Post on 30-Jul-2020
Presidential Politics and Nominations to the FederalJudiciary∗
Justin Pottle†Harvard University
Jon C. Rogowski‡Harvard University
May 13, 2019
Abstract
Few powers confer presidents with more enduring sources of in�uence than the power tonominate individuals to the federal judiciary. While existing research studies the Senate’scapacity to frustrate presidential nominations through delay and other obstruction tactics,relatively little is known about the nature and timing of presidential nomination strategies.We argue that the informational costs of vetting potential nominees lead presidents to prior-itize �lling vacancies in politically favorable jurisdictions. Studying all vacancies in federaldistrict courts from 1961 to 2018, we �nd that presidents announce nominations to vacantjudgeships at systematically higher rates in districts that provided greater electoral support.These patterns are strongest during divided government and are robust across additionalanalyses. Additional evidence illustrates how presidential nomination strategies can haveimportant distributional consequences for the courts’ institutional capacity across districts.Our results suggest mechanisms through which strategic presidential nominations are re-sponsive to local political context.
Word count: 9,727
∗We thank Bethany Blackstone, Albert Rivero, and Greg Wawro for helpful comments and suggestions. An earlierversion of this manuscript was presented at the 2019 annual meeting of the Mid-West Political Science Association.
†Ph.D candidate, Department of Government, 1737 Cambridge Street, Cambridge, MA 02138; jpottle@g.harvard.edu.
‡Assistant Professor, Department of Government, 1737 Cambridge Street, Cambridge, MA 02138; rogowski@fas.harvard.edu.
Few opportunities provide presidents with more important or enduring sources of in�uence
than the ability to nominate judges to the federal judiciary. By selecting like-minded judges and
justices to serve lifetime appointments, presidents can shape the opinions issued by the federal
judiciary for decades after they have left o�ce. In recent years, however, the number of vacan-
cies has increased, and the overall pace of president’s nominations and Senate con�rmations has
slowed.1 These prolonged vacancies have generated complaints from presidential administra-
tions, anxiety from interest groups, and occasionally bruising political battles in the Senate. The
current state of judicial nominations is, in one legal scholar’s words, “chaotic, divisive, arbitrary,
dishonest, insulting, polarizing, and damaging to the public’s con�dence in both the Senate and
the judiciary” (Stone 2011, 466).
In this paper, we study the politics of presidential nomination strategies. Though presiden-
tial delay in submitting judicial nominees to the Senate accounts for a substantial component
of vacancy length (Hollibaugh 2015; Massie, Hansford, and Songer 2004), virtually all scholar-
ship in this area focuses on the causes and consequences of con�rmation delay in the Senate
(e.g., Basinger and Mak 2010; Bell 2002; Binder and Maltzman 2002; Cameron, Kastellec, and Park
2013; Goldman 1998; Hendershot 2010; Martinek, Kemper, and Winkle 2002; Nixon and Goss
2001; Scherer, Bartels, and Steigerwalt 2008; Shipan and Shannon 2003; Stratmann and Garner
2004). While an emergent theoretical scholarship considers the strategic foundations of the tim-
ing of presidential nominations (e.g., Hollibaugh 2015; Jo 2017; Jo, Primo, and Sekiya 2017), most
empirical literature in this area focuses on presidents’ public appeals on behalf of their nomi-
nees (Cameron and Park 2011; Holmes 2007) or the nomination of female and non-white judges
(Asmussen 2011; Killian 2008). While the studies most related to our own identify how the pres-
ident’s relationship vis-à-vis the Senate a�ects their nomination strategy (Massie, Hansford, and1See, e.g., “The Growing Specter of Vacant Federal Judgeships,” Russell Wheeler, April 12, 2016, The Brook-
ings Institution (available at https://www.brookings.edu/blog/�xgov/2016/04/12/the-growing-specter-of-vacant-federal-judgeships/; accessed March 16, 2019); “Federal Judicial Vacancies: The Trial Courts,” Alicia Bannon,July 2, 2013, Brennan Center for Justice (available at http://www.brennancenter.org/sites/default/�les/publications/JudicialVacanciesReportFinal.pdf; accessed March 18, 2019).
1
Songer 2004; Shipan, Allen, and Bargen 2014), this research does not account for how presidents
prioritize nominations across vacancies during their administrations.
Moreover, previous studies neglect the signi�cant costs associated with judicial nominations.
Recent presidents have faced upwards of 50 new vacancies each year. Filling these seats requires
the White House and the Department of Justice direct time and e�ort away from other projects
and toward collecting information about viable candidates. We argue that the informational costs
of vetting lead presidents to prioritize �lling vacancies in politically favorable jurisdictions. The
costs borne by presidential administrations create opportunities for organized interests to help
subsidize the White House’s search process. Building upon scholarship that identi�es presidential
in�uence over government outputs (e.g., Berry, Burden, and Howell 2010; Kriner and Reeves 2015;
Rogowski 2016), we hypothesize that presidents will prioritize nominations to districts where
politically aligned individuals and groups can most easily assist in the vetting process.
We use data on vacancies and nominations in the federal district courts from 1961 to 2018
to establish three key �ndings. First, we show that the overall length of judicial vacancies pri-
marily re�ects the time presidents spend selecting nominees – not Senate delay in con�rming
them. Second, consistent with our argument, we �nd that presidents �ll judicial vacancies at
systematically higher rates in districts that provided greater electoral support. These patterns
are strongest during periods of divided government and are robust to a wide range of model
speci�cations, characterizations of key variables, and subsets of observations. These �ndings run
counter to theories that predict presidents strategically prioritize nominations to ideologically-
hostile courts. Third, we evaluate the implications of these �ndings for the composition and
performance of the federal judiciary. These analyses demonstrate that presidential delay is not
associated with higher quality nominees and that president’s prioritization of politically friendly
districts may indirectly contribute to disproportionate judicial burden in hostile ones. Our results
have important implications for theories of strategic presidential nominations, suggest that local
parties and interests play a larger role in federal judicial nominations than previously understood,
2
and illustrate a mechanism through which political actors can a�ect the institutional capacity of
adjoining branches of government.
The Politics of Judicial Nominations
Judicial nominations are, arguably, as politicized as they ever have been. In an era where
divided government is common and partisan con�ict within Congress is rampant, signature leg-
islative achievements have been elusive for contemporary presidents. While recent presidents
have not passed up the opportunity to convey ambitious legislative agendas, judicial nomina-
tions have become a particular priority.2 The U.S. Senate has responded in turn, scrutinizing the
president’s judicial nominees through increasingly vigorous con�rmation hearings and the fre-
quent use of obstruction and other delay tactics. Contemporary judicial nominations, therefore,
re�ect the same partisan and ideological disagreements that characterize debates over the major
policy issues of the day (Cameron, Kastellec, and Park 2013).
Though Article II powers provide presidents with the power to nominate individuals to po-
sitions within the judiciary and the executive branch, the Senate’s authority to provide advice
and consent requires interbranch agreement for successful nominations. The Senate’s role as a
potential veto point therefore may serve to constrain presidents from nominating individuals to
serve in judicial positions who are unable to withstand Senate scrutiny. Because relatively few
nominees to the judiciary are unsuccessful, much of the literature studies the predictors of Senate
delay in con�rming judges and justices. This literature focuses on how Senate delay re�ects the
level of partisan and ideological con�ict between the president and the Senate (e.g., Bell 2002;
Binder and Maltzman 2002; Shipan and Shannon 2003), the Senate Judiciary Committee (Bell2For instance, prior to his inauguration, aides to President Trump said that “remaking the federal judiciary overall
has been a priority of [Trump’s] and of Vice President-elect Mike Pence.” See “Trump to inherit more than 100court vacancies, plans to reshape judiciary,” Phillip Rucker and Robert Barnes, December 25, 2016, Washington Post(available at https://www.washingtonpost.com/politics/trump-to-inherit-more-than-100-court-vacancies-plans-to-reshape-judiciary/2016/12/25/d190dd18-c928-11e6-85b5-76616a33048d_story.html?utm_term=.c922b0c1d89a;accessed March 19, 2019.
3
2002; Binder and Maltzman 2004; Martinek, Kemper, and Winkle 2002; Stratmann and Garner
2004), and, in the context of lower court nominations, the partisanship of home-state Senators
(e.g., Binder and Maltzman 2004). Other research studies how Senate delay responds to the mo-
bilization of interest groups (Scherer, Bartels, and Steigerwalt 2008; Steigerwalt 2010) and varies
across the years of presidential terms (Bell 2002; Binder and Maltzman 2002; Martinek, Kemper,
and Winkle 2002) and nominees’ demographic and professional characteristics (Bell 2002; Binder
and Maltzman 2002; Martinek, Kemper, and Winkle 2002; Shipan and Shannon 2003).
Despite the shared authority of the president and the Senate in judicial nominations, exist-
ing scholarship dedicates substantially less attention to presidents’ nomination strategies. A few
empirical studies examine when presidents nominate non-white judges to the bench (e.g., Killian
2008) and “go public” in support of their nominees (Cameron and Park 2011; Holmes 2007), while
recent theoretical scholarship identi�es the conditions under which presidents strategically nom-
inate judges who move the court median away from the president’s preferred ideological location
(Jo, Primo, and Sekiya 2017). Studies on presidential delay, meanwhile, predict that presidents de-
lay nominations when the vetting process reveals a potential nominee is insu�ciently quali�ed
relative to the pool of alternatives (Hollibaugh 2015), the ideological distance between the presi-
dent and home-state Senators (Massie, Hansford, and Songer 2004) or the majority party median
(Shipan, Allen, and Bargen 2014) is high, and when neither home-state Senator belongs to the
president’s party (Massie, Hansford, and Songer 2004).
