Essay Should We Pay Federal Circuit Judges More (or by ikn20172


									Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission


                    Should We Pay Federal Circuit Judges More (or Less)?

                                              Scott Baker*

        On January 1, 2007, Chief Justice Roberts released his annual report on the state
of the federal judiciary.1 In the report, he claimed that inadequate judicial salaries were
precipitating a “constitutional crisis.” According to the Chief Justice, the pay gap
between federal judges and their counterparts in the private sector was becoming so large
that serving on the judiciary was no longer a reasonable option for many highly qualified
lawyers. For this reason, warned the Chief Justice, if the pay gap remained too large “the
judiciary will cease to be made up of a diverse group of the nation’s very best lawyers.
Instead, it will come to be staffed by a combination of the independently wealthy and
those following a career path before becoming a judge different from the practicing bar at
large. Such a development would dramatically alter the nature of the federal judiciary.”2

        The statements of the Chief Justice were correct at least insofar as they accurately
described the larger (and growing larger) pay differential between federal judges and
private sector lawyers. In 2005, the average partner in a prominent Chicago-based law
firm earned $2.12 million.3 By comparison, the judges of the Seventh Circuit, also based
in Chicago, earned $175,000.

        What is less clear, however, is whether the Chief Justice’s conclusion is correct
that the result of this pay gap will be to “alter the nature of the federal judiciary.”
Certainly, Chief Justice Roberts’ instinct could very well be right -- salary differences
might impact who will be willing to join the federal judiciary. Perhaps with low judicial
pay, fewer people will accept the job without accumulating a substantial nest egg
beforehand, and some people with college-age children might decline the judgeship
altogether. But the fact that some persons may no longer want to serve as federal judges
because of pay concerns does not mean that the nature of the federal judiciary will
thereby be fundamentally altered. The critical question is not whether judicial salaries
affect composition, they might,4 but whether the resulting change in composition affects

  Professor of Law, UNC Chapel Hill, School of Law; email:
  Chief Justice John Roberts, 2006 Year-End Report of the Federal Judiciary.
  Chief Justice John Roberts, 2005 Year-End Report of the Federal Judiciary. Chief Justice Roberts isn’t
the only Justice to complain about the destructive effects of low judicial salaries. See Testimony of
Associate Justice Anthony M. Kennedy before the United States Senate Committee on the Judiciary
Judicial Security and Independence, 110th Cong. (2007); Chief Justice Rehnquist, 2003 Year-End Report of
the Federal Judiciary. The sentiments of the Justices echo those of the Volcker Commission – a
commission sent up by Congress to study compensation for government employees. See Nat’l Comm’n on
the Public Service, Urgent Business for America: Revitalizing the Federal Government for the 21st Century
__ (2003).
  The AmLaw 100, 2006, The Am. Law., May 2006, at 165 (reporting 2005 profits per partner at Kirkland
& Ellis).
  Indeed, salaries are likely to affect judicial composition no matter what the level at which they are set.
Except for the limited few for whom money does not matter, the level of compensation will always affect
the desirability of a particular position.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

the “nature” of the federal judiciary, that is to say, whether relatively low judicial salaries
affect the “product” the circuit courts produce.

         This essay tests the impact of judicial pay on judicial performance. By comparing
judicial salaries to the salaries of the next best opportunity for most circuit judges –
partnership in regional law firms – it finds that judicial compensation is irrelevant to
nearly all quantifiable measures of judicial performance. No matter the political
affiliation, judges vote the same regardless of their salary against their next best
opportunity. They are equally likely to heed and cite as persuasive authority opinions by
judges from the other political party. They decide cases in the same amount of time.5
They file the same ratio of published to unpublished decisions.6 Indeed, the only two
statistically discernable differences between well-paid and poorly paid judges are (1)
well-paid judges dissent more often in controversial cases and (2) well-paid judges write
opinions that are, on average, less likely to be cited by judges from other circuits. But
even these two effects are slight. In short, pretty much nothing would happen if Congress
decided to raise judicial salaries.

         This result makes sense. There are very few federal judgeships. Many people
want them. Salary, a generous pension, and a number of non-pecuniary perks make the
circuit judgeship attractive. The president picks his nominee based on his preferences in
combination with the views of the senators, especially the home-state senators. The
composition and depth of the candidate pool makes little difference. True, someone
might turn down the job for financial reasons, but the next person in line will be
indistinguishable in their eventual judicial performance.

         Part IA sets forth the constitutional structure, statutory scheme, and history of the
law governing judicial salaries. Part IB outlines the debate over judicial salaries,
carefully considering the arguments for higher salaries. Part II details two statistical
approaches – judge-to-judge comparisons and pool-to-pool comparisons – that can be
used to unpack whether higher salaries would alter the judicial output of the circuit
courts. Part III describes the law firm salary database from which I derive the net cost of
taking the judgeship for 261 federal circuit judges. Part IV performs the statistical
analysis. It reports that judicial pay does not affect judicial votes in controversial cases or
the citation practices of judges when writing their opinions. Surprisingly, it shows that
judges who give up a lot of money for the bench write somewhat higher quality opinions,
at least according to the quality measure of average number of outside circuit citations
per opinion. By inference, then, low judicial pay (i.e., big spreads between judicial pay
and private sector pay) yields marginally better opinions. Part IV also shows that low
judicial pay leads to slightly fewer dissents. Part V deals with some potential objections
to the analysis. Part VI concludes.

I.          Judicial Salaries: Background

            A.       Constitutional Requirements and Statutory Background

    The results on time to render a decision are still tentative as of this draft.
    The results on the ratio of published to unpublished decisions are still tentative as of this draft.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

         Article III, section 1 of the Constitution provides: “The judges, both of the
supreme and inferior courts, shall hold their offices during good behavior, and shall, at
stated times, receive for their services a compensation, which shall not be diminished
during their continuance in office.” The framers wanted to insulate judges from the
whims of the legislative branch and, thus, ensure a more independent judiciary.7 But they
didn’t account for inflation. The text of the Constitution prevents Congress from
reducing judicial salaries, but it doesn’t require cost of living increases. Without such
increases, inflation diminishes the purchasing power of the judicial salary. As many
others have noted, that is exactly what has happened over the last thirty years – a decline
in the real salary for federal judges.8

        Congress has tackled the problem of judicial salaries a number of times. In 1967,
Congress enacted the Post Revenue and Federal Salary Act.9 This Act established a
commission to review the salary structure of high level members of the executive,
legislative, and judicial branches. The commission made salary recommendation to the
President. The President decided on salaries, which took effect unless Congress
expressly rejected the proposed salary structure. The Post Revenue and Federal Salary
Act resulted in a large judicial pay increase in its first year, but had little effect on salaries

        In its next foray into judicial salaries, Congress enacted in 1975 the Executive
Salary Cost of Living Adjustment Act.11 This Act provided for automatic cost of living
adjustments (COLA) for members of Congress, the executive, and the judiciary. Despite
efforts under this Act to make wage adjustments predictable and consistent, Congress
often rejected the automatic COLA increases for itself and the other branches. Such a
rejection – coupled with the rampant inflation of the late seventies -- meant that judicial
salaries fell almost 30 percent in real terms during this period.

       In 1980, a group of federal district court judges filed a lawsuit claiming that
Congress violated the constitutional guarantee of undiminished judicial salaries by
postponing or repealing previously-enacted automatic COLA adjustments. In United

  THE FEDERALIST NO. 78 (Alexander Hamilton) at 394 (stating that “[I]n a monarchy [fixed judicial
salaries] is an excellent barrier to the despotism of the prince; in a republic it is no less an excellent barrier
to the encroachments and the oppressions of the legislative body. And it is the best expedient which can be
devised in any government to secure a steady, upright, and impartial administration of the laws.”); The
Federalist No. 79 (Alexander Hamilton) at 491 (reflecting on the judicial compensation clause and stating
“[I]n the general course of human nature, a power over a man's subsistence amounts to a power over his
  See Albert Yoon, Love’s Labor Lost? Judicial Tenure Among Federal Court Judges: 1945-2000, 91 Cal.
L. Rev. 1030, 1033 (2003) (figure 1); Richard A. Posner, The Federal Courts 21-34 (1996); Kristen A.
Holt, Justice for Judges: The Roadblocks on the Path to Judicial Compensation Reform, 55 Cath. U. L.
Rev. 513, 515 (2006).
  Pub. L. No. 90-206, § 225, 81 Stat. 613, 642-45.
   See Yoon, supra note ___, at 1036 (speculating that Congress didn’t raise judicial pay after the first year
because “other policy issues gained greater salience”).
   Pub. L. No. 94-82, 89 Stat. 419.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

States v. Will,12 the Supreme Court reinstated the COLA increases for two of the four
years the judges requested. In picking among the COLA increases, the Court
distinguished between COLAs that had “vested” and those that hadn’t. The Court held
that “a salary increase ‘vests’ for purposes of the Compensation Clause only when it
takes effect as part of the compensation due and payable to Article III judges.”13 The
upshot of Will is that Congress cannot repeal COLA increases after the judges have
received them. Congress, however, can repeal a COLA increase that is simply promised,
if no money has yet to be distributed under that COLA. In tracing the history of judicial
salaries, the main effect of Will was to retroactively increase judicial salaries for 1976
and 1978.

        The Ethics Reform Act of 1989 marks the most recent Congressional activity on
judicial salaries.14 Relevant to judicial salaries, this Act accomplished three things. First,
it provided a different measure of the COLA, tying the inflation adjustment in judicial
salary to the adjustment regularly given other federal government employees. Second,
the Act fused any Congressional decision about COLA increases for judges with the
decision about COLA increases for members of Congress and high-level executive
branch officials. That is to say, under the Act, if Congress approved a COLA increase for
the judiciary, it would necessarily approve a COLA increase for itself and executive
officials. This tying effectively froze judicial salaries. The reason: Members of Congress
feared a voter backlash if they gave themselves a raise.

       Third, and unrelated to the issue of COLAs, the Act gave an immediate judicial
pay bump of forty percent. In return, the Act restricted how much judges could engage in
non-judicial activities for compensation. The Act capped the payment for teaching-style
services to 15 percent of the judicial salary. Coupled with the ethical rule restriction on
outside activities, like serving on corporate boards, the cap effectively ensures that
judges’ income will be limited to their official salaries plus some income from teaching.

