ideological distance between the president and Senate opposition have a non-monotonic relation to duration. More generally, we show that appointee selection, divided government, polarization, and the policy value of an appointment fit together in complicated ways. In short, despite the seeming prima facie simplicity of the appointment dilemma, the empirical analysis of appointments actually requires a theoretically sophisticated underpinning. Interestingly, from a policy perspective, this approach shows that, net of mandating short durations directly, there is no easy way to engineer quick action on confirmations despite much and continued popular bemoaning of the negative costs associated with delay—a reform that would decrease duration under one set of conditions will increase it under another.

In the same vein, besides lending credence to many of the predictions associated with our comparative statics, our theoretical and empirical analyses shed light on the role that political parties might play. Specifying theoretically alternative roles for parties and examining them empirically produces evidence that parties matter, although not unambiguously. For one thing, nominations under divided governments face more obstruction than would be predicted by a pure preference model—to reiterate, we provide a theoretical rationale, largely supported empirically, for divided government to matter having to do with caring about government performance. For another, pressure effectively shifts the ideological location of the relevant pivot for first-tier nominations about which parties will be largely focused on; by contrast, members appear able to vote as they choose for appointments further down the bureaucratic pecking order. Thus, in terms of the well-known parties versus preferences debate, our analysis reinforces the view that a sophisticated perspective is required, where how party influence is likely to operate is carefully specified and then whether there is confirmatory evidence is empirically assessed.

Obviously, we do not believe that our analysis will be the final word on appointments. In particular, it would be illuminating, as we make progress in measuring appointee ideology in a common space with presidents and legislatures (e.g. Clinton et al., 2012), to investigate whether the hypotheses stemming from our executive appointment model for the equilibrium ideologies of nominees are actually borne out. Only then could we be said to have fully integrated duration and ideology.

Footnotes

Appendix 1

We first solve for the agency’s policy outcomes for all possible status quos, derive the president’s (P) expected utility, and then plug equation (4) into it

E U_{p} (x_{a} | p) = {\begin{array}{l} - \frac{β {(x_{a} - x_{e})}^{2}}{4 γ^{2}} - [(x_{a} - p) - {(p - x_{a})}^{2}] if p < x_{a} < 1 (5) \\ - \frac{β {(x_{a} - x_{e})}^{2}}{4 γ^{2}} - [(p - x_{a}) - {(p - x_{a})}^{2}] if 2 p - 1 < x_{a} < p (6) \\ - \frac{β {(x_{a} - x_{e})}^{2}}{4 γ^{2}} - \frac{1}{2} [p^{2} - x_{a}^{2} + {(1 - p)}^{2}] if 0 < x_{a} < 2 p - 1 (7) \end{array}

The first result that we obtain is that it is not optimal for the president to propose a nominee with x_a > p or with x_a < x_e. As equation (5) shows, for x_a > p, the president is strictly better off nominating someone closer to him, given that p > 1/2.³³ The first derivative with respect to x_a in EU_p(x_a|p) of equations (6) and (7) is positive for any x_a < x_e, implying that the president prefers x_a = x_e to any x_a < x_e. From now on, we focus attention on the set of [x_e, p] for the president’s optimal nominees.

To solve the president’s optimal choice of nominee, x_a, we first derive the first and second derivatives with respect to the president’s choice x_a

(8)

\frac{\partial E U_{p}}{\partial x_{a}} = {\begin{array}{l} {(x_{e} - x_{a}) + α [1 + 2 (x_{a} - p)]} / α & if 2 p - 1 < x_{a} < p \\ {(x_{e} - x_{a}) + α x_{a}} / α & if 0 < x_{a} < 2 p - 1 \end{array}

and

(9)

\frac{\partial^{2} E U_{p}}{\partial x_{a}^{2}} = {\begin{array}{l} [- 1 + 2 α] / α & if 2 p - 1 < x_{a} < p \\ [- 1 + α] / α & if 0 < x_{a} < 2 p - 1 \end{array}

where α = 2γ²/β. Since the first-order conditions are complex and the expected utility functions for the cases of 2p −1 < x_a < p and 0 < x_a < 2p − 1 are different, we cannot simply obtain global maximizers with the first- and second-order conditions for only one of the cases. We then consider the sign for different cutoff points.