While scholarship on presidential nomination tactics and senatorial response has made im-
portant contributions to understanding the politics of judicial nominations, we know little about
how presidents prioritize among the vacancies that are available for them to �ll. The lack of focus
on the president’s nomination strategy is surprising given the president’s role as agenda setter in
this process and the attendant political advantages that often accompany political agenda-setters
(e.g., Romer and Rosenthal 1978). Moreover, according to Shipan and Shannon (2003, 666), ex-
clusive focus on the Senate’s response to presidential nominations does not account for “the fact
4
that the president’s choice of a nominee is a strategic, political decision, one that is shaped by
institutional con�gurations that may a�ect not only the nomination but also the con�rmation.”
We address this lacuna by studying why presidents �ll some vacancies more quickly than others
and identifying the consequences of these nomination strategies.
How Presidents Prioritize Nominations to the Judiciary
Presidential capacity is fundamentally constrained. Alongside making judicial nominations,
presidents manage the personnel and activities within the White House and executive branch,
engage in diplomatic relations with foreign heads of state, construct legislative agendas and ne-
gotiate with Congress, interact with organized interests, coordinate with their political party, and
receive brie�ngs from intelligence services and on military operations. Even with their signi�cant
institutional resources, presidents cannot simultaneously prioritize all of their responsibilities.
These capacity constraints force presidents to prioritize some responsibilities over others. In-
coming presidents inherit dozens of vacancies in the federal judiciary. Dozens more arise each
year. Despite increases in White House capacity to assemble lists of potential nominees and thor-
oughly vet the names on them, presidents are not equally likely to have a nominee at the ready
for every vacancy that arises. Nonetheless, given the importance of the judiciary for adjudicating
contemporary political disputes, presidents have a strong incentive to use their nomination pow-
ers to tilt the courts in favor of their policy interests. Presidential nomination strategies can be
viewed as “an exercise of policymaking furthering a presidential agenda” (Goldman 1997, 2). Fill-
ing the federal bench is at once a great source of presidential power and a weighty administrative
task.
However, presidents submit nominations more quickly for some seats than for others. How
do presidents prioritize among them? Identifying systematic sources of variation is important
for understanding how and why presidents contribute to vacancy length in the judiciary. We
argue that the time and energy needed to vet candidates combined with White House capacity
5
constraints create incentives for presidents to more quickly nominate individuals from politically
aligned jurisdictions. This relationship, we argue, results from information de�cits in the White
House about whom to nominate. Our argument borrows from theories of lobbying as an infor-
mation subsidy (Hall and Deardor� 2006). Because of their limited capacity, presidents will look
to maximize their in�uence on the federal judiciary by minimizing the costs of identifying viable
nominees. Two key groups can help subsidize the costs of the vetting process by providing pres-
idents and their administrations with information about quali�ed local candidates and their like-
lihood of con�rmation: local organized interests (Hollis-Brusky 2015; Scherer 2005; Steigerwalt
2010) and partisan allies in Congress (Binder and Maltzman 2004; Massie, Hansford, and Songer
2004). Activists and interest groups tapped into local legal networks may assist presidents in iden-
tifying quali�ed, ideologically sympathetic, and politically acceptable nominees for judgeships in
their district. Similarly, the president’s congressional copartisans may recommend individuals for
open seats. Norms of senatorial courtesy, in which the Senate defers to the home-state senator
of the president’s party on con�rming district court nominees, may also speed up nominations.
We theorize that districts politically aligned with the president will provide greater information
subsidies. Friendly districts more likely to elect presidential copartisans and be home to larger,
better organized networks of likeminded lawyers and interest groups than hostile ones. Addi-
tionally, presidents and their administrations may place greater trust in information that comes
from groups in politically aligned places. Capacity subsidies may also work in the negative by
reducing transaction costs. Presidents need not spend as much time strategizing nominations
in districts with weak opposition from local groups or representatives as in hostile ones where
organized resistance is stronger.
Our argument builds upon recent research that emphasizes the president’s role as party leader
(e.g., Galvin 2009) and disproportionate responsiveness to partisan incentives (Berry, Burden, and
Howell 2010; Kriner and Reeves 2015; Rogowski 2016; Wood 2009). While previous scholarship
has studied these relationships in the context of the president’s interactions with Congress and
6
bureaucracy, to our knowledge presidential prioritization of copartisan constituencies has not
been studied in the context of the judiciary. We emphasize, however, that our argument impli-
cates di�erent mechanisms than previous literature on copartisan favoritism. In particular, we
do not argue that presidents strategically prioritize judicial nominations for electoral purposes,
as they might for the distribution of federal spending (e.g., Berry, Burden, and Howell 2010) or
bureaucratic resources (e.g., Gordon 2009). Instead, our argument emphasizes how information
supplied by reliable partisan allies and weak organized local resistance enable a constrained White
House to maximize its impact on the judiciary.
Our argument makes several new contributions to existing perspectives on the politics of
presidential nominations. First, while existing scholarship focuses largely on how presidents an-
ticipate opposition from key members of the Senate, we focus on how information de�cits and
capacity constraints within the White House shape presidents’ nomination strategies. Second,
we call attention to the role that organized interests may play in judicial nominations. While re-
cent scholarship has begun to focus on executive branch lobbying (e.g., You 2017), to our knowl-
edge this phenomenon has not been studied in the context of the courts or nominations. Third,
our argument contrasts somewhat with theoretical perspectives that posit that presidents make
nominations to reshape the ideological composition of the judiciary (Krehbiel 2007; Moraski and
Shipan 1999). While we do not deny that presidents are inclined to nominate like-minded judges
and justices, our argument predicts that presidents will take the path of least resistance and pri-
oritize nominations to politically aligned districts rather than those aligned against them, where
their ideological or political impact will be arguably greater. Fourth, and �nally, our argument
implies that presidential nomination strategies have distributional consequences with respect to
judicial capacity. To the extent that greater of number vacancies weaken the institutional capac-
ity of the judiciary, our argument suggests that judicial capacity is disproportionately weakened
in jurisdictions less politically aligned with the president. This implication highlights a potential
unintended consequence of the politics of modern judicial nominations.
7
Data and Empirical Strategy
We test our argument using data characterizing all federal district court vacancies and subse-
quent nominations made between 1961 and 2018.3 Using data from the Federal Judiciary Center
on Article III judgeships, we calculate the days that elapsed between the beginning of a vacancy,
which commences when a sitting judge takes senior status, retires, dies, or is removed, and when
the president nominates the next judge for that post. Overall, 2,063 vacancies opened during the
time period under study.4 Figure 1 displays these data, where the height of each bar indicates the
total number of vacancies present in each Congress. The shaded regions at the bottom on each
bar describe the number of vacancies that were inherited from a previous congress. In the 1960s,
vacancies averaged 73 per congress, and just 39 when excluding newly-created seats. However,
since then each congress has seen an average of 120 vacancies, driven in part by the creation of
411 new judgeships between 1961 and 1990. Just over half of all vacancies took multiple con-
gresses to �ll, including 70 percent of newly created seats. The number of vacant seats inherited
by an incoming president peaked in 1992-1993, when President George H.W. Bush left o�ce with
roughly a third of the 193 vacancies open during the 102nd Congress un�lled.
Un�lled vacancies, and the missed nomination opportunities they subsequently represent,
re�ect presidents’ limited capacity to provide the time and energy needed to identify and vet
hundreds of nominees. While much has been made of senatorial footdragging in the con�rma-
tion process, our data show that presidential nomination decisions comprise the lions’ share of
seat vacancy time. The left plot in Figure 2 displays the average number of days that elapsed
between a vacancy and the president’s nomination (Nomination) and between a president’s nom-
ination and the Senate’s con�rmation vote (Con�rmation). While the plot shows that both time-
to-nomination and time-to-con�rmation have increased in recent decades, presidential delay3All summary statistics are shown in Appendix A.4About one-�fth of these vacancies originated from new judgeships that were created during this time period.
We treat the creation of these judgeships as vacancies beginning on the date when the statute was signed by thepresident. However, our results are robust to the exclusion of these newly-created positions; see Appendix B.
8
Figure 1: District Court Vacancies, 1961-2018
Note: Data show the number of vacancies in federal judgeships during each Congress. While vacancies carrying overfrom previous congresses increase most drastically following large expansions of the courts, un�lled existing seatsaccount for a small yet growing proportion of vacant positions encountered in a given Congress.
nonetheless accounts for the large majority of vacancy length. Moreover, presidents have taken
steadily longer to make nominations since the 1960s. Prior to the Reagan administration, presi-
dents made nominations within an average of about eight months from the time a seat became va-
cant. Nomination delay peaked amid the Clinton impeachment crisis during the 105th Congress,
where vacancies sat without nominees for an average of 462 days. While nomination times dipped
slightly during George W. Bush’s �rst term, average time to nomination again exceeded a year
by 2012. Across the time period under study, the average seat remained vacant for 287 days.