        B.       The Salary Debate

        Most sitting judges find the current salary system deplorable. Like any other
worker, judges want higher wages, at least enough additional cash to cover increases in
the costs of goods and services. There are three arguments conventionally given for
higher judicial salaries. The first argument centers on retention: Declining real salaries
will result in judges leaving the bench for private practice.15 Turnover might affect
judicial performance because the exit of a sitting judge creates transition costs. The
vacancy has to be filled, and the new judge brought up to speed. Until that happens, the
other judges will have a higher workload, straining the capacity of the circuit court.16

   449 U.S. 200 (1980).
   Id. at 229.
   Pub. L. No. 101-194, 103 Stat. 1752
  Roberts, supra note __, at 8; Testimony of Associate Justice Anthony M. Kennedy before the United
States Senate Committee on the Judiciary Judicial Security and Independence, 110th Cong. (2007) at ___.
   Panel Warned About Inadequate Pay for Federal Judges, 34 The Third Branch _ (2002) (quoting Justice
Breyer as making this argument).

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

        This argument assumes that declining pay leads to higher turnover. Yet, that
doesn’t appear to be the case, at least up to 2000. Albert Yoon has examined the
retirement decisions of all district court and federal court judges between 1945 and 2000.
He finds that “tenure trends among the federal judiciary have held fairly constant over the
past half century, notwithstanding the cyclical decline in inflation-adjusted salaries.”17

        The second argument for higher salaries rests on attracting the “best and
brightest” from the private bar. These lawyers give up a lot to join the bench. Few
talented lawyers in private practice, the argument goes, will make the leap if judicial
salaries remain far below those in the private sector. And that’s bad. This argument
assumes that attracting the best and brightest private sector lawyers will make the
judiciary better in some meaningful sense. These lawyers might decide cases with a
more practical bent; have a greater understanding and appreciation of the real world
consequences of their decisions; or have greater expertise in certain technical subjects,
like, say, securities law. Empirically testing this argument is hard, and this essay does
not aim to do so. It does find, however, that holding constant the net cost of taking a
judgeship, lawyers who come from private practice perform similarly across a range of
judicial output measures as those coming from government service or academia.18

        The third argument for higher salaries has to do with the preferences of those in
the candidate pool. A circuit judgeship brings with it substantial non-pecuniary benefits
and a generous pension. The job offers prestige, power, influence, control of one’s
schedule, and interesting work. It is not hard to find lawyers willing to take circuit
judgeships because the actual wage is only one -- arguably small -- component of the
total compensation package. The intuition must be that lower pay leads to worse judges,
not no judges. This intuition has a sound theoretical basis and generates testable
hypotheses. Call this theory the “salary matters” theory.

        To explore this theory in more detail, note that people care about both non-
pecuniary and pecuniary aspects of a job. Different people place more weight on the
pecuniary aspects of a job and less weight on non-pecuniary aspects of a job. For any
person, we can construct a preference profile indicating how much they subjectively
value each non-pecuniary aspect and each pecuniary aspect of any job. This profile will
differ depending on the wealth of the individual, how much they value consumption
versus leisure, and many other personal factors.

  Yoon, supra note __, at 1032.
  This finding differs from the standard one in the literature. See Lee Epstein et al., The Norm of Prior
Judicial Experience and Its Consequences for Career Diversity on the United States Supreme Court, 91 Cal.
L. Rev. 903, appendix (collecting studies showing how prior experience affects judicial behavior). The
studies reviewed by Epstein et al. look at a variety of judicial “output” measures. In addition to prior
experience, none of these studies consider the net cost of taking the judgeship, the variable of interest here.
For a study that considers the relationship between judicial pay and the production of precedent by
contrasting state court judges with federal circuit judges, see William M. Landes & Richard A. Posner,
Legal Change, Judicial Behavior, and The Diversity Jurisdiction, 9 J. Legal Stud. 367 (1980).

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

        Now take judges. Judges care about a number of things besides money: status,
prestige, leisure, power to affect policy, and public service.19 Different people attach
different weights to these non-pecuniary aspects of the job. The best theoretical case for
higher salaries is that higher judicial wages change the preference profile of those
candidates in the pool.

        The spread between the judicial salary and the wage in a candidate’s next best
opportunity proxies the strength of the preference for becoming a judge. The bigger the
spread is the stronger the preference. And people with strong preferences for the judicial
role might act differently once on the bench. A strong preference for the circuit
judgeship could correlate with a strong preference for power, leisure, prestige, or public
service. By raising salaries, Congress reduces the spread. As a result, higher salaries
might weed out some of the people with the strongest preferences for the judicial role.
Sure, the true ideologue, the leisure maximizer, and the prestige-obsessed will still be
interested in the judgeship. But so will a lot of other people. Under the salary matters
theory, the increased competition will impact who is selected and the eventual judicial
performance of the circuit courts.

        To see why this might be so, suppose that the pay for circuit judges is zero. In
this case, individuals willing to take the job must really want to be a judge. They value
the non-pecuniary aspects of the job a lot – leisure, power, prestige, etc -- money less so
(perhaps because they are wealthy already). Suppose that the pay is increased to
$150,000 a year. In that case, the people who would take the judgeship for nothing
would still compete for the judicial slot, but now people who place a lower value on non-
pecuniary perks and a higher value on wages would enter the pool. The pool would then
be composed of some with strong preferences over the non-pecuniary aspects of the job
and some with weaker preferences. Increasing pay to $2 million a year and the pool
expands even further. It includes some lawyers who don’t care about the non-pecuniary
aspects of the judgeship and a lot about money. In this way, raising judicial pay both
expands the candidate pool and changes the profile of preferences of members of that
pool. These changes, then, translate into a change in the kinds of judges reaching the
bench. In short, the “salary matters” theory is that, with low pay judges who care a lot
about policy outcomes, leisure, prestige, or public service dominant the candidate pool
and get nominated; and people with such strong preferences over the non-pecuniary
aspects of the job act differently once on the bench.

         According to this theory, the appeal of higher salaries is the end result: a judiciary
that is less ideological, harder working, and less concerned about its own prestige and
public service. Assuming that some of these traits are normatively desirable – debatable,
to be sure-- the negative impact of low salaries is clear and testable. With a high spread
between judicial pay and the next best opportunity, the judiciary is composed of people
who are more partisan, lazier, more driven by prestige, and/or place a high premium on
public service. And these judges act like it by, for example, voting more consistently
along party lines (the partisan judge), only citing judges from the same political party (the

  See Richard Posner, What Do Judges and Justices Maximize? (The Same Thing Everyone Else Does), 3
Sup. Ct. Econ. Rev. 1, 31-39 (1993) (developing a model of the judicial utility function).

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

partisan judge), not deciding cases quickly (the lazy judge), filing fewer published
opinions (the lazy judge), or investing time ensuring they write decisions that others will
cite (the status conscious judge).

        But that is only one view about the role of low judicial salaries. A plausible
alternative view is that the spread doesn’t make a difference. As a result, relatively low
judicial salaries are nothing to fret about. This would be true if political tides select the
same kind of people regardless of pool composition. In this case, deepening the pool to
include those who care less about salary doesn’t make much sense. The judiciary will
have the same number of judges with preferences to be leisure maximizers, ideologues,
and influence-peddlers independent of the wage.

        In other words, a reasonable alternative theory is that (1) politics alone drive
judicial selections and (2) the pool, at present and historic salary levels, is already
saturated with candidates that are near-perfect substitutes for each other. Under this
theory – the “substitutability theory” -- higher judicial salaries are hard to justify.
Expanding the pool doesn’t change the type of person who reaches the bench. The
President has his man or woman picked out already. Perhaps the President nominates the
most extreme ideologue he can get confirmed or the President uses the circuit judgeship
to reward party loyalists. A salary bump might be necessary to induce that person to join
the bench. But even if that person turns down the job, the next person selected will be
indistinguishable, both in terms of their preferences over the non-pecuniary aspects of the
judgeship and resulting judicial performance.

        This essay empirically tests the last reason to raise judicial salaries: the need to
weed out people with strong preferences for the non-pecuniary aspects of judging. It
compares the salary matters and substitutability theories to see which one the data
supports. In so doing, the empirical model of section IV considers whether the federal
circuit courts would perform differently – along a number of measures of judicial
performance -- if salaries were higher.

II.      Unpacking the Impact of Higher Judicial Salaries – Two Statistical Approaches

         a)       Direct Comparison Approach

         This first statistical approach used to compare the two theories holds everything
else constant and asks whether people who give up more money to become judges simply
“want the job more” than people who give up less money. Moreover, this strong
preference for the job translates into (1) a stronger desire to impose policy preferences;
(2) a stronger desire for leisure; and/or (3) a stronger desire to exert judicial influence.20

       To illustrate this approach, consider two judges, X and Y. Assume a republican
president appoints both judges. Suppose that judge X gave up $1.5 million to take the
bench, whereas judge Y gave up $700,000. Next suppose that Congress raises judicial

 I don’t test the strong preference for public service. The reason is that I am unsure what judicial output
measure would correlate with such a preference.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

salaries. With higher salaries, judge X would still compete for the judgeship. Since he
was willing to give up $1.5 million to take the judgeship, he would certainly take the
position if offered an even higher judicial wage. For the sake of this example, suppose
that after the pay raise judge X only had to give up $700,000 to take the bench. At the
same time, with increased judicial salaries, more people would be interested in judge X’s
slot. These additional people, the thinking goes, might be more like judge Y. If pool
matters in addition to politics, perhaps someone like judge Y would reach the bench
instead of judge X. That is to say, the pool expansion induced by the salary increases
might change the eventual appointment. As a result, higher pay might change the
composition of the judiciary: more judge Ys, fewer judge Xs. But that change in and of
itself does not support the “salary matters” theory.

