70 DUKE L.J. __ (2020) (forthcoming)

Pay to Play? Campaign Finance and the Incentive Gap
in the Sixth Amendment’s Right to Counsel
Neel U. Sukhatme* and Jay Jenkins**
Abstract
For nearly 60 years, the U.S. Supreme Court has affirmed that the Sixth Amendment to the
U.S. Constitution guarantees felony defendants the right to counsel, regardless of their ability to pay.
Yet nearly all criminal procedure scholars agree that indigent defense as practiced today falls far
short of its initial promise. These scholars frequently cite a lack of political support, insufficient public
funding, and a failure to address instances of inadequate legal representation, among other things,
as causes for the underlying systemic dysfunction.
We contend that these conventional critiques are incomplete. Rather, indigent defense
systems often fail due to poor design: they do not align publicly funded defense attorneys with their
clients’ best interests. This is particularly true when courts appoint private attorneys to represent
indigent defendants for a fee, as is done in hundreds of jurisdictions across the United States. We
explain how such assignment systems create an “incentive gap” that financially motivates defense
attorneys to maximize their caseloads but minimize their efforts.
We then show how campaign finance exacerbates this problem. Specifically, we provide
empirical evidence that elected trial court judges and criminal defense attorneys regularly engage in
“pay to play,” where judges appoint attorneys who donate to their campaigns as counsel for indigent
defendants. We find trial judges routinely accept such donations, often as apparent “entry fees” from
attorneys who have just become eligible for appointments. These judges, in turn, typically award
their donors more than double the cases they award to non-donors, with the average donor attorney
earning greater than a 27-fold return on her donation. Indeed, we find indigent defense appointments
can be surprisingly lucrative, with many donor attorneys earning tens or even hundreds of thousands
of dollars across the hundreds of cases assigned to them by their donee judges.
Worse, this apparent quid pro quo between judges and defense attorneys appears to directly
harm defendants. We find that defense attorneys who donate to a judge are, if anything, less
successful than non-donor attorneys in attaining charge reductions, dismissals, and acquittals, or
avoiding prison sentences. We contend donor attorneys might underperform simply because they
take on so many more cases from their donee judges, and hence spend less time on each matter.
Our study is the first empirical analysis of how campaign finance distorts criminal trial court
decisionmaking. While our data are from Harris County (Houston), Texas—the nation’s third most
populous county—we show that pay to play is probably endemic across that state. Indeed, similar
problems likely affect millions of Americans, as trial judges who control indigent defense assignments
in many other states—including California, Georgia, Maryland, Missouri, North Carolina, and Ohio,
among others—accept attorney donations to fund their electoral campaigns. Unless substantial
reforms are made to address the corrosive influence of campaign finance on criminal defense, the
Sixth Amendment’s right to counsel will continue to ring hollow for millions of indigent defendants.
* Associate Professor (J.D., Ph.D. Economics), Georgetown University Law Center; Affiliated Faculty Member,
Georgetown University McCourt School of Public Policy.
** Project Attorney – Harris County (J.D.), Texas Criminal Justice Coalition.

Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

Introduction

1

The Promise and Failure of Gideon: A Brief History of the Right to Counsel
Under the Sixth Amendment

5

The Incentive Gap in the Right to Counsel
Background on Indigent Defense Systems
Assigned Counsel
Contract Attorneys
Public Defenders
Different Systems, Different Incentives
Caseload Incentives for Assigned Counsel and Contract Attorneys
Caseload Incentives for Public Defenders
Private Attorneys
The Incentive Gap Magnified: Campaign Finance and Pay to Play
Attorney Donations and Indigent Defense in Texas
Attorney Donations in Trial Courts Elsewhere

11
11
12
13
13
14
14
17
18
19
19
22

The Incentive Gap and Pay to Play: An Empirical Analysis of Campaign Finance and
Indigent Defense in Harris County, Texas
Background on Harris County
Courts and Judicial Elections
Indigent Defense
Construction of the Data
Summary Statistics
Cases and Contributions
Donations, Appointments, and Revenues
Pay to Play: Regressions and Graphical Evidence
Donations, Appointments, and Revenues
Comparisons Within the Donor Class
Donation Timing and “Entry Fees”
Case Outcomes

26
26
26
27
29
32
32
35
37
37
39
41
42

The Pervasiveness of the Incentive Gap: Campaign Finance and Other Reforms

45

Conclusion

49

Appendix

50

The authors thank participants at the Georgetown Law Summer and Fall Faculty Workshops and The
University of Texas Law and Economics Workshop; Jason Tashea and Keith Porcaro for connecting the
authors; the Texas Indigent Defense Commission; and Gregg Bloche, Alex Bunin, Jens Dammann, David
Hyman, Greg Klass, David Luban, Paul Ohm, David Schwartz, Mike Seidman, Gerry Spann, David Super,
Josh Teitelbaum, and Abe Wickelgren for helpful comments. We are grateful to Kyle Rozema for kindly sharing
Texas bar data. We especially thank Arturo Romero Yáñez for superb research assistance on this project. Any
errors are the authors’ own.

2
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
I.

Introduction

Were they alive today, Johnny Ray Johnson and Keith Steven Thurmond would be experts
on what happens when indigent defendants are not assigned competent counsel. Both were inmates
on death row in Harris County, Texas. In 2004, Mr. Johnson’s appointed attorney, Jerome Godinich,
missed the deadline to file a petition for federal habeas corpus relief. Mr. Godinich claimed a
malfunctioning filing machine failed to date-stamp the petition, which he filed after the court had
already closed on the deadline date. His client eventually lost on appeal and was executed.1
Months after the first missed deadline, Mr. Godinich missed the same deadline in Mr.
Thurmond’s case. His excuse? The same malfunctioning filing machine. The delay foreclosed a
petition from Mr. Thurmond, who was also eventually put to death.2
Sadly, cases like these are not so rare. Although many indigent defense counsel perform
admirably for their clients, others have fallen asleep during a client’s capital murder trial,3 neglected
to address an inflammatory opening by a prosecutor,4 or allowed their client to remain locked up in
pre-trial detention for 17 months before investigating facts in a drug possession case in which no
drugs were ever found.5 There is, unfortunately, no shortage of such horror stories.
More prosaically, public defenders and assigned defense counsel often fail their clients
simply because they are spread too thin across too many cases. Many of these attorneys handle
hundreds of matters at the same time,6 practically precluding them from conducting a focused
investigation on any one case or providing truly individualized counsel to any one defendant. So it is
no surprise that the vast majority of indigent defendants simply do the most expeditious thing and
quickly plead guilty to whatever deal they are offered—often after being encouraged to do so by their
own lawyer.
This dynamic has led most criminal procedure scholars and many practicing attorneys to
assail the state of indigent defense in America. To illustrate, the Yale Law Journal held a symposium
in 2013 to commemorate the 50th anniversary of Gideon v. Wainwright,7 the landmark decision that
established a right to counsel under the Sixth Amendment to the U.S. Constitution for all state felony
1

See Lise Olsen, Slow Paperwork in Death Row Cases Ends Final Appeals for 9, Hᴏᴜs. Cʜʀᴏɴ. (Mar. 21,
2009).
2
See Thurmond v. Thaler, 2011 U.S. App. LEXIS 1608 (2011).
3
See Carol S. Steiker, Gideon at Fifty: A Problem of Political Will, 122 Yᴀʟᴇ L.J. 2694 (2012) (citing Henry
Weinstein, A Sleeping Lawyer and a Ticket to Death Row, L.A. Tɪᴍᴇs, July 15, 2000, available at
http://articles.latimes.com/2000/jul/15/news/mn-53250).
4
See id. (describing how an indigent defense attorney said nothing in response to a prosecutor claiming the
defendant had bitten a child in a rape and murder case and could be identified through bite marks; the
defendant was exonerated 17 years later).
5
See Lise Olsen, Hundreds of Indigent Inmates Jailed for Months Pretrial, Hᴏᴜs. Cʜʀᴏɴ. (Oct. 3, 2009), at
https://www.chron.com/news/houston-texas/article/Hundreds-of-indigent-inmates-jailed-for-months1566415.php (noting failure of assigned counsel to investigate on behalf of defendant charged with possession
of cocaine even though no drugs were found on him).
6
See, e.g., Steiker, supra note __, at 2696 (noting, e.g., “In Miami-Dade County, Florida, the average felony
caseload per lawyer has reached five hundred in recent years due to budget cuts.”).
7
Gideon v. Wainwright, 372 U.S. 335 (1963). Indigent capital defendants have had the right to counsel since
the Supreme Court’s decision in Powell v. Alabama, 287 U.S. 68, 68-69.

Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
defendants. The occasion, however, turned out to be more lamentation than celebration, as legal
scholars and practitioners pointed out again8 and again9 and again10 the persistent failures of indigent
defense systems across the country.
Most of these critiques rely in some part on a standard narrative that centers on money and
power.11 Critics recognize that while Gideon required states to provide counsel for indigent
defendants, this mandate was unfunded.12 Political actors, in turn, lacked the will to provide the
resources necessary to make a poor defendant’s right to counsel more than a symbolic gesture.13
Combined with a substantial increase in prosecutorial power, as well as the Supreme Court’s
reticence to put substantial teeth into the requirement that appointed counsel actually be effective,14
the manifest failures of indigent defense in America are hardly surprising.
Certainly this standard narrative contains a good bit of truth. But, as we show here, it is
incomplete, particularly in the context of assigned counsel systems, in which courts assign private
attorneys to represent defendants for a fee. Such systems, which are used in hundreds of
jurisdictions across the country, are undoubtedly underfunded on the whole. But that does not mean
they are not highly profitable for some. As we show, many defense attorneys make tens or even
hundreds of thousands of dollars in a given year across assignments from a single judge.

8

See, e.g., Roger A. Fairfax, Jr., Searching for Solutions to the Indigent Defense Crisis in the Broader Criminal
Justice Reform Agenda, 122 Yᴀʟᴇ L.J. 2150 (2012) (“[F]ifty years after Gideon was decided, there is nearuniversal acceptance of the notion that our system of indigent defense is broken.”); Pamela R. Metzger, Fear
of Adversariness: Using Gideon To Restrict Defendants’ Invocation of Adversary Procedures, 122 Yᴀʟᴇ L.J.
2550 (20120) (arguing “the Supreme Court has used Gideon to decrease the protection of Sixth Amendment
rights that constitute the core structures of the American adjudicatory process.”)
9
See Stephen B. Bright & Sia M. Sanneh, Fifty Years of Defiance and Resistance After Gideon v. Wainwright,
122 Yᴀʟᴇ L.J. 2150 (2012); see also Victoria Nourse, Gideon’s Muted Trumpet, 58 Mᴅ. L. Rᴇᴠ. 1417 (“Once
the darling of the legal academy, criminal procedure has fallen into disrepute.”); see also id. at 1431 (“The
importance of the lawyer . . . is not someone to try a case (most cases are never tried), but someone to stand
with the individual citizen unaligned to the forces that laid him low and resist claims of the natural and inevitable
superiority of government.”).
10
See Gabriel J. Chin, Race and the Disappointing Right to Counsel, 122 Yᴀʟᴇ L.J. 2236 (2012).
11
By contrast, a few critiques have cut much deeper. In particular, Paul Butler argues Gideon itself was
misguided and has led to more harm than good. He contends that by coating the trial process with a false
veneer of objectivity and fairness, the right to counsel might actually obfuscate the system’s true goal of
subjugating poor and black people. See Paul D. Butler, Poor People Lose: Gideon and the Critique of Rights,
122 Yᴀʟᴇ L.J. 2176 (2012). As Professor Butler notes, similar arguments have also been made in other areas
of criminal procedure. See, e.g., Louis Michael Seidman, Brown and Miranda, 80 Calif. L. Rev. 673, 719 (1992)
(arguing that Brown v. Board of Education, 347 U.S. 483 (1954) and Miranda v. Arizona, 384 U.S. 436 (1966)
“served to stabilize and legitimate the status quo by creating the illusion of closure and cohesion”); see also
William J. Stuntz, Tʜᴇ Cᴏʟʟᴀᴘsᴇ ᴏғ Aᴍᴇʀɪᴄᴀɴ Cʀɪᴍɪɴᴀʟ Jᴜsᴛɪᴄᴇ 3 (2011) (noting that accused persons have
fared worse as rights have expanded).
12
See, e.g., Erwin Chemerinsky, Lessons from Gideon, 122 Yᴀʟᴇ L.J. 2676, 2680 (2012) ("The Court imposed
an unfunded mandate on state governments without any enforcement mechanism, and the Court then
undermined the one remedy available to the judiciary, the ability to find ineffective assistance of counsel."); M.
Clara Garcia Hernandez & Carole J. Powell, Valuing Gideon’s Gold: How Much Justice Can We Afford?, 122
Yᴀʟᴇ L.J. 2358 (2012).
13
See Carol S. Steiker, Gideon at Fifty: A Problem of Political Will, 122 Yᴀʟᴇ L.J. 2694 (2012).
14
See infra notes __ and accompanying text (discussing Strickland v. Washington, 466 U.S. 668 (1984) and
succeeding cases).

2
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
Moreover, these attorneys often continue to receive such assignments even after they have
been revealed to be woefully incompetent. For example, in the years after Mr. Johnson and Mr.
Thurmond were executed, Mr. Godinich was appointed as counsel in state court for hundreds of
other felony defendants, continuing to amass a record of inadequate representation during this
time.15 Judges could have appointed other lawyers to these cases. Why did they continue to appoint
the same underperforming attorneys?
The proximate cause,16 we argue, may lie in campaign finance. Mr. Godinich was assigned
the majority of his felony cases from one trial court judge, Jim Wallace. Mr. Godinich was also a
significant donor to Judge Wallace. Between 2005 and 2014, Mr. Godinich donated on seven
occasions a total of $9,000 to Judge Wallace’s electoral campaigns. Between 2004 and 2018, Judge
Wallace appointed Mr. Godinich to 1,974 cases, including 5 capital cases. And since 2014, Mr.
Godinich has earned at least $872,642.50 from cases before Judge Wallace. Clearly, some indigent
defense attorneys are not just scraping by.
While Mr. Godinich is an outlier, the link between campaign finance and indigent defense
extends far beyond him. In this Article, we expose this hidden connection. In particular, we present
empirical evidence of “pay to play,” where defense attorneys donate to elected judges to obtain or
maintain access to indigent defense cases.
We accomplish this by creating the first large-scale database that links trial court campaign
contributions with multiple criminal court datasets. Specifically, our data comprise the universe of
290,633 felony cases from January 2005 through May 2018 where defense counsel were assigned
and ascertainable in Harris County, the nation’s third most populous county and the home of the City
of Houston. While a few recent papers have looked at how campaign donations might influence
elected state Supreme Court justices,17 ours is the first to explore how campaign finance might affect
trial court decision-making and criminal case outcomes.

15

See Casey Tolan, She Watched Her Husband Get Sentenced to Death. Now She’s Becoming a Lawyer to
Save Him and Others., Sᴘʟɪɴᴛᴇʀ (Oct. 24, 2016), available at https://splinternews.com/she-watched-herhusband-get-sentenced-to-death-now-she-1793863088 (“Godinich and his second chair attorney didn’t even
meet with Juan until just before the trial, and conducted almost no investigation, Yancy said. … at times during
jury deliberations, neither of Juan's lawyers were present while the judge and prosecutors responded to
questions from the jury by themselves. Hartwell said that when she confronted Godinich, he told her, ‘I have
another trial to take care of.’ (Godinich did not respond to a request for comment.)”)); see also Stephen B.
Bright, Independence of Counsel: An Essential Requirement for Competent Counsel and a Working Adversary
System, 55 Hᴏᴜs. L. Rᴇᴠ. 853, 869 (2018) (“[Jerome Godinich] was assigned 406 felony cases at the trial level
in 2017—more than twice the national standard of 150 felony cases—as well as 8 capital cases, and 7 felony
appeals.”) (citations omitted)).
16
We say proximate cause because we recognize, as critical legal studies scholars have argued, that deeper
societal problems might be common drivers of all of these phenomena. If indeed “prison is for the poor, and
not the rich,” see Butler, supra note __, at 2178, then the whole system of judicial campaign finance might be
another system that was created to be indifferent or even hostile to the interests of the poor.
17
See Michael S. Kang & Joanna M. Shepherd, The Partisan Price of Justice: An Empirical Analysis of
Campaign Contributions and Judicial Decision, 86 N.Y.U L. Rᴇᴠ. 69 (2011) (finding elected state supreme
court justices more likely to rule in favor of business interests as donations from those interests increase);
Joanna M. Shepherd, Money, Politics, and Impartial Justice, 58 Dᴜᴋᴇ L.J. 623 (2009); Damon M. Cann, Justice
for Sale? Campaign Contributions and Judicial Decisionmaking, 7 Sᴛ. Pᴏʟ. & Pᴏʟ. Q. 281 (2007).