Con�rmation time has similarly grown, increasing nearly eight times over during the same time
period, from roughly 29 days during the 87th Congress to 263 days during the 114th Congress.
Even at these record lengths, however, con�rmation delay still comprised less than half (43%) of
total vacancy time for judicial seats opening in the 114th Congress.
The right plot in Figure 2 shows how presidential nomination delay varies with whether the
9
president’s party controls the Senate. Overall, we �nd relatively little evidence that presidents
nominate federal judges at di�erent speeds based on party control of the Senate. The average
time to nomination is 292 days for nominations made under uni�ed party control and 282 days for
those made under divided control. As the bars in the boxplot show, however, there is considerable
variation in nomination time, particularly in more recent congresses. This �gure provides some
initial evidence that presidents’ nomination strategies may depend more on particular vacancy
characteristics than the global political environment.
Figure 2: Descriptive statistics on vacancy length
Note: The plot on the left shows the average number of days before a judge was nominated and con�rmed (oncenominated) for vacancies in district courts. Time to nomination accounts for a much larger percentage of the timeduring which vacancies are un�lled, though the Senate has exhibited increasing delay in recent years. The plot onthe right compares the time to nomination across congresses with divided vs. uni�ed party control. There are fewsystematic di�erences in time to nomination among presidents based on the presence of a divided rather than uni�edSenate.
Given the nature of our dependent variable, we estimate the predictors of nomination de-
lay using Cox proportional hazards models. Duration models are commonly used in literature
that models Senate response to judicial (Binder and Maltzman 2002, 2004; Massie, Hansford, and
Songer 2004; Shipan and Shannon 2003) and executive branch (Ostrander 2016) nominations. We
follow suit and treat whether a nomination is made at a given time as our dependent variable.
Coe�cients from this model indicate whether the relevant independent variable increases or de-
10
creases the hazard rate, where positive values indicate that larger values of a covariate increase
the hazard rate of a nomination, thereby decreasing nomination delay.
Following earlier studies of federal district court nominations, we record a separate observa-
tion for each congress a vacancy goes un�lled to account for time-variant covariates. For each
vacancy, the �rst observation is the congress during which the vacancy is announced, with a sep-
arate observation for each subsequent congress until the president makes a nomination. This data
structure yields 3,291 observations for 29 congresses. The count of days a vacancy remains open
is recorded up through when a nomination is made or the end of a given congress, whichever
comes �rst.5 As discussed, presidents frequently do not �ll vacancies that open during their ad-
ministration. Because we are interested in presidential decision-making and prioritization, we
restart the count of days at zero when a new president inherits a vacancy from their departing
predecessor. The time count for seats that remain open at the end of a president’s administration
are capped at the end of that congress. We do this to avoid introducing measurement error by
measuring the likelihood of a nomination based on the actions of a presidents’ predecessor. In
some cases, more common earlier in the sample, presidents nominate a judge before a vacancy
is o�cially announced. We code the vacancy duration for such nominations as zero. To account
for correlation in errors across the various shared political units in our sample, we cluster robust
standard errors at the state level.6
Our primary independent variable of interest is Presidential Alignment, which we measure
as the president’s two-party vote share in the most recent election.7 We construct this variable
from county-level returns for each federal district from 1960 through 2018, excluding territorial5For instance, a seat vacated in 2005 and �lled in 2006 would register as one observation, with Time to Nomination
as the di�erence between the vacancy and nomination date. A vacancy beginning in 2006 and ending in 2007 wouldbe recorded as two observations, one for the 109th Congress and one for the 110th, with Time to Nomination runningfrom the vacancy to January 3, 2007 for the 109th and from the vacancy to the nomination for the 110th.
6This decision re�ects the assignment of senators at the state level. While an argument may be made for clusteringat the district level, clustering at the state level yields larger – i.e., more conservative – standard errors.
7We also estimated models using the president’s share of all votes cast, which produce nearly identical results tothose reported below.
11
courts and the District of Columbia, and code them so that larger values indicate greater support
for the sitting president. Districts are geographically bounded by state borders8 and about half
of the states have only a single contiguous district court. All other states are divided into two to
four districts along county lines.
Figure 3 illustrates how presidential vote share varied across federal judicial district in the
2012 election. Darker colors indicate districts that cast larger proportions of votes for President
Obama. The �gure shows that the political characteristics of federal judicial districts vary sub-
stantially within states. In California, for instance, Obama’s vote share ranged from 75 percent in
the northern district, his highest in any district, to 51 percent in the eastern district. In Tennessee,
Obama won 52 percent of votes cast in the western district, 40 percent in the middle district, and
31 percent in the eastern district. Even in states with two districts, Obama’s support varied sub-
stantially between districts, as in the eastern and western districts of Michigan (59 and 48 percent,
respectively), Virginia (56 and 39 percent, respectively), and Washington (42 and 61 percent, re-
spectively). According to our argument, presidents prioritize nominations to politically-aligned
districts. We therefore expect to observe a positive coe�cient on Presidential Alignment, which
would indicate that vacancies are shorter in duration in districts that supported the president at
higher rates.
We also estimate models that account for several other factors that previous literature sug-
gests may a�ect a president’s nomination strategy. First, while we restart the time to nomination
when a new president inherits a vacancy, we anticipate that presidents may prioritize backlogged
vacancies di�erently relative to newly opened ones. Therefore, we distinguish vacancies as Inher-
ited if a president inherited it from previous administrations. Similarly, we expect that presidents
will pursue di�erent strategies for �lling newly created seats and existing ones, perhaps due to
their di�erent implications for court capacity. While vacated seats may drag down courts’ abil-8The District of Wyoming, which contains Yellowstone National Park, including very small parts of Montana and
Idaho, is the sole exception.
12
Figure 3: Presidential vote share by federal judicial district (2012)
0.3 0.4 0.5 0.6 0.7 0.8
Obama vote share (2012)
ity to deal with existing cases, new seats left empty only deny courts prospective increases in
capacity. To capture this, we code observations as New Seat if the seat has been newly created.
Second, previous studies on nominations to the federal judiciary focus on the potential Senate
constraints on judicial nominations. The Senate’s constitutional role to provide advice and con-
sent on presidential nominations has traditionally been accompanied by two norms of deference
to home-state senators. One is senatorial courtesy, by which other Senators defer decisions on
judicial nominations to Senators of the presidents’ party from a district’s state. Similarly, home
state senators may refuse to submit a written opinion on nominees (a “blue slip”) to the Sen-
ate Judiciary Committee, subsequently treated as a veto of their nomination.9 Following Binder
and Maltzman’s (2002; 2004) approach, we create an indicator, Senate Courtesy, that distinguishes9However, as of early 2018, Senate Judiciary Committee Chairman Charles E. Grassley has begun to push forward
nominees despite blue slips withheld by Democrats.
13
whether a vacancy’s state is represented by at least one Senator from the president’s party. We
create a second variable, Blue Slip Potential, that indicates whether the ideological distance be-
tween a home state senator and the president is more than one standard deviation larger than the
mean president-Senator ideological distance in that congress, measured using DW-NOMINATE
scores.10 These variables also account for a potential confounder; namely, that presidential vote
share may re�ect the presence of ideologically aligned or opposed senators for a given district.
This allows us to distinguish the subsidy e�ect of local political alignment from the presence of
friendly home-state representation. To account for how a president may respond to potential
obstruction from an ideologically hostile Senate more generally, we distinguish Divided Senate
based on whether the opposition party controlled the Senate for a given congress.
Third, unlike the Supreme or Appellate Courts, district court cases are usually heard by in-
dividual judges rather than panels. Nonetheless, presidents may focus their nomination strategy
around maximizing courts’ ideological alignment with their policy preferences (for applications
of this theoretical perspective to the Supreme Court, see, e.g., Krehbiel 2007; Moraski and Shipan
1999). To address this possibility, we include a measure of Party Composition, re�ecting the per-
centage of judges in a district nominated by a president of the current presidents’ party.
Fourth, presidents may also be sensitive to district-speci�c factors when making nominations.
While six federal districts contain just two seats each, America’s largest district courts – the
Central District of California and the Southern District of New York – each contain 28 judges.
These two districts contain parts of the country’s two largest cities, with millions of people and10This measurement strategy represents a middle ground between approaches found in previous research that
measure the potential for a negative blue slip either as a continuous measure of ideological distance from the presi-dent (e.g., Black, Madonna, and Owens 2014) and by distinguishing only the most ideologically distant Senators (e.g.,Binder and Maltzman 2002). We note that roll-call based estimates of President Trump were not yet available as ofthis writing. Instead, we imputed Trump’s ideology based on the mean value of the ideological estimates for theother Republican presidents who served during this time period (Nixon, Ford, Reagan, and the two Bushes). These�ve Republican presidents had very similar NOMINATE estimates, with a mean of 0.600 and a standard deviationof 0.087. However, we note that our results are not sensitive to this particular measurement strategy, as our resultsare robust to characterizing Trump’s ideology as equivalent to Gerald Ford and George W. Bush, the most moderateand conservative presidents, respectively, during this time period (see Appendix C) and when excluding the Trumppresidency (see Appendix F).