        Suppose it turns out that judge X, with a big spread, and judge Y, with the small
spread, act the same on the bench. They vote the same; they decide cases in the same
amount of time; they file the same number of published opinions; they are equally likely
to dissent; they are equally likely to cite democratic judges as persuasive authority; and
they are equally likely to be cited by other judges. Because judge Y and judge X are
interchangeable once on the bench, it doesn’t make any difference if there are more judge
Ys and fewer judge Xs. Composition -- and hence the size of judicial salary -- is
irrelevant because judge X and judge Y are perfect substitutes for each other. If the
spread between a judge’s next best opportunity and the judicial salary is a statistically
insignificant predictor of judicial performance, the world more likely consists of lots of
perfect substitute candidates who want the job. Moreover, at least one of them will take
the job at the prevailing wage. In this world, an increase in pay would not affect judicial
performance. The same person or their close substitute would reach the bench whether
or not Congress raised salaries. Such a statistical result – i.e., judges with low spreads act
the same as judges with high spreads – supports the “substitutability” theory.

       Using the direct comparison approach, I assume that, no matter the previous job
held by a candidate, anyone qualified to become a federal circuit judge could get a job as
a lawyer in an average size law firm in that person’s region of the country during that
time. For example, I assume that when taking a judgeship, a federal prosecutor based in
San Francisco forgoes more purchasing power – because of the high law firm salaries in
California – than a federal prosecutor based in St. Louis. This is true despite roughly
equivalent pay in the San Francisco and St. Louis U.S. attorney offices.

        b)       Pool Comparison Approach

        The direct comparison approach described above doesn’t capture the strength of
the pool of nominees from which the President selects. It only looks at those candidates
nominated and confirmed to the bench – a self-selected group. According to the salary
matters theory, the greater the spread for the average pool member, the more likely the
pool will be composed of individuals who place a high value on the non-pecuniary
aspects of joining the bench. A president selecting from a pool of people who value
highly the judicial role is more apt to select a person with strong preferences for the job.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

Higher salaries, under this measure, change the pool composition by bringing people with
tempered preferences and motivations into the fold.

        Suppose that judge X comes from a pool of people with high net costs of taking
the judgeship, whereas judge Y comes from a pool of people with low net cost of taking
the judgeship. Raising salaries means more candidate pools with low net costs and,
holding all else equal, more judges like judge Y will be selected. Again, we can then
just look at the actual behavior of judge Y to see what a judiciary composed of more
judge Ys would look like. Again, if judge X and Y act the same, there is no reason to
raise judicial salaries to get more judge Ys in the mix. The substitutability theory carries
the day.

       The pool measure overcomes a number of failings of the direct comparison
approach. First, the direct comparison approach measures and compares individuals, not
candidate pools. But the salary matters theory suggests that the preference profile of the
pool members matters too. Second, the direct comparison approach assumes that the
opportunity cost for candidates coming from outside the private sector equals their lost
law firm salary in an average law firm in that region of the country. The pool approach
discards this assumption. Instead, it focuses on the strength of the private sector pool that
the government lawyer, prosecutor, state court or district court judge competes against.
And this might produce different results.

        Imagine that judicial salaries are so low or, the same thing, private sector salaries
are so high that no one outside the public sector will take a job as a federal circuit judge.
The pool is all public sector lawyers, state court judges, and district court judges. As
such, the selected nominee will be a public sector lawyer or judge. More importantly,
this nominee will not have faced any private sector competition for the slot. The
hypothesis is that a public sector nominee who faces substantial private sector
competition for a slot will look different from a public sector nominee who doesn’t.
Why? Well, for one reason, the competition means the president sees more candidates
with different backgrounds and perhaps would be more inclined to pick the public sector
candidate with weaker preferences over leisure, power, status, and/or influence.

        But this is only one theory, salary matters. As noted above, the competing theory
suggests that composition of the pool doesn’t make a difference. The same people or
type of people might be selected independent of the pool. In that case, there is no reason
to raise salaries to attract more private sector candidates into the nominee pool.

III.    Opportunity Cost Measure

                 A. Direct Comparison

       The opportunity cost for a federal nominee is her wages forgone in the next best
opportunity. I construct this measure for 261 federal circuit judges appointed since 1974.
As noted above, law firm salaries in the region at the date of the confirmation serve as the
relevant benchmark. Of course, many judges come from academia, government

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

positions, and other judgeships. For these judges, any lost salary at the time of
appointment is quite small. Their current salaries and the judicial salaries don’t differ
that much. For these judges, I nevertheless use the lost law firm wages as the relevant
opportunity cost.21 I then control for prior experience to account for systematic
differences in the preferences of lawyers coming from government service or academia.
In other words, the very fact that these judges come from government service or
academia might reveal something about their preferences. Government lawyers, lower-
court judges, and academics might, for example, prefer leisure more than their private
sector counterparts. And so, holding opportunity cost constant, a judge coming from one
of these positions might, for example, write opinions less swiftly than a judge coming
directly from the private bar. The control variable captures these potential differences in

        Given these assumptions, the optimal lost wages calculation for a person
considering the bench runs like this: Calculate, at the time of the appointment, the
number of years the candidate would likely remain at the firm if they didn’t take the job
on the bench. Second, determine the likely private sector compensation for each of those
years, considering increasing compensation due to increased seniority in the firm. Third,
estimate how much law firm compensation in general is likely to increase during that
time. Fourth, discount the total amount back to present value using the appropriate real
discount rate. Fifth, estimate the anticipated judicial wage for the years of service on the
bench and discount this amount back to present value. Sixth, to get the net cost of taking
the judgeship – the financial sacrifice made -- subtract from the present value of the lost
law firm wages the present value of the anticipated judicial salary. Seventh, adjust this
net sacrifice for geographic cost of living differences, revealing, in effect, the purchasing
power of the wages forgone. Finally, place that lost purchasing power into constant
dollars, enabling the comparison of the financial sacrifices made by judges appointed at
different times.

        Data limitations make the optimal calculation impossible. That said, the data
available in the “Law Firm Survey of Law Firm Economics” gives a rough answer. The
publishers of this survey – Altman & Weil, Inc. – collect data on law firm compensation.
The survey reflects self-reports by law firms throughout the country. The number of
reporting lawyers and firms varies by year, but the numbers are fairly substantial. In

   Presumably, any government lawyer, judge, or academic considered for a circuit court clerkship is
talented enough to be a law firm partner – if they so choose – at an average size firm in that region of the
country. The evidence supports this assumption. Prosecutors move into law firms. See Richard T. Boylan,
What Do Prosecutors Maximize? Evidence from the Careers of U.S. Attorneys, 7 Am. L. & Econ. Rev.
379, 383 (2005) (“Of the 570 U.S. attorneys in the study, . . . , 19.65% took a position in a large private
practice, and 39.12% took a position in a small private practice.”). State court judges rely on contacts to
secure positions in local firms. Cite. District court judges become partners in law firms. See Emily Field
Van Tassel, Why Judges Resign: Influences on Federal Judicial Service, 1789 to 1992, Appendix Index 3
1993 (revealing that many judges left for private practice between 1789 and 1992); Joe Mandak, “More
Judges Leaving for Private Practice, Associated Press Account, Febuary 2004 (“Joe Kendall left his U.S.
District Court job in Texas when he was 47 after 10 years on bench. He told The Third Branch, the
newsletter for the federal courts system, that with two soon-to-be college-aged children, he couldn't afford
not to sell his skills to the private sector.”) Talented academics become of counsel at firms in their area.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

2005, for example, the survey includes 7,516 associates and 9,704 partners, working in
340 U.S. law firms.22 The survey breaks down lawyer compensation by region, by state,
by year admitted to the bar, by size of the firm, and by title (associates, partners, of
counsel, etc).

         I collected the volumes of the survey, starting in 1974. An example best
illustrates the construction of the net cost of taking a circuit court judgeship figure. Take
Judge Sprouse of the Fourth Circuit Court of Appeals. In 1979, President Carter
appointed Judge Sprouse. His chambers are located in Charleston, West Virginia. At the
time of his appointment, Judge Sprouse was 56 years old. Judge Sprouse graduated from
law school in 1949. As a result, he was most likely admitted to the bar in 1950.

        What would Judge Sprouse have earned had in stayed in practice rather than
taking the bench? To answer this question, start with the 1979 “Survey of Law Firm
Economics.” The survey provides data on the average partner salary for firms in the
South Atlantic – a region that includes West Virginia. The average partner in a law firm
who graduated from law school in 1950 made $97,578 that year. But that is only one
year’s worth of data: Sprouse’s lost salary from the first year on the bench. Add to this
amount what a lawyer with a year more seniority made at the firm, that is, a lawyer who
graduated from law school in 1949. This two-year sum represents the anticipated loss
from two years on the bench (accounting for increases in pay due to an extra year of
seniority at the firm). Do the same calculation for seven more years, assuming that
Lawyer Sprouse retires at 65. Discount the total back to present value using the real rate
of interest of three percent.23 This amount is a rough approximation of the present value
of Judge Sprouse’s wages forgone at the time of appointment -- $ 868,319.

        The next step is to compute Judge Sprouse’s anticipated judicial salary. In 1979,
a federal circuit judge made $ 65,000. Assuming that the real judicial wage would not
increase, Judge Sprouse could anticipate bringing home nine years worth of that salary.24

   The Survey of Law Firm Economics 2005 at 5. Unfortunately, the Law Firm Survey does not necessarily
track the same lawyers and law firms over time. Altman Weil sends the survey to law firms that have
contact with the company, specifically firms that have purchased their consulting services or subscribe to
their newsletter. Id. at 11. In addition, Altman Weil also sends forms to law firm participants in the survey
from prior years. Altman Weil does not report how many firms report every year or how many new firms
have been added to the database. That said, the information in each year provides the best available
snapshot of the national legal market for that year.
   Picking the appropriate rate at which to discount future earnings is more art than science. The analysis
uses three percent as the appropriate real rate. See [cites on discounting future earnings]. I did the same
analysis with discount rates ranging from 1 to 6 percent. The statistical results all still hold. Note that
inflation is not included in the growth rate of the law firm wages. As such, the real rate of interest is used
to discount back to present value, keeping the treatment of inflation the same in numerator and denominator
of the lost earnings equation. See O’Shea v. Riverway Towing Co., 677 F.2d 1194, 1199-1201 (7th Cir.
1982)(Posner, J.).
   As noted in section IIA, there were no significant judicial raises between 1974 and 1990. After the forty
percent pay bump in 1991, judicial salaries have remained fairly stagnant in real terms. The assumption
that the real judicial wage will not change is stark, but not that unrealistic given past trends in
compensation. That said, I replicated all the results, assuming a variety of growth rates in judicial salaries
(anywhere from 2 percent to 7 percent). The results all hold under these various assumptions.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

Subtracting the present value of the stream of payments from the judicial salary from the
present value of his stream of lost law firm wages gives $ 272,221.92.