3
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
What we find is shocking. Trial judges routinely accept donations from defense attorneys who
practice before them. Worse, judges often accept these donations as an apparent “entry fee” from
counsel soon after they become eligible for indigent defense appointments. And while donor and
non-donor attorneys appear similar in terms of their education and experience, on average, judges
assign their donors more than double the number of cases they assign to non-donors. Such
assignments are flatly inconsistent with the prescribed rotation or “wheel” assignment system
mandated by Texas law.18
In addition, we find these preferential assignment patterns enable donors to earn on average
more than double the total attorneys fees of non-donors. And the total amount of these fees can be
quite significant: the average donor earns $31,081 in attorneys fees from her donee judge.19 Put
another way, if pay to play exists, then our regression analysis suggests the average defense
attorney receives more than a 27-fold “return” on her campaign contributions.20
More troubling, the apparent quid pro quo that we observe might directly harm indigent
defendants. We find that if anything, defense attorneys who donate to judges are less successful
than those who don’t in terms of attaining charge reductions, dismissals, and acquittals, or avoiding
prison sentences for their clients. These results are not driven solely by observable differences in
assigned cases or defendant characteristics, or unobservable differences that remain fixed over
time. We suggest that donor attorneys might underperform simply because they take on so many
more cases from their donee judges, and hence have less time to spend on each matter.21
We also present qualitative evidence based on news reports and interviews with practicing
attorneys in Texas that suggests the phenomena described here are endemic across the state.
Furthermore, we explain why these problems might exist throughout the country, as trial judges in
many other states (including California, Georgia, Maryland, Missouri, North Carolina, and Ohio,
among others) can receive donations from attorneys and control assignments of indigent defense
counsel.22
Some might view our results as yet another nail in the coffin of Gideon—as further proof that
America’s grand experiment with indigent defense is a failure, and that societal forces will ensure
that it continues to be a failure absent sweeping social and political change. Perhaps this is true. But
we argue, more modestly, that our research reveals the more immediate problem is one of
misaligned incentives: as practiced today, no one’s interests are truly aligned with those of indigent
defendants. Instead, assigned counsel are often financially incentivized to dispose of indigent
18

See infra notes __ and accompanying text.
See infra notes __ and accompanying text.
20
To give some perspective on how extraordinary this return is: if you had invested $10,000 in a portfolio that
replicated the S&P 500 in March 2009, at the market low during the Great Recession, and sold in July 2019,
when the market recently peaked, your portfolio would have grown to $48,708.38—just less than a 5-fold
increase. See S&P 500 Periodic Reinvestment Calculator (With Dividends), at https://dqydj.com/sp-500periodic-reinvestment-calculator-dividends/ (last visited Sept. 3, 2019). Moreover, as we explain later, our
return estimates are if anything significantly underestimated, as we have reliable attorney revenue data only
from 2014-2018 and not from earlier years. See infra notes __ and accompanying text.
21
See infra notes __ and accompanying text.
22
See infra notes __ and accompanying text.
19

4
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
defense cases as quickly as possible, with as little effort as possible—motives that line up better with
zealous prosecutors and docket-conscious judges than with the best interests of poor defendants.
We argue that this “incentive gap” is an underappreciated fault at the heart of the right to
counsel. We further explain that while modest reforms to attorney assignment rules and campaign
finance regulations might reduce pay to play, they are unlikely on their own to solve the problem or
substantially improve the lot of poor defendants absent deeper structural changes that alter attorney
incentives.
The next section provides a brief history of indigent defense in the United States, starting
before Gideon and moving to present-day. Section III then overviews the three main models of
indigent defense in America: assigned counsel, contract, and public defender systems. It further
explains how each of these systems creates attorney incentives at odds with those of indigent
defendants. This section also provides qualitative evidence on how campaign finance interacts with
these assignment systems to further distort attorney incentives and widen the incentive gap.
Section IV turns toward empirics. After describing how the dataset was constructed, the
section provides summary statistics, regression specifications, and graphical evidence to show how
pay to play is likely influencing indigent defense appointments in Harris County. Section V briefly
lays out some policy proposals that seek to eliminate pay to play. It further explains why such
proposals are unlikely to be successful or to substantially improve the quality of indigent defense
unless they are also accompanied by deeper structural reforms. A short conclusion then follows.

II.

The Promise and Failure of Gideon: A Brief History of the Right to Counsel
Under the Sixth Amendment

The origins of the right to counsel in the United States begin far before Gideon in 1963.
Twelve of the original thirteen colonies declared some form of this right in early versions of their state
constitutions, despite English common law denying such a right existed.23 The Supreme Court
described this split from the common law in Holden v. Hardy, decided at the turn of the 20th century:
The earlier practice of the common law . . . so far as it deprived him of the assistance of
counsel and compulsory process for the attendance of his witnesses, it had not been
changed in England. But to the credit of her American colonies, let it be said that so
oppressive a doctrine had never obtained a foothold there.24
By the early 20th century, every state “had established the right to have the assistance of
counsel similar to that found in the Sixth Amendment”25 at that time and required the appointment of
counsel in capital cases.26 Nonetheless, the scope of the right was inconsistent across states, with
23

See Powell v. Alabama, 287 U.S. 45, 61-62 (1932).
Holden v. Hardy, 169 U.S. 366, 386 (1898).
25
Sɪxᴛʜ Aᴍ. Cᴛʀ., The Story of the Scottsboro Boys in Powell v. Alabama, at https://sixthamendment.org/theright-to-counsel/history-of-the-right-to-counsel/the-story-of-the-scottsboro-boys/ (citing William Beaney, Tʜᴇ
Rɪɢʜᴛ ᴛᴏ Cᴏᴜɴsᴇʟ ɪɴ Aᴍᴇʀɪᴄᴀɴ Cᴏᴜʀᴛs 19 (U of Mich, 1955)); U.S. Cᴏɴsᴛ., Sɪxᴛʜ Aᴍ. (“In all criminal
prosecutions, the accused shall enjoy the right . . . to have the Assistance of Counsel for his defence.”).
26
Id.
24

5
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
one scholar estimating that by the early 1930’s approximately half of all states had yet to extend the
right to counsel to non-capital cases.27
The calculus on the right to counsel began to shift in 1932 with the U.S. Supreme Court’s
decision in Powell v. Alabama.28 In Powell, the Court overturned the conviction of six young men
charged with rape on the grounds that they had been denied access to counsel.29 The Powell Court
found that in capital cases, the Sixth Amendment right to counsel was a fundamental right
encompassed by the Due Process Clause of the 14th Amendment.30
This “fundamental fairness” test soon became a part of Sixth Amendment jurisprudence. The
Court would next apply it in the context of a federal criminal trial in 1938’s Johnson v. Zerbst.31 In
Johnson, two men facing federal counterfeiting charges were denied access to counsel, ultimately
being convicted and sentenced.32 In overturning the convictions, the Johnson Court definitively
established a right to counsel in all federal criminal proceedings.33
Still, the Court was not yet ready to extend the right to counsel further to encompass state
proceedings. In Betts v. Brady,34 a defendant charged with a non-capital crime in Maryland state
court challenged the court’s refusal to appoint him counsel.35 Rather than require counsel to be
available for indigent defendants in all state criminal proceedings, the Betts Court pointed to the
diversity of access to counsel constitutional provisions and statutes as proof that “in the great majority
of states, it has been the considered judgment of the people, their representatives and their courts
that appointment of counsel is not a fundamental right, essential to a fair trial.”36 The Betts Court then
enumerated a “special circumstances” test as to whether a state had violated a defendant’s due
process rights by denying them access to counsel.37
Many state supreme courts and legislatures responded to Betts by affirming their support for
the right to counsel, and by the late 1950’s, thirty-five states had required counsel be appointed in

27

Sɪxᴛʜ Aᴍ. Cᴛʀ., The Story of the Scottsboro Boys in Powell v. Alabama, at https://sixthamendment.org/theright-to-counsel/history-of-the-right-to-counsel/the-story-of-the-scottsboro-boys/
(citing
Hugh
Richard
Williams, Tʜᴇ Hɪsᴛᴏʀʏ ᴏғ ᴛʜᴇ Rɪɢʜᴛ ᴛᴏ Fʀᴇᴇ Cᴏᴜɴsᴇʟ ɪɴ Aᴍᴇʀɪᴄᴀ 29 (Ill. Univ. Sch. of Law: Fall 2001)).
28
287 U.S. 45 (1932).
29
Id. at 65. After an out-of-state lawyer who had offered to help local counsel refused an appointment to act
as counsel for the defendants, the trial judge then appointed all the members of the local bar as counsel but
neglected to name an individual lawyer to act in that capacity.
30
Id.
31
304 U.S. 458 (1938).
32
Id. at 459-60. The District Attorney in the case denied the men’s request for counsel on the grounds that the
courts in South Carolina (where the trial occurred) did not appoint counsel unless the defendant was charged
with a capital crime.
33
Id. at 467 (“Since the Sixth Amendment constitutionally entitles one charged with crime to the assistance of
counsel, compliance with this constitutional mandate is an essential jurisdictional prerequisite to a federal
court's authority to deprive an accused of his life or liberty.”).
34
316 U.S. 455 (1942).
35
Id. at 456-58.
36
Id. at 467.
37
Id. at 463-64.