14
major �nancial institutions within their jurisdiction. Moreover, the marginal e�ect of a single
vacancy on a court’ capacity to process cases or its partisan composition will be greater for a
two- or three-judge court than a 20-judge one. We account for potential di�erences in presidential
strategy around large and small courts with Statutory Court Size, which characterizes the number
of judges allocated by statute to each district at the start of a given congress. This measure tracks
changes in the size of courts over time that may capture other relevant confounders at the district
level, such as population density. Arizona, for instance, saw judgeships increase from two to 13
over the course of our sample, while Nebraska grew by only one seat over the same period.
In addition, we control for presidents’ institutional capacity to vet potential nominees with a
measure of the number of Total Vacancies in need of nominees in a given congress.
Finally, in all our models we include �xed e�ects for presidential term to account for nomina-
tion strategies that may vary across and within presidents’ terms in o�ce. Given this speci�ca-
tion, our main coe�cient of interest is estimated by comparing nomination delay among districts
within a presidential term.11
Results
Before reporting the results of our regression models, we show the estimated smoothed hazard
function in Figure 4.12 This function visualizes the life cycle of federal district court vacancies.
The likelihood of survival, in this case meaning the probability a nomination is not made, de-
creases monotonically each day a seat remains open. However, the function is clearly non-linear,
with the chances of a nomination increasing quickly through the �rst year and slowly plateauing
after the thousand-day mark.11Though our modeling strategy follows previous research and is appropriate given the nature of the dependent
variable, we acknowledge that the chosen modeling strategy is accompanied by inferential limitations. Namely,presidential vote share is not randomly assigned and we must assume that we have accounted for all potentialconfounders. We return to this issue in the conclusion.
12The hazard function is estimated from model (4) in Table 1.
15
Figure 4: Estimated Hazard Function
Note: Smoothed hazard function is estimated from model (4) in Table 1. The probability a president makes a nomi-nation increases the longer the seat remains vacant.
Table 1 presents the results from our regression models. Column (1) reports results from our
baseline model which includes Presidential Alignment along with presidential term �xed e�ects.
The coe�cient estimate for Presidential Alignment is positive and statistically signi�cant, provid-
ing initial support for our hypothesis. This �nding is consistent across our models. In column (2),
we add the indicator for inherited vacancies and the indicator for newly created judgeships. The
model in column (3) includes covariates that characterize the president’s political relationship
with the Senate and column (4) includes covariates that account for factors relating to the courts
themselves. In each, we �nd that a district’s political alignment with the president is consistently
and signi�cantly related to the speed with which presidents make nominations. Using the esti-
mates from model (4), we estimate that a one standard deviation increase in district vote share
(approximately 10 percentage points) increases the hazard rate (and thus the likelihood of nomi-
16
nation) by approximately 21 percent.13 The results in Table 1 thus provide strong and consistent
evidence for our hypothesis.
The coe�cients for the other covariates are also of substantive interest. Somewhat surpris-
ingly, we �nd that the hazard rate is lower for vacancies inherited by a new president. When pres-
idents inherit seats, they are 52% less likely to make a nomination than for a seat that became
vacant during their tenure. Interestingly, we �nd modest evidence that the political composi-
tion of the Senate is associated with the speed with which presidents make judicial nominations.
While the coe�cient estimates for Blue Slip Potential are small in magnitude and indistinguish-
able from zero, we do �nd that Divided Senate has a signi�cant negative e�ect on the speed of
nomination, associated with a 34% decrease in the hazard rate. Senate Courtesy is positively and
signi�cantly correlated with presidential nomination speed. The hazard rate is approximately
41 percent higher for nominations to vacancies in states where at least one Senator shares the
president’s party, indicating that nomination delay is considerably lower in these circumstances.
This �nding provides evidence for a potential mechanism through which local political align-
ment in�uences presidential priorities. Interestingly, these results also suggest that the potential
for Senate obstruction may be counteracted by the presence of a partisan ally. Where opposition
party control of the Senate is associated with longer nomination delay, this e�ect is greatest for
courts in states where that party holds the �rmest grip.
Interestingly, we �nd evidence that presidents systematically move to �ll seats on smaller
courts faster than larger ones. While a single seat increase in Statutory Court Size produces only
a small (2%) decrease in the risk of a nomination, it points to a signi�cant di�erence in nomination
strategy around large and small courts. This re�ects a potential means through which (intention-
ally or not) presidents appear to prioritize nominations to courts where additional judges would
have the greatest consequences for court capacity.
We also �nd that nominations for individual seats occur more quickly as more vacancies are13Smoothed estimates of the hazard rate for this and other models are shown in Appendix D.
17
open in a given congress, holding all else equal. While the change in the hazard rate for a one ad-
ditional vacancy increases just over half of a percent, each congress sees on average 117.5 open
seats, pointing toward a much larger e�ect. Both the Statutory Court Size and Total Vacancies
covariates gesture at how changes in presidential and court capacity may alter president’s nomi-
nation strategies. The prioritization of certain types of courts over others points to the potential
for some non-trivial number of persistently empty seats within the federal judiciary. We return
to the implications of these �ndings in the �nal section of this paper.
18
Table 1: E�ects of Presidential Alignment on Vacancy Duration
(1) (2) (3) (4)
Presidential Alignment 1.919∗∗ 1.889∗∗ 1.294∗ 1.140∗
(0.407) (0.415) (0.521) (0.495)
Inherited −0.356∗∗ −0.392∗∗ −0.416∗∗
(0.073) (0.075) (0.070)
New Seat −0.356∗∗ −0.382∗∗ −0.449∗∗
(0.124) (0.127) (0.116)
Senate Courtesy 0.331∗∗ 0.345∗∗
(0.108) (0.106)
Blue Slip Potential −0.080 −0.074(0.083) (0.077)
Divided Senate −0.340∗∗ −0.293∗∗
(0.083) (0.093)
Partisan Composition 0.194(0.133)
Satutory Court Size −0.020∗∗
(0.004)
Total Vacancies 0.005∗∗
(0.001)
Term �xed e�ects X X X X
Observations 3,291 3,291 3,291 3,291Log Likelihood −13,775.100 −13,747.480 −13,716.170 −13,688.130
Note: Coe�cients are estimated from Cox proportional hazards model with robust standard errors in parenthesesclustered on state. President-term �xed e�ects are included but not reported.* indicates p < .05, ** p < .01 (two-tailed tests).
The results in Table 1 are robust across a variety of additional analyses. First, to evaluate
whether our models comply with assumptions of the Cox proportional hazard model, we test
and correct for nonproportionality among our covariates. These corrections do not substantially
alter our results, as Presidential Alignment does not show evidence of non-proportionality and its
19
coe�cient in the corrected model continues to be positive and statistically signi�cant.14 Second,
our results are not driven by particular districts, time periods, or observations. We estimated
model (4) from Table 1 while sequentially omitting each congress, presidential term, president,
district, and circuit. Across these 158 additional regressions, the coe�cient estimate for Presiden-
tial Alignment is positive in each, stable in magnitude, and statistically distinguishable from zero
in all but six. These additional analyses indicate that the �ndings in Table 1 are not driven by any
particular subset of observations and may be found in Appendix F.
We also explored the extent to which changes in Senate rules may have a�ected presidents’
nomination strategies. In November 2013, a majority of the Democratic-controlled Senate voted
to invoke the so-called nuclear option, which eliminated the use of the �libuster for judicial nom-
inations to federal courts other than the Supreme Court. We conducted a variety of analyses,
distinguishing vacancies that opened before and after this Senate rule change. Overall, the re-
sults indicate that the nuclear option signi�cantly increased nomination delay – perhaps contrary
to the goals of its proponents. However, this e�ect was conditioned by Presidential Alignment,
with faster post-nuclear option nominations in districts that were politically aligned with the
president. 15 We are reluctant to overinterpret these results, however, given the relatively small
number of vacancies that occurred beginning in November 2013; however, they do provide some
suggestive evidence about how political incentives and institutional procedures interact to a�ect
presidents’ nomination strategies.
Interbranch Con�ict and Presidential Priorities over Judicial Vacancies
As indicated by our �ndings above, the presence of a hostile majority in the Senate signi�-
cantly slows presidential nominations. During divided Senate control, a president may anticipate
greater scrutiny from the Judiciary Committee, especially of their nominees for districts held by14See Appendix E.15These results are shown in Appendix G.
20
the Senate majority party. Presidents will look to maximize their impact on the court by focusing
their nomination e�orts on seats that have the highest likelihood of being swiftly con�rmed. If
our argument that informational subsidies from local organizations lead presidents to prioritize
politically aligned vacancies is correct, we should anticipate that these subsidies will be of even
greater value to the White House under the stricter institutional constraints of divided govern-
ment. Presidents should therefore prioritize nominations in friendly districts to a greater degree
when faced with an unfriendly Senate.
To test how partisan con�ict between the Senate and the president conditions the e�ects of
local alignment on nomination strategy, we estimate our models while including an interaction
for divided government and presidential vote share. The results, presented in Table 2, show that
presidents heavily prioritize �lling politically-aligned vacancies when faced with a hostile Sen-
ate. In these models, under divided Senate control, nominations are more than three time less
likely at any given point than when the president’s party holds it. However, this e�ect is strongly
conditioned by Presidential Alignment. The hazard ratio for the interaction of presidential vote
share with divided government is highly signi�cant and large, with a standard deviation increase
in Presidential Alignment accounting for a 68% increase in the likelihood of a nomination. During
uni�ed government, the e�ect of presidential vote share remains positive but is no longer distin-
guishable from zero when including other senatorial factors. This suggests our initial results are
driven primarily by periods of interbranch con�ict when constraints on presidential capacity are
greatest.