(10)

When x_{a} = p, \partial E U_{p} / \partial x_{a} > 0 iff α > (p - x_{e})

(11)

When x_{a} = 2 p - 1, \partial E U_{p} / \partial x_{a} > 0 iff α > 1 - x_{e} / (2 p - 1)

(12)

When x_{a} = x_{e} > 2 p - 1, \partial E U_{p} / \partial x_{a} > 0 iff (p - x_{e}) < 1 / 2

(13)

When x_{a} = x_{e} < 2 p - 1, \partial E U_{p} / \partial x_{a} > 0

Note that (p − x_e) > 1 − x_e/(2p − 1)1 − x_e/(2p − 1) iff (p − x_e) < 1/2. We assume that (p − x_e) < 1/2, but the result is qualitatively the same if we assume the opposite.³⁴ As the location of an optimum depends on the level of α, there are three cases to discuss.

Case 1. (p − x_e) < α

$\forall x_{a} \in [2 p - 1, p]$ , the first derivative of EU_p is strictly positive, given equations (9) to (11). Therefore, $\underset{x_{a} \in [2 p - 1, p]}{arg max} E U_{p} = p$ . If x_e > 2p − 1, $\underset{x_{a} \in [x_{e}, p]}{arg max} E U_{p} = p$ . If x_e ≤ 2p − 1, we need to examine EU_p for x_a ∈ [x_e, 2p − 1]. Given α > 1 − x_e/(2p − 1) and equation (8), the first derivative of EU_p is strictly positive for x_a ∈ [x_e, 2p − 1] and $\underset{x_{a} \in [x_{e}, 2 p - 1]}{arg max} E U_{p} = 2 p - 1$ . Therefore, $x_{a}^{*} = p$ and t* = (p − x_e)/2γ.

Case 2. 1 − x_e/(2p − 1) < α < (p − x_e)

$\forall x_{a} \in [x_{e}, 2 p - 1]$ , the same result as in the previous case holds. $\forall x_{a} \in [2 p - 1, p]$ , since p − x_e < 1/2, α < 1/2 and the second derivative of EU_p is strictly negative. This implies that EU_p is strictly concave on this range of and that EU_p is strictly increasing as x_a increases from x_e to 2p − 1 but deceases as x_a approaches p. Thus, $x_{a}^{*} = (β x_{e} + 2 γ^{2} - 4 γ^{2} p) / (β - 4 γ^{2}) \in [2 p - 1, p]$ . Plugging this into equation (3) obtains t* = γ(1 − 2p + 2x_e)/(2β − 4γ²).

Case 3. 0 < α < 1 − x_e/(2p − 1)

Since α is positive, 1 − x_e/(2p − 1) has to be positive, which implies x_e < (2p − 1). Equations (9) to (11) imply that $\underset{x_{a} \in [2 p - 1, p]}{arg max} E U_{p} = 2 p - 1$ . $\forall x_{a} \in [x_{e}, 2 p - 1]$ , EU_p is strictly concave from Equations (9), (11), and (13). We then obtain $x_{a}^{*} = β x_{e} / (β - 2 γ^{2}) \in [x_{e}, 2 p - 1]$ , and therefore $\underset{x_{a} \in [x_{e}, p]}{arg max} E U_{p} = β x_{e} / (β - 2 γ^{2})$ , since $\underset{x_{a} \in [2 p - 1, p]}{arg max} E U_{p} = 2 p - 1$ .

This implies that t* = γx_e/(β − 2γ²)

Appendix 2

In the text we assume that p − x_e < 1/2. We now show that our theoretical results are quite similar even if we relax this assumption and assume that p − x_e > 1/2.

Proposition A1. Assume that p − x_e > 1/2. Let α ≡ 2γ²/β.

where $α^{*} \in [p - x_{e}, 1 - \frac{x_{e}}{2 p - 1}]$ and is defined below.

Proof. When p − x_e > 1/2, $p - x_{e} < 1 - \frac{x_{e}}{2 p - 1}$ .