6
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
non-capital cases, at least to some degree.38 This affirmation of the right to counsel belied the Betts
Court’s holding that the right was not considered fundamental. And indeed, the Court would revisit
that issue just over 20 years later in Gideon v. Wainwright.39
In Gideon, just as in Betts, the defendant faced felony charges in state court, this time in
Florida.40 As in Betts, the trial court denied the defendant’s request to provide an attorney when the
defendant was unable to hire his own.41 Reversing their decision in Betts, the Gideon Court
unanimously found that the right to counsel was essential to a fair trial in state criminal proceedings,
noting: “[R]eason and reflection require us to recognize that in our adversary system of criminal
justice, any person haled into court, who is too poor to hire a lawyer, cannot be assured a fair trial
unless counsel is provided for him.”42
Writing for the Court, Justice Hugo Black framed the issue in terms of resources and
incentives for all parties involved, noting that government actions and the behavior of wealthy
defendants rendered access to counsel essential for a fair trial:
Governments, both state and federal, quite properly spend vast sums of money to
establish machinery to try defendants accused of crime. Lawyers to prosecute are
everywhere deemed essential to protect the public's interest in an orderly society.
Similarly, there are few defendants charged with crime, few indeed, who fail to hire the
best lawyers they can get to prepare and present their defenses. That government hires
lawyers to prosecute and defendants who have the money hire lawyers to defend are
the strongest indications of the wide-spread belief that lawyers in criminal courts are
necessities, not luxuries. The right of one charged with crime to counsel may not be
deemed fundamental and essential to fair trials in some countries, but it is in ours.43
Justice Black concluded that the ideals of American government necessitated the
appointment of counsel to indigent defendants in state courts:
From the very beginning, our state and national constitutions and laws have laid great
emphasis on procedural and substantive safeguards designed to assure fair trials before
impartial tribunals in which every defendant stands equal before the law. This noble
ideal cannot be realized if the poor man charged with crime has to face his accusers
without a lawyer to assist him.44

38

See Sixth Amendment Center, The Post-Betts Era in the States’ Courts, at https://sixthamendment.org/theright-to-counsel/history-of-the-right-to-counsel/the-post-betts-era/ (citing Wɪʟʟɪᴀᴍ Bᴇᴀɴᴇʏ, Tʜᴇ Rɪɢʜᴛ ᴛᴏ
Cᴏᴜɴsᴇʟ ɪɴ Aᴍᴇʀɪᴄᴀɴ Cᴏᴜʀᴛs (U. Mich. 1955)). See also Stacey L. Reed, A Look Back at Gideon v. Wainwright
after Forty Years: An Examination of the Illusory Sixth Amendment Right to Assistance of Counsel, 52 Dʀᴀᴋᴇ
L. Rᴇᴠ. 47, 51 (2003) (citing Wayne R. LaFave et al., Criminal Procedure § 11.1 (3d ed.2000)).
39
372 U.S. 335 (1963). The Court noted in its decision that 22 states, as amici, argued that Betts was "an
anachronism when handed down," and should now be overruled. Id.
40
Id. at 336-37.
41
Id. at 337.
42
Id. at 344.
43
Id.
44
Id.

7
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
Gideon was warmly received by the public,45 inspiring the award-winning book Gideon’s
Trumpet.46 And in subsequent years, the Court continued to expand the situations in which the right
to counsel was deemed necessary, holding that it attaches from the time of indictment47 to the time
of a defendant’s first appeal.48 The Court subsequently extended the right to probation revocation
hearings49 and juvenile delinquency proceedings.50 Nearly a decade after Gideon, the Court ruled
that the right to counsel attaches in any criminal case resulting in actual imprisonment,51 putting to
rest many questions about Gideon’s scope.
Yet, even the most ardent supporters of Gideon acknowledged that significant resources
would need to be martialed to ensure the right to counsel was actually robust in practice. In a famous
1965 law review article, Henry Monaghan forewarned the coming tension between the promise of
Gideon and the practical and logistical problems that might arise: “Gideon's little regiment had no
real difficulty in running up its colors, but it is quite apparent that an army—a very large one—must
be raised if the victory is to be a lasting one.”52
By the time Gideon’s Trumpet was turned into a popular television movie starring Henry
Fonda in 1980,53 the shortcomings of that decision had become painfully apparent. Many states had
failed to take the steps needed to effectively comply with the ruling. These failures continue today,
with many scholars attributing the continued crisis in access to counsel to weaknesses and
oversights in Gideon itself.
In particular, many scholars argue that a key failure of Gideon was that it did not define, let
alone mandate, what kind of counsel defendants are entitled to. Neither did it define what states

45

See Reed, supra note __.
Anthony Lewis, Gɪᴅᴇᴏɴ's Tʀᴜᴍᴘᴇᴛ. New York: Vintage Books/Random House, 1964.
47
See Massiah v. United States, 377 U.S. 201 (1964); see also Rothgery v. Gillespie County, 554 U.S. 191
(2008) (right to counsel triggered at criminal defendant’s initial appearance before a magistrate judge).
48
See Douglas v. California, 372 U.S. 353 (1963) (a companion case to Gideon); see also Halbert v. Michigan,
545 U.S. 605 (2005) (state may not deny attorney for defendant who seeks to appeal after entering a guilty
plea); Griffin v. Illinois, 351 U.S. 12 (1956) (due process and equal protection require state to pay for trial
transcript for indigent defendant if necessary for defendant’s appeal to be heard).
49
See Mempa v. Rhay, 389 U.S. 128 (1967). More recently, the Court has held that ineffective assistance of
counsel claims can be made based on the plea bargaining process. See Missouri v. Frye, 566 U.S. 134 (2012)
(defense attorneys have a duty to convey plea bargain offers to defendants); Lafler v. Cooper, 566 U.S. 156
(2012) (prosecutor might be required to reoffer plea if defendant originally rejected it due to ineffective
assistance of counsel).
50
In re Gault, 387 U.S. 1 (1967).
51
See Argersinger v. Hamlin, 407 U.S. 25 (1972); see also Scott v. Illinois, 440 U.S. 367 (1979) (clarifying that
the right to counsel is required whenever a defendant was actually sentenced to prison, but not if a prison
sentence was merely authorized under the charging statute); Alabama v. Shelton, 535 U.S. 654 (2002)
(suspended prison sentence may not be imposed if defendant lacked an attorney at trial); see also Shaun
Ossei-Owusu, The Sixth Amendment Façade: The Racial Evolution of the Right to Counsel, 167 U. Pᴀ. L. Rᴇᴠ.
__ (2019) (“During this decade, which saw larger criminal procedure reforms, the Court had the most generous
approach toward indigent defense.”).
52
Henry P. Monaghan, Gideon's Army: Student Soldiers, 45 B.U. L. Rev. 445, 446 (1965).
53
Gideon’s Trumpet. Directed by Robert L. Collins, performances by Henry Fonda, Jose Ferrer, John
Houseman, and Fay Wray, Worldvision Enterprises (1980).
46

8
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
must do to ensure this type of counsel is available for indigent defendants. Put differently, Gideon
gave states considerable freedom to shirk the large unfunded mandate that the case had created.
These issues were exacerbated by subsequent Court decisions, perhaps most notably
Strickland v. Washington.54 There, the Court determined that claims of ineffective assistance of
counsel can succeed only if there were an initial determination that a defense attorney’s performance
was so lacking “that counsel was not functioning as the ‘counsel’ guaranteed the defendant by the
Sixth Amendment.”55 Moreover, the defendant must show the ineffectiveness prejudiced him
severely enough such that there is a reasonable possibility that the outcome of the defendant’s case
would have been different without the deficient performance of counsel.56
Justice Thurgood Marshall noted in his Strickland dissent that this exacting standard would
be hard to meet, since “it is often very difficult to tell whether a defendant convicted after a trial in
which he was ineffectively represented would have fared better if his lawyer had been competent.”57
And he was right: while ineffective assistance of counsel claims are regularly made, they are rarely
successful.58 The end result is that indigent defendants might get stuck with counsel who are not
“ineffective,” but whom paying clients would never select.59
With ineffective assistance claims largely rendered futile, some sought to remedy the
dramatic underfunding of indigent defense through systemic, structural litigation. But the potential for
such federal litigation was limited, as federal courts relied on the abstention doctrine to avoid
intervening in ongoing state cases.60 Structural litigation in state courts has been similarly
unsuccessful, with courts demanding a showing of prejudice or the existence of a conflict on an