21
Table 2: Interbranch Con�ict and the E�ects of Presidential Alignment on Vacancy Duration
(1) (2) (3) (4)
Presidential Alignment 1.058∗∗ 1.028∗ 0.439 0.441(0.384) (0.404) (0.453) (0.472)
Divided Senate −1.534∗∗ −1.580∗∗ −1.655∗∗ −1.392∗∗
(0.369) (0.375) (0.376) (0.351)
Presidential Alignment × Divided Senate 2.349∗∗ 2.314∗∗ 2.450∗∗ 2.053∗∗
(0.657) (0.676) (0.678) (0.643)
Inherited −0.371∗∗ −0.386∗∗ −0.411∗∗
(0.075) (0.079) (0.073)
New Seat −0.383∗∗ −0.387∗∗ −0.451∗∗
(0.124) (0.125) (0.116)
Senate Courtesy 0.343∗∗ 0.352∗∗
(0.106) (0.104)
Blue Slip −0.072 −0.066(0.080) (0.074)
Partisan Composition 0.188(0.132)
Statutory Court Size −0.018∗∗
(0.004)
Total Vacancies 0.005∗∗
(0.001)
Term �xed e�ects X X X X
Observations 3,291 3,291 3,291 3,291Log Likelihood −13,760.290 −13,729.570 −13,707.020 −13,682.000
Note: Coe�cients are estimated from Cox proportional hazards model with robust standard errors in parenthesesclustered on state. President-term �xed e�ects are included but not reported.* indicates p < .05, ** p < .01 (two-tailed tests).
Overall, we �nd strong and consistent evidence that the political alignment between pres-
idents and the judicial vacancies they have the constitutional authority to �ll is an important
predictor of presidents’ nomination strategies. Presidents appear to prioritize nominations to dis-
22
tricts with which they are politically aligned, and this relationship is particularly strong during
periods of institutional con�ict. Consistent with our argument, these patterns provide new evi-
dence that presidential nomination strategies may be contingent on the political characteristics
of the available vacancies. In particular, the �ndings suggest that presidents privilege vacanies
in politically supportive districts where the informational costs of nominee selection and vetting
are likely to be lowest.
The Consequences of Nomination Delay
In this section, we explore two potential consequences of the patterns documented above for
the federal judiciary. First, we study the implications of presidents’ political incentives for the
characteristics of the nominees they put forward, focusing speci�cally on nominee quality. The
normative consequences of the patterns shown above may depend on the extent to which po-
litical expediency substitutes for high-quality judges. If longer nomination times result in the
selection of better-quali�ed judges due to, for instance, more thorough vetting, this might sug-
gest that the patterns documented above have pernicious implications for the composition of the
federal bench. Second, we study how presidential delay contributes to the district courts’ institu-
tional capacities. To the extent presidents prioritize vacancies to some districts over others, our
argument may imply that judicial burdens are greatest in districts that are not aligned with the
current president’s partisanship.
Nomination Delay and the Quality of Judicial Nominees
Our argument suggests that the connection between presidential alignment and nomination
delay re�ects the di�culty of �nding well-quali�ed, but ideologically sympathetic nominees in
politically hostile federal districts. The pool of quali�ed liberal lawyers in the Eastern District of
Arkansas or conservative ones in Vermont, for example, is likely quite small, suggesting presi-
23
dents will take longer to identify viable candidates than in more ideologically-aligned districts
where like-minded lawyers are more plentiful. Therefore, presidents’ prioritization of vacan-
cies in politically-aligned districts is in part a function of the size of the potential pool and the
president’s information about the candidates in it. Our expectation contrasts with a potential
competing hypothesis that suggests presidents prioritize politics over quali�cations. According
to this explanation, the decreased transaction costs of nominations to politically-aligned districts
may permit presidents to get away with nominating less quali�ed candidates to those seats. Dis-
tinguishing between these competing accounts allows us to evaluate the normative implications
of our main results for the composition of the federal judiciary.
We examine whether greater nomination delay increases the likelihood of selecting better
quali�ed nominees to federal district courts. To do so, we use quali�cation ratings from the
American Bar Association.16 Since 1963, the ABA’s Standing Committee has given each nominee
for the judiciary a rating of “Exceptionally Well Quali�ed,” “Well Quali�ed,” “Quali�ed,” or “Not
Quali�ed.” For simplicity, we collapse these categories to distinguish nominees who were rated
as “Exceptionally Well Quali�ed” or “Well Quali�ed” from other nominees, which represents
slightly more than half of the sample. We model whether the president’s nominee received a
Well Quali�ed endorsement as a as a function of Time to Nomination.
The bivariate relationship is displayed in Figure 5.17 Contrary to the hypothesis that longer
vacancies are associated with higher quality nominees, we �nd that lengthy seat vacancies are
associated with a small but statistically signi�cant decrease in the likelihood that a president’s16We acknowledge two limitations of the ABA’s ratings for evaluating nominee quality. First, previous scholarship
has noted that given that conservatives (e.g., Smelcer, Steigerwalt, and Vining 2011) and minority and women nom-inees (e.g., Sen 2014) receive systematically lower ratings than liberals and white men, respectively. Second, amongpresidents’ con�rmed nominees, ABA quality ratings do not correlate with other measures of judicial performance,such as reversal rates (Sen 2014). Despite these limitations, the ABA ratings provide the most comprehensive datasource for studying the professional quality of judicial nominees during the period of our study and, unlike othermeasures of judge performance, allow us to evaluate the quali�cation ratings of nominees who were not ultimatelycon�rmed.
17We �nd nearly identical results when using the logged value of the number of days to nomination (see FigureH.1).
24
nominee was rated as highly quali�ed by the ABA. This relationship persists when estimating a
fully speci�ed logistic regression that includes Time to Nomination and all variables from model (4)
from Table 1 as covariates.18 In this model, vacancy length remains negative and statistically sig-
ni�cant, indicating that a one standard deviation increase in time to nomination (approximately
267 days) decreases the probability that the president nominates a “highly quali�ed” nominee
by about �ve percentage points. While there is no signi�cant relationship between Presiden-
tial Alignment and nominee quality, the results suggest that political factors – here, the political
alignment of federal judicial districts – may have an indirect e�ect on the make-up of the court.
Figure 5: Nomination Delay and Nominee Quality
Note: Data shows nominees to federal judgeships’ designation as "well quali�ed" and "exceptionally well quali�ed"by the length of time to nomination. To avoid measurement error, time to nomination is measured only for the nom-inating president’s administration in the case of inherited vacancies time spent on the nomination during previouspresidencies.
Our results thus provide no evidence that the political incentives that lead presidents to prior-18The results of this model can be found in Appendix H. Moreover, in models that contain quadratic and cubic
expressions of Time to Nomination, only the linear term is statistically signi�cant.
25
itize certain nominations result in the selection of less-quali�ed individuals. This indicates delays
do not re�ect a more thorough search for high-quality judges. To the contrary, extended delay
appear to be associated with the selection of less-quali�ed candidates. These �ndings suggest that
shortened search times in politically aligned districts stem in part from the greater pool of qual-
i�ed like-minded lawyers and the president’s reliance on local information sources to identify
them, giving further credence to our argument that district-level variation in the informational
costs of vetting shape presidents’ nomination strategies.
Judicial Vacancies and Institutional Capacity
If presidents prioritize politically favorable districts when �lling the federal judiciary, how
does the unequal distribution of vacancies shape how the courts do their job? The above �ndings,
we suggest here, hold important implications for the constitutional system of the separation of
powers. Not only do presidents shape the political and ideological leanings of courts by selecting
who sits on them, but they also help determine whether there are judges sitting at all. While
judges who take senior status continue to hear cases, they do so at signi�cantly diminished rates
relative to their active peers. The number of active judges, therefore, has implications for the
capacity of the courts to hear the cases before them. According the American Bar Association,
“Persistent vacancies in a busy court increase the length of time that litigants and businesses
wait for their day in court, create pressures to ‘robotize’ justice, and increase case backlogs that
perpetuate delays in the future.”19 This raises the possibility that presidential prioritization of
politically aligned districts has distributive consequences for courts’ institutional capacity and
quality of decision-making.
However, the e�ect of vacancies on individual court performance has not been well docu-
mented empirically. Should vacancies have no e�ect on how the courts perform their consti-19See “Judicial Vacancies,” November 18, 2018, American Bar Association (available at https://www.
americanbar.org/advocacy/governmental_legislative_work/priorities_policy/independence_of_the_judiciary/judicial_vacancies/; accessed March 27, 2019).
26
tutional duty, then presidential prioritization could simply re�ect presidents’ well-documented
interest in appointing ideological allies to the court in the most e�cient way possible. If vacancies
do in�uence court capacity, however, our initial �ndings would suggest presidents’ appointment
powers shape the performance of adjoining branches of government. To further explore the im-
plications of presidential nomination strategies, we examine how vacancies a�ect court capacity.