        To take account of geographic differences in cost of living, I adjust the net cost of
the judgeship for relative purchasing power using the ACCRA index.25 This index
compares the cost of living in different cities at different points in time. The Charleston,
WV index is 99.3. This means that the cost of living in Charleston is roughly 99.3
percent of the cost of living in the average city in the United States. After making this
geographic cost of living adjustment, the purchasing power Judge Sprouse lost by taking
the bench is $ 274,140.90 in 1979 dollars. The consumer price index (CPI) is used to
translate this amount into inflation-adjusted 2004 dollars. After this adjustment, Judge
Sprouse gave up roughly $ 949,120.79 to take the bench.

      Table 1 provides summary statistics on the net cost of taking the judgeship

         [Insert Table 1 here]

        The descriptive statistics reveal a couple of under-appreciated points in the
judicial salary debate. First, the debate focuses, naturally, on a comparison of annual
judicial salary versus annual salary in the private firms or academia, with a heavy focus
on the large and ever-increasing first year associate salaries in the major markets. There
is a shock value and a rhetorical value to this approach. In 2006, first year associates at
New York City law firms made almost as much as circuit court judges. How could a
judge be of the same value as a first year associate? For a person considering the bench,
this annual comparison is immaterial. It ignores differences in cost of living. The
judicial salary doesn’t vary by the location. Law firm salaries generally do. By
comparing judicial pay for a judge sitting in, say, Omaha, Nebraska with law firm
salaries in Washington, DC or New York City, misses the point that a dollar buys a lot
more in Omaha. Because few judges ever leave the bench, the use of annual comparisons
also hides differences in lost lifetime earnings – the true wages forgone.

        Second, judges appointed early in life had the highest net cost of taking the
judgeship. The four judges who made the biggest sacrifice -- Judges William Pryor, Jerry
Smith, Lavenski Smith, and Karen Henderson -- were all appointed in their early or mid-
forties. The extra years of lost earnings swamp differences in geographic cost of living
and differences in real salaries.

          Nominal judicial wages have of course increased over time, from $ 42,500 in 1974 to $171,00 in
2005. Inflationary pressures drove much of this judicial wage growth, albeit not enough to make the
judicial wage constant in real terms. As with lost law firm salaries, in computing the present value of the
lost judicial wage, I didn’t bump the wage up to account for inflationary increases. At the same time, the
real, not nominal, discount rate is used to discount back to present value. The treatment of inflation is thus
the same in the numerator and the denominator of the lost judicial salary computation.
   Cite explaining ACCRA.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

        Third, the net cost of taking the bench has not increased substantially over time.
There is a lot of variance in each year, but only a small upward trend. Although law firm
salaries have increased in real terms, the age of appointment has bounced around.
President Reagan appointed relatively young federal judges (average age 49). President
George W. Bush appointed some older judges and some younger judges (average age
52). Comparing the two sets on the Fifth Circuit, for example, shows that the some
Reagan appointees sacrificed more purchasing power than some George H. Bush
appointees, despite the upward real trend in salaries in Texas, Mississippi, and Louisiana.
The data does not support the idea -- implicit in the arguments by proponents of higher
salaries -- that appointees from ten or twenty years ago paid a small price to take the
bench, whereas appointees today pay a hefty price. The truth is that the lost purchasing
power depends on the judge’s age and his geographic cost of living, not just the absolute
salary in the private law sector. Every judge appointed before the age of forty-five took a
serious financial hit to take the bench. Again, annual comparisons to the salaries of
lawyers in large market firms, law school professors or law school deans are not
revealing. If low relative judicial salaries are a problem, they always were a problem.
That hasn’t changed much over time.

        B.       Pool Comparisons

         Pool comparisons require a measure of the strength of the pool. To do this, I
constructed the net cost for the typical 49 year old lawyer in a region at the time of
appointment of the judge. To wade into the nominee pool, this typical lawyer would
have to give up sixteen years of law firm income, adjusted for increased seniority in the
firm. As with the judge-to judge comparison, from this figure the discounted value of
the likely judicial wage is deducted. The net cost figure is then adjusted for geographic
cost of living differences and inflation. The end result is a measure of the “typical” loss
in purchasing power for a lawyer who decided to take a judicial appointment at that time
in that region of the country. If the typical lawyer gave up relatively little purchasing
power, lost income shouldn’t be a large barrier to entry into the judicial nomination
process. In this sense, lost purchasing power coextends with the size of the pool. With a
small financial sacrifice for the average lawyer in that region at that point in time, more
people should be willing to consider the judgeship. Raising judicial salaries in this sense
enlarges the candidate pool.

        Table 2 provides descriptive statistics on the net costs for the pools the president
selected from.

Insert Table 2 here

        A couple of points are worth noting. First, the DC Circuit judges are not included
in the pool-to-pool comparisons. Since the president selects these judges from the

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

national market, there isn’t a natural regional pool. As such, it was hard to decide the
relevant region that a “typical” DC circuit judge might come from.

        Second, the president looks to specific states for the regional circuit appointments.
To capture this fact, the strength of the pool is measured by state. The judge-to-judge
comparisons accounted for geographic cost of living differences by assessing the relative
costliness of the city where a specific judge lived. The pool comparisons are adjusted for
geographic cost of living differences by averaging the geographic cost of living index
statewide. For example, the cost of living adjustment for Judge Sprouse was 99.8,
reflecting the relative costliness of Charleston, West Virginia. For the pool Judge
Sprouse competed against, I averaged the cost of living in all of West Virginia, giving an
index of 98.3.

         Third, on average, the financial barriers to entry on the circuit courts are the most
substantial for the Fifth Circuit and weakest on the Tenth Circuit. Some might be
surprised that appointment on the Ninth and Second Circuits didn’t require the greatest
financial sacrifice. This is because the pool members in the Ninth and Second Circuits
had dramatically higher costs of living. As a result, each dollar lost from taking the
judgeship didn’t cost the candidate that much extra consumption. The higher cost of
living, in other words, swamped the relatively higher law firm salaries in those regions of
the country.

IV.     Would the Circuit Courts Look Any Different with Higher Judicial Salaries?

        This section tests several hypotheses concerning the link between higher salaries
and judicial performance. The three hypotheses are: (1) paying circuit judges more
creates a less ideological judiciary; (2) paying circuit judges more creates a harder
working judiciary; and (3) paying circuit judges more creates a judiciary less concerned
with its own influence.

       Under the “salary matters” theory, if the primary non-pecuniary reason for taking
a judgeship is the power to impose policy preferences, the data should support the first
hypothesis. If the primary non-pecuniary reason for taking the bench is a desire to
engage in leisure, the second hypothesis should hold. Finally, if the primary non-
pecuniary reason for taking the bench is a desire to have influence or prestige, the data
should support the third hypothesis. With a strong preference for power, leisure, or
influence, the intensity of the preference – and the exercise thereof -- should correlate
with the amount of money the judge gave up. The overall effect of the strong preference
can then be unmasked by comparing the actual performance of judges who gave up a lot
of money against those that gave up a little money.

         Under the “substitutability” theory, paying judges higher salaries should have no
effect whatsoever. As noted above, the evidence supports this theory if judges with large
spreads and small spreads between their judicial salaries and private sector salaries
perform the same on the judicial output measures as judges. More formally, if this theory
is right, we should fail to reject each of the three hypotheses set forth above.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

        A.       Hypothesis One: Paying Circuit Judges More Creates a Less Ideological

        Measuring judicial ideology is a tricky business. The common perception is that
some judges are conservative, like, say, Judge Edith Jones of the Fifth Circuit, while
other judges are liberal, like, say, Judge Stephen Reinhardt of the Ninth Circuit. But
what traits make Judge Jones conservative and Judge Reinhardt liberal? And, more to the
point, can those traits be quantified? In short, testing whether judicial pay impacts
judicial ideology requires some measure of ideology. I get at ideology two different
ways in this section. The first subsection considers whether judicial pay impacts judicial
voting in controversial cases. The thinking is that a more ideological judiciary will
engage in predictably partisan voting patterns across a variety of cases. The judge who is
a true conservative ideologue will always cast a conservative vote. The opposite holds
for the liberal ideologue. Under this measure, a more ideological judiciary consists of
republican appointees who more consistently cast conservative votes and democratic
appointees who more consistently cast liberal votes.

        The second subsection examines the relationship between judicial pay and citation
practices. Judges do more than vote; they write opinions. These opinions often cite
outside circuit judicial opinions to support their analysis. The thinking here is that a
judge’s ideology can be unmasked by closely examining citations to persuasive authority.
A judge who is a true ideologue would never find the reasoning of a judge of the other
political stripe persuasive and, hence, citable. In contrast, a judge open to all arguments
would cite judges from either party indiscriminately, only looking for judges who had the
best-reasoned outside circuit opinion. Under this measure, a more ideological judiciary
consists of judges who seldom, if ever, recognize the opinions of judges from the other
party as persuasive.