54

466 U.S. 668 (1984).
Id. at 687.
56
Id.
57
Id. at 710 (Marshall, J. dissenting).
58
A 2013 analysis found that Supreme Court findings of ineffective assistance of counsel have been virtually
non-existent since Strickland. Erwin Chemerinsky, Lessons from Gideon, 122 Yᴀʟᴇ L.J. 2676, 2689 (2013).
Professor Chemerinsky was able to identify just two Supreme Court findings of ineffective assistance of
counsel in the 25 years following Gideon—Wiggins v. Smith and Rompilla v. Beard, both capital cases. He
also noted that the findings in Wiggins and Rompilla were called into question by the Court’s ruling in Cullen
v. Pinholster, 563 U.S. 170 (2011), which held that evidence of ineffective assistance of counsel could not be
presented at federal habeas proceedings. See also Padilla v. Kentucky, 559 U.S. 356 (2010) (holding that
criminal defense attorneys must inform non-citizen defendants of deportation consequences of guilty pleas).
59
As one private criminal defense firm in Texas colorfully (and probably self-servingly) noted on its website, “If
you’re appointed a lawyer in a criminal case in Texas, your court-appointed lawyer may be an incompetent
hack, or he may be a truly outstanding attorney. Some of the best criminal-defense lawyers in Houston
represent indigent defendants . . .By the same token, if you hire a lawyer for a criminal case in Texas, your
retained lawyer might be either a fine attorney or a bungler.” See Bennett & Bennett, Court Appointed Lawyers,
at https://bennettandbennett.com/about/court-appointed-lawyers/ (last visited Sept. 28, 2019).
60
Chemerinsky at 2687; see also Younger v. Harris, 401 U.S. 37, 43–54 (1971) (federal courts to abstain from
issuing rulings that interfere with ongoing state criminal prosecutions); Margaret A. Costello, Fulfilling the
Unfulfilled Promise of Gideon: Litigation as a Viable Strategic Tool, 99 Iᴏᴡᴀ L. Rᴇᴠ. 1951, 1962 (2014);
Cara H. Drinan, The Third Generation of Indigent Defense Litigation, 33 N.Y.U. Rᴇᴠ. L. & Sᴏᴄ. Cʜᴀɴɢᴇ 427,
438-43 (2009).
55

9
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
individual case-by-case basis from public defenders seeking to remedy funding disparities.61 The
heightened evidentiary requirements have limited the number of structural indigent defense litigation
suits in state courts, with one scholar stating “it is estimated that no more than ten of these suits were
filed between 1980 and 2000.”62
Gideon’s failures have also been pinned on its inability to change underlying power dynamics
and political considerations that adversely affect indigent defendants. In particular, scholars point to
asymmetries in monetary support and available resources for prosecutors and indigent defense
counsel. Prosecutors work closely with law enforcement agencies, whose job is to investigate in
support of prosecution; defense counsel are often left to fend for themselves should they desire to
investigate on behalf of their clients.63 Prosecutors are afforded large budgets dedicated to
investigation; indigent defense attorneys are rarely (if ever) afforded such a luxury.64 And some
judges rely heavily—perhaps inappropriately—on prosecutors, following their recommendations and
even asking prosecutors to write orders for them,65 a benefit that is rarely accorded to counsel
advocating on behalf of indigent defendants.
The politics of mass incarceration in the decades following Gideon have further entrenched
the built-in advantages that prosecutors enjoy over defense counsel. While the 1960’s and 1970’s
brought only moderate increases in the size of America’s criminal justice system, the “tough-oncrime” 1980’s and 1990’s created a system with an incarceration rate higher than any other
industrialized nation.66 The demand for indigent defense likewise exploded, financially straining a
system that was already starved for resources.
The political battle for indigent defense resources that pitted popular tough-on-crime policies
versus poor people accused of crimes was decidedly and expectedly one-sided. As one scholar put
it when discussing capital defendants: “The individuals adversely affected by this crisis—those
accused of aggravated murder—are the most hated and the least politically powerful in the country,
and political actors, including judges, are not highly motivated to make unpopular decisions that
would benefit them.”67 With political considerations weighing heaviest on publicly elected prosecutors
and judges, there was little to gain professionally from insisting on a level playing field.68

61

See Chemerinsky, supra note __ at 2687 (citing Florida v. Public Defender 12 So. 3d 798 (Fla. Dist. Ct. App.

2009)).
62
See Drinan, supra note __, at 468.
63

Id. at 2156. Moreover, discovery is limited in criminal cases, so defendants are also limited in terms of what
information they can obtain from prosecutors and what prosecutors are obligated to disclose to defendants.
See Brady v. Maryland, 373 U.S. 83 (1963).
64
Id.
65
Id.
66
See Chemerinsky at 2686 (citing Michelle Alexander, Tʜᴇ Nᴇᴡ Jɪᴍ Cʀᴏᴡ: Mᴀss Iɴᴄᴀʀᴄᴇʀᴀᴛɪᴏɴ ɪɴ ᴛʜᴇ Aɢᴇ
ᴏғ Cᴏʟᴏʀʙʟɪɴᴅɴᴇss (2012)).
67
Id. (citing Douglas W. Vick, Poorhouse Justice: Underfunded Indigent Defense Services and Arbitrary
Death Sentences, 43 Bᴜғғ. L. Rᴇᴠ. 329, 459 (1995)).
68
Id. at 2157-58. Some scholarly critiques of Gideon also focus on its impact on other criminal procedural
rights. See Justin F. Marceau, Gideon’s Shadow, 122 Yᴀʟᴇ L.J. 2482 (2013) (arguing Gideon has been used
as a cudgel to curb other procedural rights); Pamela Metzger, Fear of Adversariness: Using Gideon to Restrict
Defendants’ Invocation of Adversary Procedures, 122 Yᴀʟᴇ L.J. 2550 (demonstrating how Gideon and

10
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
III.

The Incentive Gap in the Right to Counsel

As we have seen, the long history of the right to counsel is largely defined by false hopes and
unmet expectations. And as is commonly argued, inadequate funding, tepid political support, and
the practical inability of defendants to succeed on most ineffective assistance of counsel claims are
important reasons why indigent defense has not evolved in the way that many advocates hoped it
would after Gideon.
But indigent defense in America doesn’t just suffer from a lack of money or a lack of political
support; it also suffers from poor design.69 Indeed, most indigent defense systems are built on a
shaky foundation that puts publicly funded defense attorneys at odds with those they are supposed
to represent. This incentive gap, in turn, adversely affects the legal representation these defendants
receive.70
In this section, we begin exploring this dynamic. We start by summarizing the different models
of indigent defense adopted by various jurisdictions, and then discussing how they distort attorney
incentives in different ways. Next, we explain how the relationship between judges and defense
attorneys, mediated through the mechanism of campaign finance, exacerbates the incentive gap,
thereby further adversely impacting the provision of indigent defense. Finally, we provide qualitative
evidence that indigent defense attorneys and trial judges might participate in pay to play in a number
of jurisdictions.

A. Background on Indigent Defense Systems
Gideon and its progeny famously failed to direct states on how to provide counsel to poor
defendants, thereby giving them substantial latitude in crafting indigent defense systems. States are
fairly evenly split between administering indigent defense at the local (usually county) level, or at the
state level.71

Strickland have weakened the adversarial process). Others have focused on its failure to adequately protect
black defendants. See, e.g., Gabriel J. Chin, Race and the Disappointing Right to Counsel, 122 Yᴀʟᴇ L.J. 2126,
2236 (2013) (arguing the Gideon court failed to directly address racial discrimination and this failure portended
larger failures to protect the rights of African Americans subject to the Jim Crow system of criminal justice).
Still others suggest Gideon might actually be counterproductive to the plight of the poor in the justice system.
See Butler, supra note __, at __ (“Poor people lose, most of the time, because in American criminal justice,
poor people are losers. Prison is designed for them. This is the real crisis of indigent defense. Gideon obscures
this reality, and in this sense stands in the way of the political mobilization that will be required to transform
criminal justice.”); see also John H. Blume & Sheri Lynn Johnson, Gideon Exceptionalism?, 122 Yᴀʟᴇ L.J. 2126
(2013) (attempting to reconcile the state of indigent defense in America with the general reverence for Gideon
by declaring it to be both a “shining city on a hill” . . . and a sham.”).
69
Of course, these phenomena might be interrelated—for example, some poor design choices in indigent
defense systems might be deliberately made by those looking to render the system impotent. And a lack of
funding is one way to make sure that poor design leads to worse outcomes.
70
A rare paper that does look at how assigned counsel incentives are affected by payment systems is Benjamin
Schwall, More Bang for Your Buck: How to Improve the Incentive Structure for Indigent Defense Counsel, 14
Oʜɪᴏ Sᴛ. J. Cʀɪᴍ. L. 553 (2017).
71
See Suzanne M. Strong, State-Administered Indigent Defense Systems, 2013, Bureau of Justice Statistics
(Nov. 2016) (“In 2013, 28 states and the District of Columbia had state-administered indigent defense programs
for the delivery of criminal defense services.”).

11
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
Jurisdictions also are split among three different models of representation: assigned counsel,
contract, and public defender.72 Each of these models is commonly used in the United States. Public
defender systems are used most often in larger cities, whereas assigned counsel systems are used
in the most jurisdictions overall, particularly in rural parts of the country. We describe each of these
systems below.