With the expectation that departing judges’ workload will be transferred to the remaining judges,
we hypothesize that vacancies will decrease the courts’ capacity to e�ciently hear and decide on
cases. If our hypothesis is correct, it implies that presidents’ nomination powers lead to the dis-
proportionate concentration of judicial capacity in districts where voters supported for the sitting
president.
We study the capacity of the federal courts with original data drawn from federal caseload
reports for the 89 state district courts from 2001 to 2018.20 These reports, published quarterly
by the Administrative O�ce of the U.S. Courts, document court performance over the previous
year. They also record the number of months any seat remained vacant within that period for
each court. A seat that went un�lled for an entire year is coded as 12, as would two separate
six-month vacancies. Using a �xed e�ects panel design, we estimate the e�ect of the number
of vacancy months in a given year on two measures of court performance: the number of trials
completed in a year and the percentage of ongoing civil cases over three years old. We anticipate
that longer vacancies mean courts complete fewer trials and have more long-lasting cases.
Because a court’s performance is likely correlated with the number and di�culty of cases it
receives in a given year, we control for each courts’ pending cases at the end of the previous year
and the number of �lings it received in the current year, weighted by time-intensiveness.21 We20These years are selected based on the Administrative O�ce of the U.S. Courts’ practice of reporting �ndings
annually for six year intervals. We use the most recently available reports, September 2018, and aggregate data fromSeptember-to-September years. The year 2001 is dropped from our models due to calculating lagged variables withinour panel.
21Used by the Judicial Resources Committee since 2004, an average case is given a weighting of 1, while low-e�ortcases (such as student loan defaults) are scaled down to between 1 and 0. More time intensive cases, such as a death-penalty habeas corpus case, are weighted to just under 13. Thus, weighted �lings provide a more accurate estimation
27
include district �xed e�ects to capture all other unobserved confounding for individual districts,
such as size or location. Year �xed e�ects are included to account for nationwide trends over time,
and we present robust standard errors clustered by district to account for potentially correlated
errors. Therefore, our coe�cients of interest are identi�ed with within-district change in vacancy
months across years.
The results in Table 3 show that longer vacancies have negative consequences for court per-
formance. More vacancy months decrease the number of trials courts completed in a given year
and increase the percentage of long-lasting cases. The estimates suggest that districts with va-
cancies that last for a full year reduce the number of trials they complete by about four, which
corresponds to a reduction of approximately three percentage points relative to a district average
of 139. Similarly, our model predicts that districts with an empty seat for a full year will expe-
rience a one percentage point increase in the number of cases that last for three years or more.
Though relatively small, the magnitude of this �nding is not trivial as it represents roughly 13%
of the sample mean. With many courts balancing several thousand cases at a time, this suggests
that hundreds cases of cases may be stretched out each year due to nomination delay. While we
do not wish to overstate the results given the relatively small magnitudes of the estimated e�ects,
we also note that these �ndings do not take into account how courts decide on cases. Both small
changes in the number of completed trials and length of ongoing cases may re�ect how judges
prioritize shorter cases or “robotize” their decisions when operating below full capacity.
of caseload than the raw total of �lings.
28
Table 3: Vacancies and Judicial Performance
Completed Trials % Cases 3+ Years Old
Vacancy Months −0.331∗∗ −0.321∗ 0.079 0.091∗
(0.127) (0.133) (0.042) (0.041)
Weighted Filings (in Thousands) −0.365 −1.252∗∗
(1.024) (0.350)
Previous Pending Cases (in Thousands) 0.008 0.825∗∗
(0.684) (0.296)
Unit �xed e�ects X X X XYear �xed e�ects X X X X
Observations 1,513 1,513 1,513 1,513
Note: Entries are linear regression coe�cients with standard errors clustered on district in parentheses.* indicates p < .05; ** p < .01 (two-tailed tests).
These results cast the potential implications of our initial �ndings in a starker light. Pres-
idents’ delays or prioritization not only shape who eventually sits on the bench, but also how
remaining judges keep up in their absence. While presidents are more responsive to the small
courts most a�ected by vacancies, the �ndings also suggest another strategy available to presi-
dents. Not only do presidents prioritize aligned districts for e�ciency reasons, but they may also
have strategic incentive to leave seats in politically hostile districts open to hamper those courts’
ability to take up cases in an e�ective manner. The �ip side of presidential prioritization could
give rise to patterns of (potentially unintentional) politically motivated court-jamming. We leave
this only as a suggestion and a gesture toward how these �ndings bear on the political dynamics
that a�ect how the system of separated powers operates in practice.
29
Conclusion
“Our nation is disadvantaged when our federal judiciary does not have su�cient judges to
hear cases and resolve disputes in a thorough and timely fashion,” reports the American Bar
Association.22 While in recent years the Senate has taken much of the blame for the slow pace of
�lling vacancies on the federal bench, the onus is on the presidency as well. So far as the sta�ng of
the federal judiciary is necessary for the courts to uphold their end of the American constitutional
system, presidents’ choices about nominees to the bench shape not only its ideological leanings,
but also its institutional capacity. However, presidents face their own capacity constraints. With
the potential for hundreds of vacancies at any one time, selecting viable nominees requires that
the White House shift precious time and resources away from its myriad other political goals.
In this paper, we provide evidence that judicial nominations are responsive to districts’ polit-
ical alignment. We argue that political alignment with the president re�ects the relative strength
of allied legal groups in given districts, who help subsidize search costs by providing the White
House with valuable information about prospective nominees. Alternately, opposition groups in
hostile districts may make getting nominees con�rmed more di�cult, especially during divided
government, requiring presidents to spend even more time �nding viable candidates. Theories
of judicial selection which posit that presidents prioritize ideological or policy in�uence when
selecting nominees may underestimate the capacity constraints presidents face when �lling the
dozens of seats that open each year. Rather than place ideologically friendly judges in politically
hostile districts, we �nd presidents often follow the path of least resistance and prioritize districts
where they enjoy the greatest support.
These presidential incentives have downstream implications for the functioning of the judi-
cial branch. Presidents’ discretion to decide when to nominate, and by extension, the Senate’s to22See “Judicial Vacancies,” November 18, 2018, American Bar Association (available at https://www.
americanbar.org/advocacy/governmental_legislative_work/priorities_policy/independence_of_the_judiciary/judicial_vacancies/; accessed March 27, 2019).
30
con�rm, shape the institutional capacity of the third branch of government. Our �ndings point
to an imbalance in which the courts’ strength as a check on the other branches is rendered con-
ditional on those branches’ choices.
We note several limitations of our own research and opportunities for further study. First,
while our empirical strategy follows those commonly used to study the pace at which the White
House announces nominations, it limits the causal inferences we can make. While an alternative
modeling strategy (such as di�erences-in-di�erences) could generate sharper causal estimates,
the nature of the dependent variable and the lack of panel data in this setting introduces di�erent
sets of limitations. Second, our argument emphasizes the importance of locally-connected in-
terest groups, political parties, and o�ceholders for providing information to the White House.
Collecting granular data on the activities of these individuals and organizations would be an im-
portant next step for evaluating our proposed mechanisms. Third, the evidence provided here
suggests that the executive branch, and the White House in particular, is not immune to informa-
tion de�ciencies or the costs of acquiring information. While presidents are sometimes attributed
with informational advantages (Howell, Jackman, and Rogowski 2013; Gailmard and Patty 2012),
in the case of judicial nominations, presidents may rely on other actors who can subsidize infor-
mation costs. Fourth, while our empirics focus on the federal district courts, our argument applies
more generally to other federal courts as well as appointments within the executive branch. Ex-
tending the argument to other settings is an important opportunity for future research. Finally,
we have demonstrated how one political institution – here, the presidency – can a�ect the per-
formance of adjoining branches of government. In doing so, we shed new light on previously
undocumented sources of presidential in�uence. Identifying other means through which inter-
institutional interactions shape institutional capacity is important for more fully understanding
and evaluating systems of separated powers.
31
References
Asmussen, Nicole. 2011. “Female and Minority Judicial Nominees: President’s Delight and Sena-
tors’ Dismay?” Legislative Studies Quarterly 36(4): 591–619.
Basinger, Scott, and Maxwell Mak. 2010. “The Changing Politics of Federal Judicial Nominations.”
Congress & the Presidency 37(2): 157–175.
Bell, Lauren Cohen. 2002. “Senatorial Discourtesy: The Senate’s Use of Delay to Shape the Federal
Judiciary.” Political Research Quarterly 55(3): 589–607.
Berry, Christopher R., Barry C. Burden, and William G. Howell. 2010. “The President and the
Distribution of Federal Spending.” American Political Science Review 104(4): 783–799.
Binder, Sarah A., and Forrest Maltzman. 2002. “Senatorial Delay in Con�rming Federal Judges,
1947-1998.” American Journal of Political Science 46(1): 190–199.
Binder, Sarah A., and Forrest Maltzman. 2004. “The Limits of Senatorial Courtesy.” Legislative
Studies Quarterly 29(1): 5–22.
Black, Ryan C., Anthony J. Madonna, and Ryan J. Owens. 2014. “Quali�cations or Philosophy?
The Use of Blue Slips in a Polarized Era.” Presidential Studies Quarterly 44(2): 290–308.
Cameron, Charles, and Jee-Kwang Park. 2011. “Going Public When Opinion Is Contested: Evi-
dence from Presidents’ Campaigns for Supreme Court Nominees, 1930-2009.” Presidential Stud-
ies Quarterly 41(3): 442–470.