                 (1)      Voting Patterns in Controversial Cases

        The data on circuit court voting patterns comes from the Chicago Judge’s Project.
The project tracks recent published judicial decisions of the circuit courts in controversial
cases. The cases involve “abortion, capital punishment, the American with Disability
Act, takings, criminal appeals the Contracts clause, affirmative action, Title VII race
discrimination with African American plaintiffs, sex discrimination, cases in which
plaintiff sought to pierce the corporate veil, industry challenges to environmental
regulations, and federalism challenges to congressional enactments under the Commerce

   Cass R. Sunstein, David Schkade, & Lisa Michelle Ellman, Ideological Voting on Federal Court of
Appeals: A Preliminary Investigation, 90 Va. L. Rev. 301, 304 (2004). For a more complete discussion of
the dataset, see Cass R. Sunstein, David Schkade, & Lisa Michelle Ellman, Are Judges Political? An
Empirical Investigation of the Federal Judiciary (Brookings Institution Press, 2006). The complete dataset
is available at

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

        The dataset includes 4958 decisions and 14,874 individual judicial votes. Each
judges’ vote is coded “liberal” or “conservative.” The labels are imprecise. They do
track, however, common notions of liberal and conservative jurisprudence. For example,
a liberal vote in a sex discrimination case is a vote for the employee; a conservative vote
is a vote for the employer. The database also includes information about whether the
judge dissented and the political make-up of the panel of judges, that is, whether the
panel members were appointed by a republican or democratic presidents.

       As is, the database is too broad for my inquiry. It includes votes by district court
judges sitting by designation and circuit judges appointed before 1974 for whom
opportunity cost data is unavailable. Truncating the dataset left 8699 judicial votes.

         To see whether judicial pay impacts voting patterns, any analysis must control for
other factors that influence judicial votes. One of the most important factors is the
politics behind the nomination process. No matter the level of judicial pay, a judge
appointed by a republican president facing a republican-controlled senate is likely to cast
more conservative votes than a judge appointed by a democratic president facing
democratically-controlled senate. Politics, in other words, play a large role in what kind
of judge reaches the bench. Presidential politics dictate the type of nominee sent to the
senate. The politics of the senate dictate whether that nominee is confirmed. Indeed,
because of the blue slip process and presidential courtesy to senators from the nominee’s
home-state, the politics of the senators from the nominee’s home state occasionally carry
a lot of weight in the confirmation process.27

        Controlling for politics is hard. Just using the political party of the appointing
president as a proxy for the ideology of the appointed judge misses much of the process
behind nominations. Although most republican presidents are conservative, not all
republican presidents are equally conservative. The same holds for democrats.
Furthermore, because of dynamics between the president and senate, a republican
president facing democrat senators from the nominee’s home state would be able to push
through a different judge than a republican president facing republican home-state
senators. Fortunately, Micheal Giles, Virginia Hettinger, and Todd Peppers have
constructed a measure of the ideology of the appointing president and confirming senate,
controlling for the possibility of senatorial courtesy and the blue slip process.28

       Giles et al measure the ideology of the appointing president based on his votes on
various pieces of legislation. In the political science literature, this is called the common
space score.29 The same type of score measures the ideology of the important senators.
The index weights each of these factors and comes up with a measure the ideology of the
nominee. The index runs from -1 to 1. Absent senatorial courtesy, the nominee’s
ideological score equals the common space score of the appointing president. The more

   Cite Gerhardt’s book, some pol-sci stuff.
   See Micheal W. Giles et al. Picking Federal Judges: A Note on Policy and Partisan Selection Agendas,
54 Pol. Res. Quart. 623 (1998); Michael W. Giles et al., Measuring the Preferences of Federal Judges:
Alternatives to Party of the Appointing President, (Working draft)
   Explain the common space score.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

conservative the president the closer the index is to 1. If there was senatorial courtesy for
the nomination, the ideological score weights the common space scores of the president
and the home state senators. Again, the closer the index is to 1, the more conservative the
weighted common space scores.

        Combining the data from the Chicago Judges Project with the Giles et al. measure
reveals a consistency between the two datasets, demonstrated in Table 3. The dependent
variable is the probability that the judge casts a liberal vote. The independent variables
include the Giles et al measure of the confirmation process (“selpref”) and circuit dummy
variables to control for unobserved differences across circuits. The Giles et al. measure is
negative and highly statistically significant. The more conservative the players in the
political process of the nomination, the more likely the judge will be to cast a
conservative vote in a controversial case.

[Insert Table 3 here]

        I now turn to the hypothesis that higher judicial pay would lead to a less
ideological judiciary. Tables 4 and 5 present the result of the judge-to-judge
comparisons. I first divided the sample into votes by democratic appointees and votes by
republican appointees. The dependent variable is the probability the judge voted liberal
in the votes collected by the Chicago Judge’s Project. If the hypothesis is correct, the
sign of the coefficient for net cost variable (“NETCOST”) should be positive and
significant for democratic appointees and negative and significant for republican
appointees. Included as controls are (1) if available, the judge’s net worth at the time of
appointment, adjusted for inflation and geographic cost of living differences; (2) circuit
court dummy variables (with the DC circuit left out to the constant term); (3) a dummy
variable for whether the appointed judge came from private practice; (4) the Giles et al
measure of the ideology of the nomination process; (5) the nominee’s age at the time of
appointment; and (6) the nominee’s gender.

[Insert Tables Four and Five Here]

       I cannot reject the hypothesis that the size of a judge’s financial sacrifice has no
impact on voting patterns. This is true in both the subsample, for which net worth data is
available, and the full sample, which includes all judges even if the net worth data is

      Table 6 presents the result of the pool-to-pool comparison. The net cost variable
(“NETCOSTPOOL”) is not statistically significant for either democratic or republican

[Insert Table 6 here]

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

         In sum, raising salaries lowers the financial cost of taking a job on the circuit
courts. Yet the available data suggests that lowering this cost would not impact judicial
voting patterns in controversial cases. The sample size is relatively large. If there was an
effect, the analysis should reveal it. This “non-result” supports the substitutability theory:
At current and historic salary levels, largely interchangeable candidates flood the
candidate pool. As a result, low pay doesn’t lead to the appointment of more partisan

        (2) Uncovering Ideology through Citation Practices in Opinion Writing

        Voting patterns is the most studied metric of judicial ideology.30 Recently,
however, Mitu Gulati and Steve Choi looked at judicial ideology through different lens,
citation practices.31 Gulati and Choi collected data on judicial opinions rendered between
January 1, 1998 and December 31, 1999. In so doing, they amassed data on the citation
practices of 98 circuit judges. In particular, they examine who cites whom as persuasive
authority. More specifically, they look and see which judicial decisions from outside the
circuit a judge cites as persuasive authority. As noted above, Gulati and Choi think that a
true ideologue would not be inclined to cite an opinion by a judge of a different political
party. The reason is that the ideologue is never persuaded by arguments from the other
side of the political spectrum.

        Gulati and Choi find evidence of citation bias. Specifically, they find that judges
tend to cite opinions from judges of the same political stripe, especially in “hot button”
cases, such as civil rights and campaign finance. They also find that dissent exacerbates
bias. Dissenting judges and judges writing majority opinions in the face of dissent
engage in more biased citation practices. The bias gets a further boost if presidents of
opposing parties appointed the majority judges and the dissenting judge.

        Choi and Gulati define citation bias as follows: They first construct the mean
fraction of cites for a judge’s opinions to outside circuit judges from the opposite political
party. If, for example, a judge cites to outside circuit judges of the same political stripe
75 percent of the time (averaged over all his opinions), the mean fraction of cites to
judges of the opposite party would be 25 percent. Second, Choi and Gulati need to
control for the pool of potentially citable opinions. If most judges are republican-
appointees, most outside circuit citations will be to republican-appointed judges. In this
case, a failure of a republican judge to cite democrat appointees wouldn’t indicate bias.

   See, for example, Donald R. Songer, The Policy Consequences of Senate Involvement in the Selection of
Judges in the United States Courts of Appeals, 35 W. Pol. Q. 107 (1982); Donald R. Songer & Martha
Humphries, Assessing the Impact of Presidential and Home State Influences on Judicial Decisionmaking in
the United States Courts of Appeals, 55 Pol. Res. Q. 299 (2002); Frank B Cross, Decisionmaking in the
U.S. Circuit Courts of Appeals, 91 Cal. L. Rev. 1457 (2003); Richard L. Revesz, Environmental
Regulation, Ideology, and the D.C. Circuit, 83 Va. L. Rev. 1717 (1997).
   Stephen J. Choi & G. Mitu Gulati, Bias in Judicial Citations: A New Window into the Behavior of
Judges? NYU Law and Economics Working Paper No. 06-21 (2007); Stephen J. Choi & G Mitu Gulati,
Ranking Judges According to Citation Bias (as a Means to Reduce Bias), __ Notre D. L. Rev. __
(forthcoming 2007)

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

Instead, it would reflect the lack of opinions authored by democrat appointees in the
citable pool. To control for this, Choi and Gulati construct a mean fraction of democrat-
appointee and republican-appointee opinions in the pool. Citation bias is the distance
between the mean fraction of opposite party cites a judge makes and the mean fraction of
republican opinions (for democrats) or democrat opinions (for republicans) in the pool.
The closer the distance is to zero, the less prevalent the citation bias is.

        If judges that give up more purchasing power are more ideological, low judicial
salaries should result in increased citation bias. To test this hypothesis, I regressed the
citation bias measure from the Gulati/Choi dataset against (1) the Giles et al measure of
politics; (2) the net cost measures; (3) whether the judge came from private practice; (4)
the age at the time of appointment; (5) gender; and (6) circuit dummy variables.

        [Insert Table 7 here]

        Table 7 reports the results. The net cost measure is statistically insignificant both
in the judge-to-judge and pool-to-pool comparisons. At least with the data available, one
cannot reject the hypothesis that low judicial salaries have no impact on whom a judge
cites as persuasive authority. At least on this measure, there is little evidence that low
relative judicial salaries result in a judiciary more prone to ideological thinking.