1. Assigned Counsel
Assigned counsel systems rely on private attorneys to represent indigent defendants.73 The
oldest model is the “Ad Hoc Assigned Counsel” system,74 where “appointment of counsel is generally
made by the court, without benefit of a formal list or rotation method and without specific qualification
criteria for attorneys.”75 Popular in smaller, more rural jurisdictions, ad hoc assigned counsel systems
are frequently criticized “for fostering patronage and lacking control over the experience level of
qualified attorneys.”76
Other iterations of assigned counsel systems address some, but not all, of the ad hoc model’s
shortcomings. One such system is sometimes referred to as the “wheel” system, in which “private
attorneys, acting as independent contractors and compensated with public funds, are individually
appointed from a list of qualified attorneys using a system of rotation to provide legal representation
and service to a particular indigent defendant accused of a crime or juvenile offense.”77 Wheel and
ad hoc assigned counsel systems generally require attorneys to petition the court for any caserelated expenses, including funds for investigation, expert witnesses, and other litigation costs.
These expenses often requires prior approval from the court and are limited by a maximum, courtdetermined payout.78
The last distinct form of assigned counsel system is known as a managed or coordinated
assigned counsel program. Rather than let the judge select an indigent defense attorney, this
decision is left to an independent body, such as “a governmental entity, nonprofit corporation, or bar
association operating under a written agreement with a county for the purpose of appointing counsel
to indigent defendants.”79 Such systems are the rarest of assigned counsel systems, largely due to
the additional administrative burdens that accompany them.80

72

Id. at 32.
Id.
74
Id. at 33.
75
Id.
76
Id.
77
Tx Iɴᴅɪɢᴇɴᴛ Dᴇғ. Cᴏᴍᴍ., Primer on Managed Assigned Counsel Programs (Sept. 2017), at
http://www.tidc.texas.gov/media/57815/tidc_primer2017.pdf, p.3; see also Scott Ehlers, Texas Indigent
73

Defense Commission: “Challenges of Implementing Fair Defense Act Requirements & Indigent Defense Grant
Opportunities” (Aug. 10, 2018). As discussed in more detail below, Harris County, Texas, where we conduct
our empirical analysis, is supposed to be a “wheel” jurisdiction. See infra notes __ and accompanying text.
78

Spangenberg, supra note __ at 33.
Tx Iɴᴅɪɢᴇɴᴛ Dᴇғ. Cᴏᴍᴍ., Primer on Managed Assigned Counsel Programs, supra.
80
See Spangenberg, supra note __ at 33.
79

12
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
2. Contract Attorneys
Contract programs differ from assigned counsel systems in that the jurisdiction “enters into
contracts with private attorneys, law firms, bar associations, or non-profit organizations to provide
representation to indigent defendants.”81 Still, in practice, many modern contract systems operate
like assigned counsel systems, with the most significant difference being that there is no rotation
requirement among attorneys in a contract system.82
An early model of contract attorney systems was the fixed price contract model, where “the
contracting lawyer, law firm, or bar association agrees to accept an undetermined number of cases
within an agreed upon contract period . . . for a single flat fee.”83 In a fixed-price contract model, the
contracting entities are responsible for the cost of investigation and expert witnesses in all the cases,
even if the caseload in the jurisdiction was higher than expected for the term of the contract.84 Fixedprice contract models are frequently faulted for their failure to take into account the time that the
attorney is expected to spend representing indigent clients, the competency of the attorney, the
complexity of the case, or the lack of support costs provided to the attorney.85
Another older model of contract attorney program is known as the fixed-fee-per-case contract
model, in which lawyers are paid by the case rather than contracting to provide an unlimited number
of cases for a flat fee.86 In this model, funds for investigation, expert witnesses, and support services
are all included in the contract.87 Just as with the fixed price model, the fixed-fee-per-case model has
been criticized for its lack of quality control.88 Nonetheless, jurisdictions are often willing to overlook
the shaky legal and ethical footing of a contract attorney system to reap the benefit of being able to
project costs by limiting the amount of money apportioned in each contract.89

3. Public Defenders
Finally, public defender systems are probably the best-known model of indigent defense
provision. Such systems employ staff attorneys who are paid a fixed salary to provide representation
to poor defendants.90 More precisely, public defender systems are often defined as “a governmental
entity or nonprofit corporation that: 1) operates under written agreement with a county rather than an
individual judge or court; 2) uses public funds; and 3) provides legal representation and services to
indigent defendants accused of a crime or juvenile offense.”91
81

Id. at 34.
Tx Iɴᴅɪɢᴇɴᴛ Dᴇғ. Cᴏᴍᴍ., Primer on Managed Assigned Counsel Programs, supra at 3.
83
Id.
84
Id.
85
Id., citing State v. Smith, 681 P.2d 1374 (Ariz. 1984), where the Arizona Supreme Court found that the fixedprice contract models used in several Arizona Counties were unconstitutional.
86
Id.
87
Id.
88
Id.
89
Id.
90
Spangenberg at 36.
91
Scott Ehlers, Tᴇxᴀs Iɴᴅɪɢᴇɴᴛ Dᴇғᴇɴsᴇ Cᴏᴍᴍɪssɪᴏɴ: “Challenges of Implementing Fair Defense Act
82

Requirements & Indigent Defense Grant Opportunities,” August 10, 2018.
13

Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
The first public defender program started in Los Angeles in 1913, but the model did not gain
widespread popularity until after Gideon and after national studies were published that supported
this approach in the 1970’s.92 Unfortunately, the increased demand for indigent defense services
beginning in the 1980’s means that many public defenders carry overwhelmingly large caseloads,
sometimes resulting in ineffective representation.93
In addition, some jurisdictions use public defenders alongside assigned counsel or contract
attorneys.94 This might be, for example, because there are insufficient numbers of public defenders
who are able to handle the flow of indigent defense cases. In some instances, public defenders
handle only a small fraction of such cases; as discussed in more detail below, Harris County, Texas,
is an example of one such jurisdiction.

B. Different Systems, Different Incentives
Apart from being poor and typically young and male, indigent defendants differ from one
another in myriad ways, such as in terms of their backgrounds, criminal histories, employment and
family status, risk preferences, and crimes for which they are charged. Still, most defendants,
whether poor or not, invariably share some predictable goals. Clearly most would like to avoid a
criminal conviction and punishment; failing that, they want to minimize any punishment they receive,
whether in terms of prison, probation, or fines. And like most people, indigent defendants would like
to avoid pre-trial detention regardless of outcome.
Defense attorneys who represent indigent defendants also share some predictable goals.
Some of those goals might align with those of their clients—for example, a desire to win a case or to
vindicate the rights of the innocent who are wrongfully accused of a crime. Other goals, however,
might not be so compatible. This in turn might open up an “incentive gap” that causes attorneys to
behave in ways that are detrimental to their clients.
This incentive gap is shaped by the particular indigent defense system in which attorneys
operate.95 As such, how this system is designed might affect, among other things: how many cases
attorneys choose to take on; how much time and resources attorneys spend on each case they are
assigned; when and whether an attorney recommends settling a case to a client and on what terms;
and the quality of lawyer who chooses to participate in such a system in the first place. We explore
these issues in more detail below.

1. Caseload Incentives for Assigned Counsel and Contract Attorneys
First, and perhaps most importantly, the design of an indigent defense system can affect
attorneys’ incentives when it comes to the size of their caseloads. In all assigned counsel and many
92

Spangenberg at 36.
Id. at 36-37.
94
See, e.g., James M. Anderson & Paul Heaton, How Much Difference Does the Lawyer Make? The Effect
Of Defense Counsel on Murder Case Outcomes, 122 Yᴀʟᴇ L.J. 154 (2012) (one in five murder defendants in
Philadelphia were randomly assigned a public defender; the rest received assigned counsel).
95
See Schwall, supra note __, at 553 (recognizing that “[t]he payment system and related incentive structure
can have a major effect on an attorney's behavior, and this impact is somewhat predictable.”).
93

14
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
contract attorney systems, payments to an attorney typically bear some relation to the number of
cases that the attorney is assigned. The payment might be, for example, a flat fee for each case that
is assigned, a flat fee for each court appearance or day worked on a case, or an hourly fee for time
worked on a case. Such systems often include caps on the maximum amount an attorney can be
paid on a particular matter.
Consider first a jurisdiction in which an attorney is paid a flat fee for each case he is assigned,
as is done in some assigned counsel and contract attorney systems96 If the flat fee is too low to
justify taking a case—such as, for example, if the attorney has more profitable outside options for
her time—then the attorney will choose not to take the case. But if the flat fee is large enough, we
can immediately see that the attorney will have an incentive to take on a large number of cases,
since each case adds to his revenue without significantly increasing costs.97 By contrast, working
more intensely on any one particular case gives the attorney no additional revenue, since he is being
compensated a fixed amount regardless of his effort or the case’s outcome. So we might expect
paying attorneys a flat fee per case would push them toward large caseloads for which they exert
little effort on any one case. In addition, we might expect attorneys to settle cases quickly, since
additional time spent on a case is just wasted from a financial perspective.
Note a paradox that emerges here: on the surface, it might seem that the large number of
cases assigned to these attorneys is the primary reason why they are unable to spend much time or
exert much effort on any one case. But the real reason these attorneys are overextended is that they
have chosen to accept an overload of cases in the first place. In this setting, it is not a lack of funds
but the poor design of the indigent defense system itself that is driving the dysfunction.
While the per-case payment scheme might be an obvious example of how a large incentive
gap might open up between attorney and client, other payment schemes introduce distortions in less
obvious ways. For example, consider when attorneys are paid based on the number of days they
spend on a case. This is, for example, how assigned counsel in Harris County, Texas, are paid, as
demonstrated in a fee table for an actual case, shown in Figure 1 below. We can see the attorney in
this case was paid $125/day for each court appearance, regardless of how long he actually spent in
court on the case itself (or regardless of how many other cases he might have handled on those
same days). Those five court appearances entitled the attorney to a total of $625 in attorneys fees.