Cameron, Charles M., Jonathan P. Kastellec, and Jee-Kwang Park. 2013. “Voting for Justices:
Change and Continuityin Con�rmation Voting, 1937-2010.” Journal of Politics 75(2): 283–299.
Gailmard, Sean, and John W. Patty. 2012. Learning While Governing: Expertise and Accountability
in the Executive Branch. Chicago: University of Chicago Press.
32
Galvin, Daniel J. 2009. Presidential Party Building: Dwight D. Eisenhower to GeorgeW. Bush. Prince-
ton: Princeton University Press.
Goldman, Sheldon. 1997. Picking Federal Judges: Lower Court Selection from Roosevelt through
Reagan. New Haven: Yale University Press.
Goldman, Sheldon. 1998. “The Judicial Con�rmation Crisis and the Clinton Presidency.” Presiden-
tial Studies Quarterly 28(4): 838–844.
Gordon, Sanford C. 2009. “Assessing Partisan Bias in Federal Public Corruption Prosecutions.”
American Political Science Review 103(4): 534–554.
Hall, Richard L., and Alan V. Deardor�. 2006. “Lobbying as Legislative Subsidy.” American Political
Science Review 100(1): 69–84.
Hendershot, Marcus E. 2010. “From Consent to Advice and Consent: Cyclical Constraints within
the District Court Appointment Process.” Political Research Quarterly 63(2): 328–342.
Hollibaugh, Gary E. 2015. “Vacancies, Vetting, and Votes: A Uni�ed Dynamic Model of the Ap-
pointments Process.” Journal of Theoretical Politics 27(2): 206–236.
Hollis-Brusky, Amanda. 2015. Ideas with Consequences: The Federalist Society and the Conservative
Counter-Revolution. New York: Oxford University Press.
Holmes, Lisa M. 2007. “Presidential Strategy in the Judicial Appointment Process: ‘Going Public’
in Support of Nominees to the U.S. Courts of Appeals.” American Politics Research 35(5): 567–
594.
Howell, William G., Saul P. Jackman, and Jon C. Rogowski. 2013. TheWartime President: Executive
In�uence and the Nationalizing Politics of Threat. Chicago: University of Chicago Press.
33
Jo, Jinhee. 2017. “Now or Later? A Dynamic Analysis of Judicial Appointments.” Journal of
Theoretical Politics 29(1): 149–164.
Jo, Jinhee, David M. Primo, and Yoji Sekiya. 2017. “Policy Dynamics and Electoral Uncertainty in
the Appointments Process.” Journal of Theoretical Politics 29(1): 124–148.
Killian, Mitchell. 2008. “Presidential Decision Making and Minority Nominations to the U.S.
Courts of Appeals.” Presidential Studies Quarterly 38(2): 268–283.
Krehbiel, Keith. 2007. “Supreme Court Appointments as a Move-the-Median Game.” American
Journal of Political Science 51(2): 231–240.
Kriner, Douglas L., and Andrew Reeves. 2015. The Particularistic President: Executive Branch
Politics and Political Inequality. New York: Cambridge University Press.
Martinek, Wendy L., Mark Kemper, and Steven R. Van Winkle. 2002. “To Advise and Consent: The
Senate and Lower Federal Court Nominations, 1977-1998.” Journal of Politics 64(2): 337–361.
Massie, Tajuana D., Thomas G. Hansford, and Donald R. Songer. 2004. “The Timing of Presidential
Nominations to the Lower Federal Courts.” Political Research Quarterly 57(1): 145–154.
Moraski, Bryon J., and Charles R. Shipan. 1999. “The Politics of Supreme Court Nominations: A
Theory of Institutional Constraints and Choices.” American Journal of Political Science 43(4):
1069–1095.
Nixon, David C., and David L. Goss. 2001. “Con�rmation Delay for Vacancies on the Circuit
Courts of Appeals.” American Politics Research 29(3): 246–274.
Ostrander, Ian. 2016. “The Logic of Collective Inaction: Senatorial Delay in Executive Nomina-
tions.” American Journal of Political Science 60(4): 1063–1076.
34
Rogowski, Jon C. 2016. “Presidential In�uence in an Era of Congressional Dominance.” American
Political Science Review 110(2): 325–341.
Romer, Thomas, and Howard Rosenthal. 1978. “Political Resource Allocation, Controlled Agen-
das, and the Status Quo.” Public Choice 33(4): 27–43.
Scherer, Nancy. 2005. Scoring Points: Politicians, Political Activists and the Lower Federal Court
Appointment Process. Stanford: Stanford University Press.
Scherer, Nancy, Brandon L. Bartels, and Amy Steigerwalt. 2008. “Sounding the Fire Alarm: The
Role of Interest Groups in the Lower Federal Court Con�rmation Process.” Journal of Politics
70(4): 1026–1039.
Sen, Maya. 2014. “How Judicial Quali�cation Ratings May Disadvantage Minority and Female
Candidates.” Journal of Law and Courts 2(1): 33–65.
Shipan, Charles R., and Megan L. Shannon. 2003. “Delaying Justice(s): A Duration Analysis of
Supreme Court Con�rmations.” American Journal of Political Science 47(4): 654–668.
Shipan, Charles R., Brooke Thomas Allen, and Andrew Bargen. 2014. “Choosing When to Choose:
Explaining the Duration of Presidential Supreme Court Nomination Decisions.” Congress & the
Presidency 41: 1–24.
Smelcer, Susan Navarro, Amy Steigerwalt, and Richard L. Vining. 2011. “Bias and the Bar: Eval-
uating the ABA Ratings of Federal Judicial Nominees.” Political Research Quarterly 65(4): 827–
840.
Steigerwalt, Amy L. 2010. Battle Over the Bench: Senators, Interest Groups, and Lower Court Con-
�rmations. Charlottesville: University of Virginia Press.
Stone, Geo�rey R. 2011. “Understanding Supreme Court Con�rmations.” The Supreme Court Re-
view 2010(1): 381–467.
35
Stratmann, Thomas, and Jared Garner. 2004. “Judicial Selection: Politics, Biases, and Constituency
Demands.” Public Choice 118: 251–270.
Wood, B. Dan. 2009. The Myth of Presidential Representation. New York: Cambridge University
Press.
You, Hye Young. 2017. “Ex Post Lobbying.” Journal of Politics 79(4): 1162–1176.
36
Supplementary Materials for“Presidential Politics and Nominations to the Federal
Judiciary"
ContentsA Summary Statistics 1
B Previously Held Seats 2
C Blue Slip Speci�cation 3
D Smoothed Hazard Functions 4
E Non-Proportionality Checks 6
F Sequential Observation Dropping 8
G Nuclear Option 11
H Predictors of Nominee Quality 12
A Summary Statistics
Table A.1: Summary Statistics of Dependent and Independent Variables, Table 1
Mean Stan.Dev. Min. Max.Days to Nomination 307.091 298.743 0.000 2920.000Presidential Alignment 0.537 0.091 0.000 0.809Inherited 0.150 0.357 0.000 1.000New Seat 0.251 0.433 0.000 1.000Senate Courtesy 0.730 0.444 0.000 1.000Blue Slip Potential 0.327 0.469 0.000 1.000Divided Senate 0.427 0.495 0.000 1.000Partisan Composition 0.480 0.257 0.000 1.000Statutory Court Size 9.836 7.314 0.000 28.000Total Vacancies 126.527 36.785 40.000 194.000
Table A.2: Summary Statistics of Dependent and Independent Variables, Table 3;
Mean Stan.Dev. Min. Max.Completed Trials 139.981 101.341 6.000 836.000Perc. Cases Over 3 Y.O. 7.352 9.700 0.000 90.300Vacancy Months 7.681 11.178 0.000 85.400Weighted Filings (in Thousands) 3.657 3.378 0.414 25.51Previous Pending (in Thousands) 4.036 5.577 0.304 74.620
1
B Previously Held SeatsWe might anticipate that presidents make nominations to seats recently created by statutesdi�erently than vacancies in established seats. For example, vacancies of previously held seatsand new ones are likely to re�ect di�erent capacity costs on the courts themselves – where avacant established seat may have an actively downward e�ect on court capacity, an un�llednew seat is only the delay of a prospective increase in court capacity. To test this and therobustness of our model to the inclusion of new seats, we estimate our model excluding newlycreated seats.
Table B.1: E�ects of Presidential Alignment on Vacancy Duration Among Previously Held Seats
(1)Presidential Alignment 1.191∗
(0.539)Inherited -0.486∗∗
(0.077)Senate Courtesy 0.407∗∗
(0.104)Blue Slip Potential -0.154
(0.102)Divided Senate -0.301∗∗
(0.098)Partisan Composition 0.097
(0.143)Statutory Court Size -0.023∗∗
(0.004)Total Vacancies 0.003
(0.002)Term �xed e�ects X
Observations 3,291Log Likelihood -11,452.050
Note: ∗∗p<0.1; ∗p<0.05
2
C Blue Slip Speci�cationGiven that NOMINATE ideology estimates for Donald Trump are not available at the time ofwriting, we measure blue slip potential for the Trump presidency by treating his ideology as themean estitmate for the �ve other Republicans holding the presidency between 1961 and 2018.This gives us a score of 0.6 with a standard deviation of 0.087. To ensure out model is robust todi�erent speci�cations of Trump’s ideology, we reestimate our models with blue slip indicatorsholding Trump’s ideology equal to Gerald Ford’s (the most moderate Republican in the sample,with a score of 0.506) and to George W. Bush’s (the most extreme at 0.693). Neither speci�cationsubstantially alters our results and in neither case is the blue slip potential indicator signi�cant.Similarly, our �ndings are robust to excluding Trump’s presidency entirely (see Appendix F).