B:      Hypothesis Two: Paying Circuit Judges More Creates a Harder Working Judiciary

        Testing whether increased judicial pay would result in a harder working judiciary
requires measuring the “work effort” of circuit judges. Actual effort is unobservable,
however. We don’t know how many hours each judge works, the number of weekends
she takes off, etc. Proxies are needed. Specifically, I need quantifiable measures of
judicial output that correlate with judicial effort level. The next three subsections look at
the relationship between judicial pay and three such proxies: (1) dissent rates in
controversial cases; (2) how long it takes a judge to file a published opinion after the
notice of appeal is filed; (3) the ratio of published to unpublished opinions a judge

        (1) Dissents in Controversial Cases

       Dissenting takes work. It requires separate drafting, disagreeing publicly with the
majority of the panel, and finding and articulating the flaws in the majority opinion.
Dissent also imposes more work on the majority, who often alter the majority opinion to
address the points raised by the dissent. It imposes other costs, too. A dissenting
colleague might be seen as less collegial or as someone unwilling to find common
ground. Despite its cost, however, dissent has value. Dissents sharpen the reasoning of
the majority opinion. Circuit court dissent might convey important information to the
Supreme Court about the state of the law, encouraging the grant of certiorari.32 Dissent

  Andrew F. Daughety & Jennifer F. Reinganum, Speaking Up: A Model of Judicial Dissent and
Discretionary Review, 14 Sup. Ct. Econ. Rev. 1 (2006)

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

can also influence the way the majority opinion is viewed by other circuit and district
courts.33 Finally, dissent can serve as a form of judicial self-expression.

        Most of the benefits of dissent accrue to other judges in the circuit or people
outside the judiciary. Because the individual judge bears the cost of dissent and much of
the benefits flow to others, one might suspect that a judge inclined toward leisure would
write fewer dissents. The data bears this out.

        Table 8 presents the dissent results. The Chicago Judge’s Project provides the
dependent variable: the probability of a judge writing a dissent in a controversial case.
The independent variables include (1) the net cost variable; (2) the Giles et al measure of
the ideology make-up of those in the nomination process; (3) the age at appointment; (4)
the judge’s gender; (5) whether the judge came from private practice; and (6) circuit
dummies variables. For the judge-to-judge comparison, the sign of the coefficient on net
cost (NETCOST) is negative and statistically significant for the entire sample and for the
subsample where net worth data is available. The sign on the coefficient on net cost
(NETCOSTPOOL) in the pool-to-pool comparison is not statistically significant.
Although the coefficient on NETCOST is statistically significant, its magnitude is tiny.

        [Insert Table 8 here]

        This finding suggests that poorly paid judges – those with a large spread between
judicial salaries and private sector salaries -- dissent slightly less often. Because higher
judicial salaries results in fewer poorly paid judges on the bench, raising salaries should
lead to a few more dissents. This finding is consistent with the view that low judicial pay
attracts some nominees who place a premium on leisure and act as such. Higher judicial
pay will thus result in fewer leisure-seeking judges on the bench and a slightly harder
working judiciary overall.

        That said, the results from table 7 are consistent with another story: a judiciary
composed of judges trying to find common ground. In other words, it is not just the lazy
judge who writes fewer dissents, but also, perhaps, the more considerate judge. The
dissent results support both stories. For this reason, the following two subsections
consider other proxies for judicial work effort, trying to disentangle to two possibilities.

        (2) Time it Takes to Render a Published Opinion, controlling for caseload

        (3) Ratio of Published to Unpublished Decisions, controlling for caseload

  Indraneel Sur, How Far Do Voices Carry: Dissents from Denial of Rehearing En Banc, 2006 Wis. L.
Rev. 1315, 1346.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

D:      Hypothesis Three: Paying Circuit Judges More Creates a Judiciary Less
        Motivated by its Own Influence

         Outside circuit citations roughly capture a judge’s influence. Rules of precedent
dictate inside circuit citations; circuit precedent must be followed and cited. By contrast,
judges cite outside circuit opinions as persuasive authority to bolster the arguments in
their opinions. True, occasionally opinions criticize or distinguish outside circuit
opinions. But the need for such treatment demonstrates its impact. After all, an opinion
that is ignored is less influential than an opinion that the judge feels he has to deal with.

        A judge who greatly valued his own influence would write more published
opinions and try to ensure each opinion attracted more outside citations. The idea is that
this judge – the influence maximizer—would write more opinions that “sell” in the
opinion citation market. Perhaps the influence maximizer would write shorter opinions;
delegate less opinion writing to clerks; or spend more time ensuring that the reasoning of
the opinion was sound and persuasive. In contrast to the judge the valued leisure, the
judge that valued influence would write more opinions and spend a lot of time on each

        So, the salary theory goes, if low pay leads to a judiciary that places a higher
value on influence, judges that gave up a lot of money to take the bench should be more
influential than judges who gave up a little bit of money.

        To test this claim, I use citation data collected by William Landes, Larry Lessig,
and Mike Solimine.34 They gathered data for 205 federal circuit judges on the bench in
1992. They look at the number of citations to the opinions authored by these judges.
To measure impact, they consider two different models of outside circuit citation. First,
they construct a model of total influence. In this model, Landes et al measure the raw
number of citations to a judge’s opinions and then control for, among other things, the
length of judicial tenure (obviously a judge who has been around longer will have more
citations). The second model – average influence – measures the number of citations per
opinion, controlling for other factors. A judge that scores well in average influence but
low in total influence writes fewer opinions, but each one is a gem. The opposite is true
for a judge that scores well in total influence and low in average influence. This judge
floods the market with opinions, each one garnering relatively modest outside attention.

        Landes et al. then measure judicial influence, above what is predicted by his
tenure, status, and various other control variables. For example, in terms of total
influence, the estimated coefficient for Judge Posner is 4.41. This means that Judge
Posner’s influence is four percent higher than predicted by his tenure, status and other

        Tables 9 and 10 report the results of the total influence and average influence
regressions respectively. If low salaries attracted more people interested in their own

  William M . Landes, Lawrence Lessig, and Michael E. Solimine, Judicial Influence: A Citation Analysis
of Federal Court of Appeals Judges, 27 J. Leg. Stud. 271 (1998).

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

influence, the coefficient on lost purchasing power should be positive and significant. In
the total influence regressions the coefficients on “NETCOST” and “NETCOSTPOOl”
are insignificant. In the average influence regressions, the coefficients are positive and
significant, but the magnitude of the impact is tiny. The take away is that the citation
data provides some limited support for the salary matters theory: higher wages would
result in slightly lower quality opinions. That is to say, judges paid poorly as against
their next best opportunity produce slightly better opinions. At the same time, the effect
is small and only statistically significant for one measure of opinion quality.

[Insert Tables 9 and 10 here]

IV:     Objections

       There are several objections to the foregoing discussion. This short section deals
with those objections.

        First, the opportunity cost measure may be imprecise. For example, Judge
Easterbrook’s opportunity cost equals the anticipated future earnings of the average law
firm partner in the Midwest in 1985, the year of his appointment. Given his
qualifications had Judge Easterbrook decided to leave academia and enter private
practice, he would have probably received a job at a highly prestigious Chicago-based
firm. This firm would have paid much more than the average firm in the Midwest.
Indeed, as a principal in LexEcon, the litigation consulting group associated with several
members of the University of Chicago faculty, Easterbrook might have made even more
money than at the highest paying Chicago law firm.35 In short, the net cost measure used
underestimates Judge Easterbrook’s true loss in purchasing power from taking the bench.
For other judges, the opportunity cost measure likely overestimates their earning potential
in the private sector. This is a pitfall in the data.

         The other leading measure of law firm salaries – the American Lawyer 100 and
the American Lawyer 200 – reports salaries from the prominent national firms only. But
for a judge like Judge Easterbrook, salary in a prominent firm is a closer measure of his
true opportunity cost. While perhaps getting a clearer picture of Judge Easterbrook’s lost
earnings, the Am Law 100 and Am Law 200 have other problems. Unlike the Law Firm
Survey, the Am Law 100 and Am Law 200 do not report anticipated increases in
compensation due to increased seniority in the firm, an important part of the net cost
calculation. Second, the Am Law 100 and Am Law 200 do not provide information for
many of the judges on the federal bench. For example, there are simply no Am Law 100
or Am Law 200 firms operating in Cheyenne, Wyoming (Judge O’Brien) or Columbia,
South Carolina (Judge Hamilton). The Law Firm Survey provides data for these judges;
it is the most comprehensive overview of the national legal market. And to boot, it has

  Compare George Anders, An Economist's Courtroom Bonanza, Wall St. J., March 19, 2007, at A1
(noting that David Speece, who runs a litigation consulting competitor to LexEcon, estimates his consulting
earnings to be at least $50 million).

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

also been published over a longer period of time than any other law firm salary

        The analysis also assumes that Judge Easterbrook would have stayed in the
Midwest. But why couldn’t Lawyer Easterbrook move to New York City? If he didn’t
take the bench or remained an academic, Judge Easterbrook might have been a partner at
Wachtell, Lipton, Rosen & Katz, making even more cash.37 The regional restriction, in
other words, seems too harsh. For a judge, like Judge Easterbrook, the relevant
opportunity cost is the highest salary in the biggest legal market. A regional assumption,
however, makes sense for the vast majority of circuit court appointees. A quick analysis
shows that most appointees have lived and worked in the same community for a number
of years before taking the bench.38 To assume that these appointees would move to a
totally different legal market is a stretch. For this reason, I use a regional measure of law
firm partnership salaries as the relevant benchmark.

        The data is imprecise in another way, too. I assume that the appointees would
have made the “average” salary for a law firm partner of their age. Circuit judges might
be above-average lawyers, not average lawyers. The average partner salary, then, might
underestimate their true opportunity cost. If, as is plausible, the average salary for a law
firm partner in a region strongly correlates with the law firm salary for the above-average
lawyer, the analysis still works. The variance in the average partnership salary tracks the
variance in the salary for the super-lawyer. But this doesn’t capture that some appointees
might be better law firm partners (and hence give up more cash) than other appointees.
To control for this, I ran all the statistical tests with a dummy variable for whether the law
school the judges attended ranked among the top ten. The thinking was these appointees
might systematically give up more money than appointees for lower ranked schools; they
might be the more talented law firm lawyers. The dummy variable captures this effect.39
All the results were the same.

        In short, my response to the objection about the precision of the data is this: it is
the best I can do. It is far from perfect. But it is better than any of the alternatives.