96

In this example, and the others that follow, assume that the underlying cases are otherwise identical and
that an attorney is motivated largely, if not exclusively, by financial gain. That is, the attorney seeks to maximize
the total revenue she earns net of costs across all indigent defense cases she takes on. Attorneys in the real
world are of course not one-dimensional like this; nonetheless, our empirical analysis suggests attorneys do
respond to financial incentives, indicating the profit motive is an important determinant of attorney behavior.
97
This assumes the cost of working on a case goes up linearly as well; if costs increase in a non-linear manner,
then the attorney will eventually reach a point where adding an additional case will be more costly than the
revenue she earns from it. The general point still holds, however, that the attorney will have an incentive to
take on more cases until she reaches that point.

15
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

Figure 1
At first glance, a pay-per-day system might seem like it benefits clients, since an attorney has
an incentive to spend many days on each case, thereby increasing total wages. Nonetheless, this
incentive does not necessarily help a defendant for at least two reasons.
First, like the system where attorneys are paid by the case, attorneys under this system still
earn more money if they maximize the number of cases they take on. This, once again, leads to
concerns about defense attorneys choosing to accept an excessive workload.
Second, a daily payment system encourages attorneys to “touch” as many case files as
possible in a single day, to ensure they are eligible to be paid for that day. But the amount of work
they spent on each case might be minimal. The optimal strategy for an attorney seeking to maximize
revenues and to minimize effort is to barely do any work on a case on any given day, but to spread
this minimal work out over many days. Again, this does not work to the benefit of indigent defendants.
In fact, it could even be detrimental to a defendant who is incarcerated while awaiting trial: not only
does the attorney have little incentive to spend quality time on a defendant’s case, but now she has
an incentive to drag the case out while her client languishes in jail.
An hourly payment system would seem preferable to these two alternatives. Here, an
attorney is paid based on the amount of time she spends on a case. There is, at first glance, no
incentive to take on more cases—you can spend more time on the same case and earn the same

16
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
marginal revenue as you would by taking on a new one.98 It might even be better to work more on
an existing case, since there are fixed costs in picking up a new matter that might make it more
onerous than working an additional hour on a case that you’re already familiar with.
Again, in practice, this system does not work as well as one might hope. While in theory an
hourly system would incentivize attorneys to put in more effort on a particular case, there are typically
caps on what an attorney can charge for that case. Once an attorney reaches that cap, she has little
incentive to put in more work on that case and instead gains more revenue by seeking out fresh
cases. Hence, in practice, defense attorneys in assigned counsel systems still have incentives to
maximize their caseloads even when they are paid by the hour.

2. Caseload Incentives for Public Defenders
One might view these distortions as proof that the best way to provide indigent defense is
through a pure public defender system for all indigent defendants. Unlike assigned counsel systems,
salaries for a public defender do not depend on the number of cases they are assigned; hence, from
a financial perspective, there is no incentive to take on an excessive number of cases. And there is
some empirical evidence that suggests public defenders do, in fact, outperform assigned counsel in
terms of outcomes for criminal defendants.99
Unfortunately, public defenders must deal with a different set of forces that also push their
incentives away from those of their clients. While public defenders have no financial incentive to take
on additional cases, they often have little control over the number of cases they are assigned. If a
public defender’s office is understaffed—as is invariably true in jurisdictions across the United
States—then the supply of public defenders is insufficient to meet the influx of indigent criminal
defendants who need representation. Faced with a flood of cases, public defenders are forced to
apportion their time among those cases.100 And once again, that creates pressure to dispose of

98

There is some empirical evidence to support this point. See Schwall, supra note __, at __ (showing that
South Carolina attorneys put in less effort after a shift from hourly payments to a flat-fee system).
99
See James M. Anderson & Paul Heaton, How Much Difference Does the Lawyer Make? The Effect
Of Defense Counsel on Murder Case Outcomes, 122 Yᴀʟᴇ L.J. 154 (2012) (exploiting random assignment of
indigent murder defendants in Philadelphia to public defenders and court appointed private attorneys and
finding public defenders “reduce their clients’ murder conviction rate by 19% and lower the probability that their
clients receive a life sentence by 62%,” as well as “reducing overall expected time served in prison by 24%.”);
Radha Iyengar, An Analysis of the Performance of Federal Indigent Defense Counsel, NBER Working Paper
No. 13187 (June 2007) (exploiting random case assignment to determine that salaried federal public defenders
outperform appointed private who are compensated hourly in terms of conviction rates and sentence lengths);
see also David S. Abrams & Albert H. Yoon, The Luck of the Draw: Using Random Case Assignment to
Investigate Attorney Ability, 74 U. Cʜɪ. L. Rᴇᴠ. 1145 (2007) (exploiting random assignment in Las Vegas, NV,
and finding that being assigned a “veteran public defender with ten years of experience reduces the average
length of incarceration by 17 percent relative to a public defender in her first year.”); Fabelo, Harris County
Public Defender Survey.
100
For an eye-opening discussion on the problems facing public defender’s offices nationwide, watch Last
Week Tonight with John Oliver: Public Defenders (Sept. 13, 2015), available at
https://youtu.be/USkEzLuzmZ4.

17
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
cases quickly—or at least quicker and with less substantive investigation than if time were not a
worry.101
That public defenders are not immune from such time pressures has not gone unnoticed by
commentators and scholars.102 Notably, the American Bar Association (ABA) has addressed the
issue head on,103 stating explicitly in a report on the matter that “defender organizations, individual
defenders, assigned counsel or contractors for services . . . must take such steps as may be
appropriate to reduce their pending or projected caseloads, including the refusal of further
appointments” when they find accepting new cases or continuing with old cases “will lead to the
furnishing of representation lacking in quality or to the breach of professional obligations.”104

3. Private Attorneys
Now contrast the incentives of publicly funded attorneys with those of private attorneys—
lawyers who represent private-paying clients. The two primary models by which private attorneys
are paid are through up-front fees, in which the attorney asks a prospective client to pre-pay for
criminal defense representation, and a retainer system in which the client pays money up front into
escrow, with funds deducted from the account and transferred to the attorney as she racks up hours
on the defendant’s case. The pre-payment model is commonly used for lower stakes or more routine
cases; the retainer model is commonly used for more complex criminal cases.
The pre-payment model is like the per-case payment model described for assigned counsel;
hence, private counsel have an incentive to take on a large number of cases just like assigned
counsel do in that setting. The retainer model does not include this kind of distortion, since the
attorney is paid for each hour worked regardless of which case she works on.105
Private attorneys, like most assigned counsel, have significant control over the number of
cases they accept. But unlike assigned counsel, private attorneys do not face the same financial
incentives to take on an overwhelming number of cases. This is because private attorneys typically
operate in a market with other competing defense attorneys.

101

Worse, some have argued that given this excessive caseload, implicit bias may cause public defenders to
“triage” their caseload in ways that hurt stigmatized groups such as black defendants. See L. Song Richardson
& Phillip Atiba Goff, Implicit Racial Bias in Public Defender Triage, 122 Yᴀʟᴇ L.J. 2626 (2013).
102
Professor David Luban vividly captures the setting in which most public defenders and panel attorneys
operate: “a world of lawyers for whom no defense at all, rather than aggressive defense or even desultory
defense, is the norm; a world of minuscule acquittal rates; a world where advocacy is rare and defense
investigation virtually nonexistent; a world where lawyers spend minutes, rather than hours, with their clients;
a world in which individualized scrutiny is replaced by the indifferent mass-processing of interchangeable
defendants.” See David Luban, Are Criminal Defendants Different? 91 Mɪᴄʜ L. Rᴇᴠ. 1729 (1993).
103
See, e.g., Securing Reasonable Caseloads: Ethics and Law in Public Defense. Norman Lefstein 2011.
104
See, e.g., id. Standard 5-5. As noted in its report, the ABA intentionally used the word “must” in referring to
the need to take appropriate action when caseloads become excessive, rather than the word “should,” which
is used for all of the remaining standards from the same chapter. Id. at 38. ABA Standard 5-5 has largely
remained unchanged since it was first issued in 1980, though the issue might be revisited in future ABA
releases. Id. at 37
105
Of course, the attorney might have an incentive to overcharge in this setting by working maximally on tasks
that are “easiest,” thereby minimizing effort but that maximizing revenue.