Table C.1: E�ects of Presidential Alignment on Vacancy Duration Across Blue Slip Speci�cation
(1) (2)Presidential Alignment 1.136∗ 1.158∗
(0.497) (0.498)Inherited −0.416∗∗ −0.416∗∗
(0.070) (0.070)New Seat −0.449∗∗ −0.449∗∗
(0.116) (0.116)Senate Courtesy 0.343∗∗ 0.352∗∗
(0.105) (0.103)Blue Slip Potential (Ford) −0.078
(0.079)Blue Slip Potential (Bush) −0.058
(0.079)Divided Senate −0.293∗∗ −0.291∗∗
(0.093) (0.093)Partisan Composition 0.194 0.194
(0.133) (0.132)Statutory Court Size −0.020∗∗ −0.020∗∗
(0.004) (0.004)Total Vacancies 0.005∗∗ 0.005∗∗
(0.001) (0.001)Term �xed e�ects X X
Observations 3,291 3,291Log Likelihood -11,452.050 -11,452.050
Note: ∗∗p<0.1; ∗p<0.05
3
D Smoothed Hazard Functions
Figure D.1: Full model predicted survival curve of federal district court vacancies with high andlow presidential alignment. All other covariates held at their means.
4
Figure D.2: Interaction model predicted survival curve of federal district court vacancies withhigh and low presidential alignment. All other covariates held at their means.
Figure D.3: Interaction model predicted survival curve of federal district court vacancies fordivided and uni�ed government, with high and low levels of presidential alignment. All othercovariates held at their means.
5
E Non-Proportionality ChecksThe primary assumption of the Cox proportional hazards model is the proportionality ofcovariates. We test for non-proportionality by checking for the independence of Schoenfeldresiduals for each covariate and time. While the e�ect of Presidential alignment does not changesigni�cantly over time, these tests show several controls in our model have signi�cantnon-proportional e�ects on vacancy length. To correct for this, we follow Box-Ste�ensmeierand Zorn (2001) in running a non-proportional hazards model by interacting our covariateswith log transformation of the duration of a vacancy.
Table E.1: Non-Proportionality Test the E�ects of Presidential Alignment on Vacancy Duration
(1)Presidential Alignment 0.581∗
(0.269)Inherited 0.099
(0.067)New Seat 0.987
(0.590)Senate Courtesy 7.636∗∗
(2.376)Blue Slip -0.164∗
(0.080)Divided Senate 2.187∗∗
(0.472)Partisan Composition 0.349∗∗
(0.092)Statutory Court Size 0.146∗∗
(0.023)Total Vacancies 0.097∗∗
(0.014)New Seat × ln(Days) -0.213
(0.113)Senate Courtesy × ln(Days) -1.296∗∗
(0.390)Divided Senate × ln(Days) -0.406∗∗
(0.076)Statutory Court Size × ln(Days) -0.028∗∗
(0.004)Total Vacancies × ln(Days) -0.016∗∗
(0.002)Term �xed e�ects X
Observations 3,291Log Likelihood -11,452.050
Note: ∗∗p<0.1; ∗p<0.05
6
The results of the non-proportional hazards model re�ects estimates of hazards ratios whentime-correlated covariates are allowed to vary with time. While the results are in line with ourmain �ndings in Table 1, they lead us to reassess our interpretation of several variables. While itis not itself time-varying, Presidential alignment remains signi�cantly correlated with shorternomination periods in the non-proportionality-corrected model. However, the coe�cient isslightly smaller than in the model presented in Table 1. A standard deviation increase inpresidential vote share is associated with a 7.8% increase in the likelihood of nomination. As forthe time-varying controls, Senate courtesy, Divided Senate, Statutory court size, and Totalvacancies all initially increase the likelihood of nomination, but see this e�ect decrease orreverse over time.
7
F Sequential Observation DroppingTo ensure that our �ndings are not being driven by any outliers, we sequentially drop eachcongress, presidential term, president, district, and circuit from our analysis and plot the results.The �gures below plot the coe�cient estimate for Presidential alignment for each. In a handfulof cases, the signi�cance dips below the 0.05 threshold, but in each case they are directionallyconsistent and approach standard levels of signi�cance.
8
9
10
G Nuclear Option
Table G.1: The 2013 Nuclear Option and the E�ects of Presidential Alignment on Vacancy Du-ration
Whole Sample Post-2009 113th CongressPresidential Alignment 0.854 -0.345 -5.190∗
(0.495) (0.950) (2.248)
Nuclear Option -1.446∗ -1.498∗ -3.179∗∗(0.617) (0.720) (1.201)
Presidential Alignment × Nuclear Option 2.526∗ 2.560 5.071∗(1.158) (1.307) (2.026)
Inherited -0.422∗∗ -0.689∗∗ -19.156∗∗(0.071) (0.228) (0.980)
New Seat -0.454∗∗ -1.846∗∗ -1.223∗∗(0.117) (0.306) (0.404)
Senate Courtesy 0.318∗∗ 0.391 0.858∗(0.106) (0.262) (0.372)
Blue Slip Potential -0.079 -0.779∗∗ -0.996∗∗(0.077) (0.246) (0.309)
Divided Senate -0.279∗∗ -1.005∗∗(0.092) (0.379)
Partisan Composition 0.182 -0.019 0.048(0.134) (0.357) (0.501)
Statutory Court Size -0.021∗∗ -0.021 0.008(0.004) (0.011) (0.022)
Total Vacancies 0.005∗∗ 0.012∗(0.001) (0.005)
Term �xed e�ects X X
Observations 3,291 663 135Log Likelihood -13,630.160 -1,631.006 -373.407
Note: Nuclear option is an indicator for whether a vacancy appeared after the nuclear option was invoked in November2013. The �rst column shows results from the entire time period (1961-2018). The second column shows results whenfocusing just on the Obama presidency (2009-2016). The third column shows results from just the 113th Congress(2013-2014). President-term �xed e�ects are included but not reported.* indicates p < .05 (two-tailed tests), ** p < .01 (two-tailed tests).
11
H Predictors of Nominee QualityOur �ndings of a negative bivariate relationship between nomination delay and nomineequality are consistent when we log the number of days to nomination, showing a decrease inthe likelihood of nominating a Well quali�ed nominee as the logged Time to nomination. Weshow the bivariate relationship below.
Figure H.1: Nomination Delay and Nominee Quality
Note: Data shows nominees to federal judgeships’ designation as "well quali�ed" and "exceptionally well quali�ed"by the logged length of time to nomination. To avoid measurement error, time to nomination is measured only forthe nominating president’s administration in the case of inherited vacancies time spent on the nomination duringprevious presidencies.
To accompany bivariate analysis of the relationship between nomination delay and candidatequality, we run preliminary logistic models to test which variables meaningfully alter thechances of selecting a quali�ed candidate. To do this, we structure our data as a singleobservation for each nomination made between 1961 and 2018 where an ABA rating was given.We measure Time to nomination as the time elapsing between when a seat opened and when anomination was made, or if inherited from an earlier president, from the start of the inheritingpresidency to when a nomination was made. We treat an ABA rating of Well Quali�ed as ourdependent variable. We include all other covariates from our initial models as controls,measuring time-varying covariates at the Congress in which a nomination is made. Time tonomination is signi�cantly and negatively associated with the likelihood of a high qualitynomination, in line with our earlier visual analysis. We also log Time to nomination and getsimilar results.
12
Table H.1: E�ects of Vacancy Duration on Nominee Quali�cation
Dependent variable:
Well Quali�ed(1) (2) (3) (4)
Time to Nomination -0.001∗∗ -0.001∗∗(0.0002) (0.0002)
ln(Time to Nomination) -0.100∗∗ -0.125∗∗(0.031) (0.034)
Presidential Alignment -0.580 -0.512(0.590) (0.589)
Inherited 0.222 0.156(0.169) (0.164)
New Seat 0.053 0.073(0.136) (0.137)
Senate Courtesy 0.026 0.049(0.132) (0.132)
Blue Slip Potential 0.050 0.037(0.119) (0.118)
Divided Senate 0.224 0.240(0.182) (0.181)
Partisan Composition -0.173 -0.196(0.202) (0.202)
Statutory Court Size 0.021∗∗ 0.022∗∗(0.007) (0.007)
Total Vacancies -0.005 -0.005(0.004) (0.004)
Constant 1.010∗∗∗ 1.560∗∗ 1.306∗∗∗ 1.859∗∗(0.174) (0.576) (0.220) (0.583)
Term �xed e�ects X X X X
Observations 1,951 1,951 1,951 1,951Log Likelihood -1,230.346 -1,222.089 -1,232.793 -1,224.776
Note: Coe�cients are estimated from a logistic regression model with robust standard errors in parentheses.President-term �xed e�ects are included but not reported.* indicates p < .05 (two-tailed tests), ** p < .01 (two-tailed tests).
13