       The second potential objection is that, for many measures of judicial performance,
the impact of differences in the cost of taking the judgeship – and, by construction,
changes in judicial salaries -- is statistically indistinguishable from zero. Why might this
   The American Lawyer first published the AmLaw 100 in 1993 and the AmLaw 200 in 1999. The
National Association of Law Placement (NALP) is the other common source of lawyer salary information.
While more geographically comprehensive than the American Lawyer Series, the NALP data suffers a
different flaw. NALP reports first year associate salaries only. See Nat'l Ass'n for Law Placement, 2006-
2007 NALP Directory of Legal Employers (2006). Obviously, a comparison to first year associate salaries
would understate the opportunity cost for a seasoned lawyer deciding to take the federal bench.
   In 2005, the AmLaw 100 reported that a partner at Wachtell, Lipton, Rosen & Katz made $3.79 million,
the most of any law firm in the country.
   The dummy variable “private practice” also deals in part with this objection. If lawyers that come from
private practice are systematically better law firm partners (and hence give up greater than average
earnings) than lawyers who come from government service or academia, then, this effect is embedded in
the dummy variable.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

be? To start, for some of the measures of judicial performance (citation bias
specifically), the sample size is relatively small, which will limit the power of the
statistical test.

         Moreover, suppose that some people who give up a lot of money are motivated by
power, others motivated by influence, others motivated by a desire for leisure, and still
others motivated by a call to public service. In this case, each of these people will
perform differently on the various measures of judicial performance. As a result, the
statistical tests will contain a lot of noise. The policy-motivated judge who cares little
about his influence will vote his policy preferences, but not invest energy in writing
opinions that other judges cite. The leisure-maximizing judge will not vote his policy
preferences, but instead take a long time to write his opinions. The influence-motivated
judge will write well-cited opinions, but not vote strictly along party lines. Because of
the variance in motivations, it is difficult for the statistical tests to tease out any one
“true” motivation and impute resulting actions of those with high financial sacrifice. This
results in a failure to reject the null hypothesis of zero effect of different opportunity

        In this event, the analysis shows that raising judicial salaries will bring in a
hodgepodge of folks with different motivations. These people will perform differently
along various metrics of judicial performance. Those different performances will largely
cancel each other out. The end result is relatively little overall impact on the functioning
of the circuit courts from higher salaries.

        Third, I have only measured that which is measurable – voting patterns, citation
counts, dissents, time to decision, etc. It does not immediately follow from the core
finding that the “measurables” wouldn’t change much that the judiciary wouldn’t look
different with higher salaries. I haven’t measured everything that goes into judicial
performance. And the things measured imperfectly correlate with the “true” judicial

         There are many other non-measured attributes that go into making a good judge.
Enhanced salaries might attract nicer people, people who value treating attorneys fairly
and with respect. Enhanced salaries might attract people committed to the judiciary as an
institution, people just trying to do a good job without baser motives. Allowing judicial
salaries to significantly lag behind private sector salaries might signal that circuit judging
is less valuable than run-of-the-mill lawyering. The weak signal could then impact how
the public feels about the judiciary. Along related lines, judges might get demoralized
because they make less than their clerks do in their first year after leaving the chambers.
Under this critique, relative pay is what matters to the judge, not absolute pay. With low
relative pay, judges feel undervalued and, as a result, do a worse job.

       Each of these objections is potentially valid. I don’t test them, but that doesn’t
mean they are unimportant. What this paper does is shift the burden to the proponents of
higher pay. The ball is in their court to show that the tiny effect of higher salaries on

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

measurable aspects of judicial performance is outweighed by the impact on softer
variables and concerns.

V.      Conclusion

        Chief Justice Roberts and his brethren have issued many statements about the
corrosive nature of low judicial salaries. The alarming rhetoric itself is telling: “low
judicial pay threatens the independence of the federal judiciary;” and “low judicial pay
has reached the level of a constitutional crisis.”

        This essay tests whether judicial salary impact judicial performance. The short
answer is no. The effect of low judicial pay is non-existent, when judicial pay is
measured against the next best employment opportunity for most circuit judges,
partnership in regional law firms. Low pay doesn’t impact voting patterns, citation
practices, speed of case disposition, or the ratio of published to non-published decisions.
Low pay does lead to slightly fewer dissents and slightly better opinions, as measured by
average outside circuit citations. While statistically significant, these effects are
nonetheless weak.

        Low judicial salaries might have a corrosive character, but the data suggests that
source of the corrosion lies outside judicial performance. What Justice Roberts says is
half right: low judicial salaries represent a barrier to entry for some candidates. But that
barrier is irrelevant because there are many other equally well suited candidates willing to
take the job at the prevailing wage. In the end, the analysis suggests that the federal
circuit courts operate the same no matter the wages paid to the federal judges.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                 Table 1
                                           Summary Statistics
                              Judge’s Net Cost of Taking The Bench (“NC”)

                     No obs          Average NC         Var NC              Min NC          Max NC
  Full Sample         261             1,148,157        656,837.8              0            3,603,807
        1               10            1,033,113         332,984             379,974        1,466,571
        2               26            782,441.7         394,567             209,344        1,708,354
        3               21            1,188,235         557,249                 0          2,474,461
        4               17            1,255,314         464,421             593,846        2,152,587
        5               27            1,365,941         912,313                 0          3,112,091
        6               26            1,117,551         511,608             246,836        2,104,809
        7               15            1,277,400         560,018             337,301        2,202,034
        8               18            1,037,208         690,710              32.570        3,113,461
        9               47             943,180          654,386                 0          2,715,934
       10               17            1,188,050         595,922             350,948        3,001,508
       11               19            1,518,708         759,703             177,210        3,603,807
       12               18            1,395,165         730,447             136,421        3,048,630

                                             Table 2
                                        Summary Statistics
                   Average Pool Member’s Net Cost of Taking the Bench (“NCPool”)

                    No obs          Avg NCPool         Var NCPool        Min NCPool      Max NCPool
 Full Sample         243             1,465,556          469,445           306,423         2,932,260
       1               10             1,435,677          317,257             955,168       1,955,180
       2               26             1,428,937          278,546             985,690       2,196,387
       3               21             1,543,771          280,206            1,107,419      2,225,188
       4               17             1,585,621          333,056            1,160,698      2,209,480
       5               27             1,791,910          612,301             437,773       2,715,140
       6               26             1,450,678          225,647            1,079,356      1,877,353
       7               15             1,484,65           268,213             958,831       1,969,727
       8               18             1,346,870          498,334             649,935       2,807,475
       9               47             1,287,493          630,860             306,423       2,932,260
      10               17             1,243,428          512,065             526,321       2,630,343
      11               19             1,630,690          331,987            1,204,928      2,216,145

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                Table 3

   Relationship Between Giles et al. Measure of the Confirmation Process and Judicial Voting Patterns

                                              Logit Model

Dependent Variable: Probability Judge Casts a Liberal Vote


selpref           -0.632
circdum1          -0.126
circdum2          -0.041
circdum3          0.219
circdum4          -0.408
circdum5          -0.605
circdum6          -0.287
circdum7          -0.653
circdum8          -0.679
circdum9          0.240
circdum10         -0.161
circdum11         -0.139
Constant          0.221

Observations       8669
Robust z statistics in parentheses
* significant at 5%; ** significant at 1%

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                       Table 4

                   Relationship Between Democratic Financial Sacrifice and Voting Patterns

                                                Logit Model

Dependent Variable: Probability Democratic-Appointee Casts a Liberal Vote

                                                   Model 1                 Model 2
                                                (Full Sample)       (Subsample with Net Worth)

NETCOST                                               2.49e-08                  9.28e-08
                                                      (0.27)                    (0.70)
Private Practice                                      -0.047                    0.031
                                                      (0.59)                    (0.22)
Selpref                                               0.150                     0.928
                                                      (0.47)                    (1.92)
Sex                                                   -0.074                    0.060
                                                      (0.92)                    (0.47)
Age                                                   -0.000                    -0.001
                                                      (0.05)                    (0.06)
circdum1                                              -0.139                    -0.635
                                                      (0.64)                    (0.63)
circdum2                                              -0.154                    -0.146
                                                      (0.79)                    (0.53)
circdum3                                              0.269                     0.189
                                                      (1.08)                    (0.56)
circdum4                                              -0.594                    -0.576
                                                      (2.68)**                  (2.15)*
circdum5                                              -0.799                    -1.062
                                                      (3.84)**                  (4.20)**
circdum6                                              -0.382                    -0.736
                                                      (2.09)*                   (2.82)**
circdum7                                              -0.747                    -0.609
                                                      (4.10)**                  (1.41)
circdum8                                              -0.539                    -0.531
                                                      (2.82)**                  (1.76)
circdum9                                              0.370                     0.377
                                                      (2.00)*                   (1.54)
circdum10                                             -0.159                    -0.642
                                                      (0.83)                    (1.94)
circdum11                                             -0.198                    -0.400
                                                      (0.90)                    (1.39)
Net Worth                                             N/A                       5.55e-09
Constant                                              0.601                     0.845
                                                      (1.24)                    (0.96)

Observations                                          3314                      1697
Robust z statistics in parentheses
* significant at 5%; ** significant at 1%

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                 Table 5

             Relationship Between Republican-Appointee Financial Sacrifice and Voting Patterns

                                               Logit Model

Dependent Variable: Probability Republican-Appointee Casts a Liberal Vote

                                                  Model 1                 Model 2
                                               (Full Sample)       (Subsample with Net Worth)


NETCOST                                              3.88e-08                  1.70e-07
                                                     (0.48)                    (1.39)
Private Practice                                     -0.116                    0.024
                                                     (1.56)                    (0.21)
selpref                                              -0.121                    -0.245
                                                     (0.70)                    (0.70)
Sex                                                  0.174                     0.244
                                                     (1.71)                    (1.77)
Age                                                  0.017                     0.031
                                                     (2.21)*                   (2.60)**
circdum1                                             -0.124                    -0.081
                                                     (0.71)                    (0.29)
circdum2                                             0.188                     -0.164
                                                     (0.99)                    (0.52)
circdum3                                             0.267                     0.277
                                                     (1.45)                    (1.00)
circdum4                                             -0.287                    -0.254
                                                     (1.59)                    (0.78)
circdum5                                             -0.575                    -0.908
                                                     (3.77)**                  (3.95)**
circdum6                                             -0.327                    -0.338
                                                     (1.99)*                   (1.30)
circdum7                                             -0.511                    -0.506
                                                     (3.49)**                  (2.22)*
circdum8                                             -0.668                    -0.733
                                                     (4.33)**                  (2.97)**
circdum9                                             0.055                     0.025
                                                     (0.32)                    (0.09)
circdum10                                            -0.139                    -0.376
                                                     (0.78)                    (1.40)
circdum11                                            -0.204                    -0.218
                                                     (1.28)                    (0.93)
NW RA 2004                                           N/A                       -4.03e-08
Constant                                             -0.877                    -1.695
                                                     (1.82)                    (2.09)*