The proof for $α > 1 - \frac{x_{e}}{2 p - 1}$ and α < p − x_e is very similar to that for Cases 1 and 3 in Proposition 1. Therefore, we focus our attention on the case where $p - x_{e} < α < 1 - \frac{x_{e}}{2 p - 1}$ .

When $\forall x_{a} \in [2 p - 1, p]$ , EU_p is strictly convex, since p − x_e < α (from equation (9)). The local maximizers for this range are either 2p − 1 or p, so the global maximizer depends on EU_p for x_a ∈ [x_e, 2p − 1]. For this range, we know from equations (11) and (12) that EU_p is strictly concave, so $\underset{x_{a} \in [x_{e}, 2 p - 1]}{arg max} E U_{p} = \frac{x_{e}}{1 - α}$ . Therefore, the global maximizer is either $\frac{x_{e}}{1 - α}$ or p. Define $α^{*} = {α \in [p - x_{e}, 1 - \frac{x_{e}}{2 p - 1}] | E U_{p} (x_{a} = \frac{x_{e}}{1 - α}) = E U_{p} (x_{a} = p)}$ . It can be shown that α* always exists and that $E U_{p} (x_{a} = \frac{x_{e}}{1 - α}) < E U_{p} (x_{a} = p)$ iff α > α*. Our results are, therefore, obtained.

Acknowledgements

This paper was previously presented at the 2008 Annual Meeting of the Midwest Political Science Association, Chicago, IL. We thank Nolan McCarty for providing data; all errors remain the responsibility of the authors.

Funding

Chiou’s international travel for this research has been generously supported by receipt of the Wu Da-You Research Award.

Notes

References

Aberbach

Rockman

(2009) The appointments process and the administrative presidency. Legislative Studies Quarterly 39: 38–59.

Asmussen

(2011) Female and minority judicial nominees: president’s delight and senators’ dismay? Legislative Studies Quarterly 36: 591–619.

Bertelli

Feldmann

(2006) Strategic appointments. Journal of Public Administration Research and Theory 17: 19–38.

Binder

Maltzman

(2002) Senatorial delays in confirming federal judges, 1947–1998. American Journal of Political Science 46: 190–199.

Binder

Smith

(1997) Politics or Principle? Filibustering in the United States Senate. Washington (DC): Brookings.

Bond

Fleisher

Krutz

(2009) Malign neglect: evidence that delay has become the primary method of defeating presidential appointments. Congress and the Presidency 36: 226–243.

Brookings Institution (2002) Presidential appointee initiative urges Senate to end the confirmation drought. News release, 25 July 2002.

Chiou

F-Y

Rothenberg

(2006) Preferences, parties, and legislative productivity. American Politics Research 34: 705–731.

Chiou

F-Y

Rothenberg

(2009) A unified theory of US lawmaking: preferences, institutions, and party discipline. Journal of Politics 71: 1257–1272.

10.

Clinton

Bertelli

Clinton

. (2012) Separated powers in the United States: the ideology of federal executives and their agencies. American Journal of Political Science 56: 341–354.

11.

Console-Battilana

Shepsle

(2009) Nominations for sale. Journal of Theoretical Politics 21: 413–449.

12.

Derouen

Jr Peake

Ward

(2005) Presidential mandates and the dynamics of Senate advice and consent, 1885–1996. American Politics Research 33: 106–131.

13.

Epstein

O’Halloran

(1999) Delegating Powers: A Transaction Cost Politics Approach to Policy Making under Separate Powers. New York: Cambridge University Press.

14.

Epstein

Segal

(2007) Advice and Consent: The Politics of Judicial Appointments. New York: Oxford University Press.

15.

Gerhardt

(2000) The Federal Appointments Process: A Constitutional and Historical Analysis. Durham (NC): Duke University Press.

16.

Hammond

Hill

(1993) Deference or preference? Explaining Senate confirmation of presidential nominees to administrative agencies. Journal of Theoretical Politics 5: 23–59.

17.

(2007) Congressional agency control: the impact of statutory partisan requirements on regulation. Unpublished manuscript. Stanford (CA): Stanford University.