Observations                                         5355                      2439
Robust z statistics in parentheses
* significant at 5%; ** significant at 1%

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                 Table 6

                         Relationship Between Pool Strength and Voting Patterns

                                               Logit Model

Dependent Variable: Probability Judge Casts a Liberal Vote

                                          Model (1)                                     Model (2)
                                   Democratic-Appointees                        Republican-Appointees

NETCOSTPOOL                                 1.26e-07                                    -8.93e-08
                                            (1.45)                                      (1.04)
selpref                                     -0.060                                      -0.035
                                            (0.22)                                      (0.21)
circdum1                                    -0.094                                      0.095
                                            (0.49)                                      (0.68)
circdum2                                    -0.063                                      0.366
                                            (0.42)                                      (2.36)*
circdum3                                    0.329                                       0.456
                                            (1.46)                                      (2.97)**
circdum4                                    -0.501                                      -0.178
                                            (2.90)**                                    (1.18)
circdum5                                    -0.662                                      -0.417
                                            (4.33)**                                    (3.33)**
circdum6                                    -0.281                                      -0.123
                                            (1.98)*                                     (0.94)
circdum7                                    -0.658                                      -0.359
                                            (4.46)**                                    (3.34)**
circdum8                                    -0.415                                      -0.495
                                            (2.72)**                                    (4.30)**
circdum9                                    0.505                                       0.196
                                            (3.60)**                                    (1.41)
circdum10                                   -0.031                                      -0.068
                                            (0.20)                                      (0.46)
Constant                                    0.223                                       -0.068
                                            (1.18)                                      (0.39)

Observations                               3314                                         5355
Robust z statistics in parentheses
* significant at 5%; ** significant at 1%
Votes by DC Circuit Judges not included; 11th circuit dummy left to the constant term

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                 Table 7

                        Relationship Between Financial Sacrifice and Citation Bias

                                               OLS Model

Dependent Variable: Extent of Citation Bias

                                                     Model (1)                           Model (2)
                                                     (Judge to Judge)                    (Pool to Pool)

NETCOST                                              -7.32e-10                           N/A
                                                     (0.06)                              N/A
Private Practice                                     0.008                               N/A
                                                     (0.62)                              N/A
selpref                                              -0.004                              -0.006
                                                     (0.28)                              (0.51)
sex                                                  -.0004859                           N/A
                                                     (0.04)                              N/A
Age                                                  0.001                               N/A
                                                     (0.93)                              N/A
circdum1                                             -0.021                              -0.018
                                                     (1.20)                              (1.19)
circdum2                                             -0.003                              0.002
                                                     (0.14)                              (0.07)
circdum3                                             0.010                               0.008
                                                     (0.55)                              (0.45)
circdum4                                             -0.004                              -0.008
                                                     (0.22)                              (0.51)
circdum5                                             -0.008                              0.001
                                                     (0.46)                              (0.06)
circdum6                                             -0.014                              -0.016
                                                     (0.67)                              (0.84)
circdum7                                             -0.019                              -0.020
                                                     (1.20)                              (1.32)
circdum8                                             -0.022                              -0.019
                                                     (1.43)                              (1.32)
circdum9                                             0.047                               0.043
                                                     (1.62)                              (1.52)
circdum10                                            -0.016                              -0.022
                                                     (0.92)                              (1.30)
circdum11                                            0.002                               N/A
                                                     (0.09)                              N/A
NETCOSTPOOL                                          N/A                                 -1.51e-08
                                                     N/A                                 (1.08)
Constant                                             0.009                               0.084
                                                     (0.14)                              (3.33)**

Observations                                         96                                  96
R-squared                                            0.21                                0.20
Robust t statistics in parentheses
* significant at 5%; ** significant at 1%
In model (2), votes by DC Circuit Judges not included and 11th circuit dummy left to the constant term.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                     Table 8

                             Relationship Between Financial Sacrifice and Dissent Rates

                                               Logit Model

Dependent Variable: Probability Judge Files a Dissent

                           Model (1)                 Model (2)                            Model (3)
                           (Judge-to-Judge)          (Judge-to-Judge)                     (Pool-to-Pool)
                           (Full Sample)             (Net Worth Subsample)


NETCOST                    -4.65e-07                 -1.25e-06                            N/A
                           (2.99)**                  (4.88)**                             N/A
Private Practice           0.090                     0.068                                N/A
                           (0.66)                    (0.34)                               N/A
selpref                    -0.019                    0.659                                -0.126
                           (0.10)                    (2.34)*                              (0.70)
sex                        0.137                     -0.115                               N/A
                           (0.87)                    (0.52)                               N/A
Age                        -0.017                    -0.066                               N/A
                           (1.32)                    (3.34)**                             N/A
circdum1                   -0.099                    -1.463                               0.614
                           (0.28)                    (2.15)*                              (1.60)
circdum2                   -0.741                    -1.267                               0.109
                           (1.98)*                   (2.44)*                              (0.27)
circdum3                   -0.389                    -1.109                               0.329
                           (0.96)                    (2.08)*                              (0.75)
circdum4                   0.205                     -0.863                               0.915
                           (0.62)                    (1.88)                               (2.48)*
circdum5                   -0.180                    -0.829                               0.438
                           (0.56)                    (2.19)*                              (1.22)
circdum6                   0.681                     -0.005                               1.432
                           (2.39)*                   (0.01)                               (4.39)**
circdum7                   -0.343                    -1.153                               0.328
                           (1.18)                    (2.81)**                             (0.98)
circdum8                   -0.637                    -1.221                               0.076
                           (1.97)*                   (2.97)**                             (0.21)
circdum9                   0.415                     -0.660                               1.253
                           (1.41)                    (1.64)                               (3.76)**
circdum10                  -0.521                    -1.432                               0.264
                           (1.44)                    (2.79)**                             (0.67)
circdum11                  -0.696                    -1.362                               N/A
                           (1.84)                    (3.04)**                             N/A
Net Worth                  N/A                       1.79e-08                             N/A
                           N/A                       (0.89)                               N/A
NETCOSTPOOL                                                                               -1.35e-07
Constant                   -1.856                    2.175                                -3.726
                           (2.33)*                   (1.70)                               (10.10)**

Observations                8669                     4136                                 8169
Robust z statistics in parentheses

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

* significant at 5%; ** significant at 1%; In model (3), votes by DC Circuit Judges are not included and
11th circuit dummy is left to the constant term.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                  Table 9

           Relationship between Financial Sacrifice and Total Number of Outside Circuit Citations

                                                OLS Model

Dependent Variable: Total Influence Measure

                                             Model (1)                                    Model (2)
                                             (Judge-to-Judge)                             (Pool-to-Pool)


NETCOST                                      1.13e-07                                     N/A
                                             (1.16)                                       N/A
Private Practice                             -0.126                                       N/A
                                             (1.79)                                       N/A
age                                          -0.004                                       N/A
                                             (0.46)                                       N/A
sex                                          0.062                                        N/A
                                             (0.68)                                       N/A
selpref                                      -0.232                                       -0.250
                                             (2.07)*                                      (2.52)*
Yale                                         0.328                                        N/A
                                             (2.93)**                                     N/A
Harvard                                      0.175                                        N/A
                                             (1.46)                                       N/A
Stanford                                     -0.209                                       N/A
                                             (1.97)                                       N/A
NYU                                          -0.257                                       N/A
                                             (0.67)                                       N/A
Columbia                                     0.066                                        N/A
                                             (0.33)                                       N/A
Other top ten law school                     0.042                                        N/A
                                             (0.40)                                       N/A
Law school ranked 10-20                      0.102                                        N/A
                                             (1.28)                                       N/A
ABA qualification                            -0.008                                       N/A
                                             (0.18)                                       N/A
NETCOSTPOOL                                                                               1.59e-07
Constant                                     3.025                                        2.828
                                             (5.40)**                                     (24.70)**

Observations                                  141                                          132
R-squared                                     0.19                                         0.04
Robust t statistics in parentheses
* significant at 5%; ** significant at 1%; In model (2), DC Circuit Judges are not included.

Working Draft Duke/UNC Conference; please don’t distribute, quote or cite without permission

                                                  Table 10

          Relationship between Financial Sacrifice and Average Number of Outside Circuit Citations

                                                OLS Model

Dependent Variable: Average Influence Measure

                                    Model (1)                           Model (2)
                                    Judge-to-Judge                      Pool-to-Pool


NETCOST                             1.47e-07                            N/A
                                    (2.00)*                             N/A
Private Practice                    -0.062                              N/A
                                    (1.37)                              N/A
selpref                             -0.293                              -0.285
                                    (2.94)**                            (2.71)**
age                                 0.003                               N/A
                                    (0.56)                              N/A
sex                                 -0.001                              N/A
                                    (0.01)                              N/A
Yale                                0.193                               N/A
                                    (2.68)**                            N/A
Harvard                             0.049                               N/A
                                    (0.70)                              N/A
Stanford                            -0.016                              N/A
                                    (0.18)                              N/A
NYU                                 -0.300                              N/A
                                    (1.53)                              N/A
Columbia                            -0.043                              N/A
                                    (0.35)                              N/A
Other top ten law school            0.054                               N/A
                                    (0.71)                              N/A
Law School ranked 10-20             0.042                               N/A
                                    (0.74)                              N/A
ABA qualification                   -0.019                              N/A
                                    (0.42)                              N/A
NETCOSTPOOL                         N/A                                 1.18e-07
                                    N/A                                 (2.03)*
Constant                            -0.025                              0.097
                                    (0.07)                              (1.12)

Observations                          142                               133
R-squared                             0.18                              0.08
Robust t statistics in parentheses
* significant at 5%; ** significant at 1%; In model (2), DC Circuit Judges are not included.


To top