18.

Huber

Shipan

(2002) Deliberate Discretion? The Institutional Foundations of Bureaucratic Autonomy. New York: Cambridge University Press.

19.

Rothenberg

(2012) Rational incompetence. Journal of Theoretical Politics 24: 13–18.

20.

Jones

McDermott

(2005) Americans, Congress, and Democratic Responsiveness: Public Evaluations of Congress and Electoral Consequences. Ann Arbor: University of Michigan Press.

21.

Koger

(2006) Cloture reform and party government in the Senate, 1918–1925. Journal of Politics 68: 708–719.

22.

Koger

(2010) Filibustering: A Political History of Obstruction in the House and Senate. Chicago: University of Chicago Press.

23.

Krehbiel

(1991) Information and Legislative Organization. Ann Arbor: University of Michigan Press.

24.

Krehbiel

(1998) Pivotal Politics: A Theory of US Lawmaking. Chicago: University of Chicago.

25.

Krutz

Bond

Fleisher

(1998) From Abe Fortas to Zoe Baird: why some presidential nominations fail in the Senate. American Political Science Review 92: 871–882.

26.

Lewis

(2008) The Politics of Presidential Appointments: Political Control and Bureaucratic Performance. Princeton: Princeton University Press.

27.

Lewis

(2011) Presidential appointments and personnel. Annual Review of Political Science 14: 47–66.

28.

Mayhew

(2003) Supermajority rule in the Senate. PS 36: 31–36.

29.

McCarty

(2004) The appointments dilemma. American Journal of Political Science 48: 413–428.

30.

McCarty

Razaghian

(1999) Advice and consent: Senate response to executive branch nominations. American Journal of Political Science 41: 1122–1143.

31.

McCarty

Poole

Rosenthal

(2001) The hunt for party discipline in Congress. American Political Science Review 95: 673–687.

32.

McCubbins

Noll

Weingast

(1989) Structure and process, politics and policy: administrative procedures as instruments of policy control. Virginia Law Review 75: 431–482.

33.

Moe

(1989) The politics of bureaucratic structure. In: Chubb

Peterson

(eds) Can the Government Govern? Washington (DC): Brookings.

34.

Nixon

(2001) Appointment delays for vacancies on the Federal Communications Commission. Public Administration Review 61: 483–492.

35.

Nixon

(2004) Separation of powers and appointee ideology. Journal of Law, Economics, and Organization 20: 438–457.

36.

Nixon

Bentley

(2006) Appointment delay and the policy environment of the National Transportation Safety Board. Administration & Society 37: 679–694.

37.

Nokken

Sala

(2000) Confirmation dynamics: A model of presidential appointments to independent agencies. Journal of Theoretical Politics 12: 91–112.

38.

O’Connell

(2010) Waiting for Leadership: President Obama’s Record in Staffing Key Agency Positions and how to Improve the Appointments Process. Washington (DC): Center for American Progress.

39.

Ornstein

(2009) Slow confirmation process is hurting US government. Roll Call, 24 June.

40.

Poole

(1998) Recovering a basic space from a set of issue scales. American Journal of Political Science 42: 954–993.

41.

Rohde

Shepsle

(2007) Advising and consenting in the 60-vote Senate: strategic appointments to the Supreme Court. Journal of Politics 69: 664–677.

42.

Shipan

Shannon

(2003) Delaying justice(s): a duration analysis of Supreme Court confirmations. American Journal of Political Science 47: 645–668.

43.

Snyder

Jr Groseclose

(2000) Estimating party influence in congressional roll-call voting. American Journal of Political Science 44: 193–211.

44.

Snyder

Weingast

(2000) The American system of shared powers: the President, Congress, and the NLRB. Journal of Law, Economics, and Organization 16: 269–305.

45.

Wawro

Schickler

(2006) Filibuster: Obstruction and Lawmaking in the US Senate. Princeton: Princeton University Press.

46.

Wood

Marchbanks MP

III

(2008) What determines how long political appointees serve? Journal of Public Administration Research and Theory 18: 375–396.