18
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
If a private attorney fleeces a client, or attains bad results for him, this might adversely affect
her reputation and future clients might be less likely to hire that attorney.106 To the extent the market
serves as a disciplining force, it constrains private attorneys from acting too far out of line from their
clients’ interest. Since attorneys who are overloaded cannot represent their clients as effectively as
those are not, this market force constrains the number of cases that private attorneys are likely to
take on.107
By contrast, no such market force exists to discipline counsel in indigent defense cases,
whether assigned, contract, or public defender. Since indigent defendants are told who their lawyer
is rather than given the ability to select who should represent them, they are typically at the whim or
whomever is chosen by a court or independent body as their representative. And as we shall see
below, that attorney might be far from randomly chosen and instead be determined based on
financial forces that have little regard for a defendant’s well-being or best interests.

C. The Incentive Gap Magnified: Campaign Finance and Pay to Play
As we have seen, assigned counsel and some contract attorneys have financial incentives
to amass large numbers of indigent defense cases in order to maximize their revenues. But this can
happen, of course, only if the indigent defense system assigns them those cases in the first place.
In theory, we would expect a well-functioning system to limit the number of appointments that
assigned counsel can accept. While attorneys would still have incentives to “max out” this limit, if the
ceiling is set low enough, then presumably the attorneys would have sufficient time to devote to each
of their cases (assuming they were inclined to do this in the first place).
The assignment system fails, however, if counsel can exert control over how many cases
they are assigned. And unfortunately, this appears to be happening with regularity in many assigned
counsel systems. Anecdotal evidence we have collected suggests indigent defense counsel often
“buy” case assignments by providing campaign donations to assigning judges. In this section, we
describe this evidence, which comprises news articles and interviews we have conducted with
attorneys in Texas and other states. Campaign finance appears to be a tool that magnifies the
incentive gap in these jurisdictions.

1. Attorney Donations and Indigent Defense in Texas
Harris County, Texas, has struggled with indigent defense case assignments for over two
decades. Then-State Senator Rodney Ellis, now County Commissioner for Precinct 1 in Harris
County, notes a common refrain among judges during this time: “Judges who will remain nameless
106

This dynamic might be even greater today than in year’s past, since websites like avvo.com include detailed
reviews of criminal defense attorneys by their clients. One can see numerous instances where a bad review
by a client is swiftly met by an explanation or a mea culpa by the offending attorney. See www.avvo.com. Like
Yelp reviews for restaurants, one credible bad review can be devastating for business.
107
Since private criminal defense attorneys typically earn higher wages than publicly funded defense counsel,
their leisure time is worth more than publicly funded attorneys, which might cause them to substitute leisure
for more hours of work. On the other hand, if their demand for leisure increases as they get richer, which seems
likely, then this off-setting “income effect” might lower the number of cases they choose to take on. Put in
simple terms, if private attorneys are earning high enough wages, they might decide it’s not worth it to take on
an excessive number of cases even though they could earn more by doing so.

19
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
still try and tell me that the judge picking the lawyer is better [than a statewide public defender
system]. . . because they pick people who are capable. How do you say that with a straight face?”108
Attorneys practicing in Harris County echo Ellis’ question about the efficacy and ethics that
underlie indigent appointments there. In 2015, Houston criminal defense attorney, Robert Fickman,
described one of the ways that attorneys seeking appointments curry favor with the judges who
assign them cases:
There’s a certain number of court-appointed lawyers who appear to be appointed
primarily for their ability to move the docket. The trade-off is that the judge is appointing
Lawyer X to lots of cases, and in return for the appointments, Lawyer X is moving those
cases, which meets the judge’s objective.109
Sometimes the quid pro quo is even more explicit. Drew Willey, a criminal defense attorney
who practices in Houston and nearby Galveston County, told us how he witnessed pay to play first
hand while working as a new lawyer under a retired judge who took many felony appointments in
Harris County:
A few months into working on his cases, [the attorney] told us that he'd be charging us
a monthly fee out of our hourly pay to donate to [the] judge’s campaign funds. He said
these donations were necessary to keep his lights on and keep allowing him to pay us.
Me and one colleague were abhorred enough to speak up. We approached this attorney
separately, strongly voicing our opposition to this systematized pay for play conflict. He
told me and that one colleague that we were exempt from being charged his donation
fee, but continued to make the donations, openly admitting it was so that he could
continue to get more appointments in those courts. . . . [After I left his office,] I know the
pay for play continued, because that attorney later, in passing, told me that the elections
of new judges meant that he "lost" some courts and had to begin donating more to
different judges in order to keep getting appointments.110

108

See Neena Satija, How Judicial Conflicts of Interest Are Denying Poor Texans Their Right to an Effective
Lawyer, Tᴇxᴀs Tʀɪʙ. (Aug. 19, 2019), at https://www.texastribune.org/2019/08/19/unchecked-power-texasjudges-indigent-defense/, (last visited Sep. 26, 2019). A recent time-use study showed that Texas indigent
defense attorneys spend remarkably little time on cases they are assigned. Dottie Carmichael, Austin Clemens,
Heather Caspers, Miner P. Marchbanks, III, & Steve Wood, Guidelines for Indigent Defense Caseloads, Tᴇxᴀs
A&M Pᴜʙ. Pᴏʟ. Rᴇs. Iɴsᴛ. (Jan. 2015) (conducting time-use study of Texas indigent defense attorneys and
finding the following average disposal times for these classes of cases: Class B misdemeanor (4.7 hours);
Class A misdemeanor (7.6 hours); state jail felonies (10.8 hours); third-degree felonies (12.9 hours); seconddegree felonies (15.2 hours) and first-degree felonies (22.3 hours)). The report suggested that any of the
following would comprise a full-time caseload that attorneys should not exceed: “236 Class B Misdemeanors,
216 Class A Misdemeanors, 174 State Jail Felonies, 144 Third Degree Felonies, 105 Second Degree Felonies,
[or] 77 First Degree Felonies.” Id.
109
Emily Deprang, Poor Judgment, Tᴇxᴀs Oʙs. (Oct. 12, 2015) https://www.texasobserver.org/poor-judgment/
(last visited Sept. 26, 2019); see also Bright, supra note __, at __ (“Some judges appear more concerned about
cost containment and administrative efficiency than insuring a zealous defense.”).
110
Statement of Drew Willey, (Sept. 26, 2019) (on file with authors).

20
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
Willey also reports that the prevalence of pay to play adversely affects the quality of
representation that indigent defendants receive:
[T]he disparities [in caseloads among attorneys] are too obvious that pay to play is
continuing in some very specific courts. This level of corruption leads to human lives
being forgotten in jail without anyone to stand up for them. We've handled cases in
which these overloaded attorneys have not even visited their clients for 6-10 months at
a time—pretrial.111
These types of behavior are not limited to Harris County’s felony courts. A 2008 Houston
Chronicle analysis of the juvenile appointment system found, “A relatively small group of attorneys,
some of them old friends and all financial backers of judges handing out work, regularly receives
close to half of all the tax-funded appointments to represent the poor in the juvenile courts.”112 These
concerns continue today, as a recent report indicates that “a handful of private attorneys—some of
whom happened to be generous contributors to judges’ campaign coffers—got more than 300 [case
assignments]” while “juvenile public defenders received fewer and fewer appointments over several
years, so that in 2017 they each had an average load of 140 juvenile cases, which is below the
office’s imposed limit of 200 . . .”113
In addition, there have been reports of similar instances of apparent pay to play occurring in
other counties across the state. For example, in Bexar County, the fourth most populous county in
Texas and home to nearly 2 million people, a recent San Antonio Express-News editorial decried the
phenomenon of attorneys contributing to judges handing out appointments, calling it “a dynamic that
raises inherent questions of fairness in the justice system and undermines public trust.”114
Campaign finance was similarly linked to a dysfunctional indigent defense system in Travis
County, home to the state capital of Austin.115 When a Travis County judge was asked how a lawyer
she appointed to over 400 cases in a year could possibly give his clients a full defense, the judge
replied, “That’s a lot of cases . . . Lawyers have a personal responsibility. They know what they can
handle. Do we really need to tell a lawyer, ‘Don’t do that’?”116

111

Id.
Sarah Viren, “Lawyers Picked for Juveniles Give Campaigns Cash,” Hᴏᴜsᴛᴏɴ Cʜʀᴏɴ., (Apr. 20, 2008)
https://www.chron.com/news/houston-texas/article/Lawyers-picked-for-juveniles-give-campaigns-cash1759443.php (last visited Sept. 26, 2019). See also Neena Satija, Harris County Juvenile Judges and Private
Attorneys Accused of Cronyism: “Everybody Wins But the Kids”, Tᴇxᴀs Tʀɪʙ. (Nov. 1, 2018)
https://www.texastribune.org/2018/11/01/harris-county-texas-juvenile-judges-private-attorneys/ (last visited
Sept. 26, 2019).
113
Satija, supra note __.
114
Attorney Contributions to Judges Taints the System, Sᴀɴ Aɴᴛᴏɴɪᴏ Exᴘʀᴇss-Nᴇᴡs Eᴅ. Bᴅ. (Nov. 10 , 2018)
at
https://www.mysanantonio.com/opinion/editorials/article/Attorney-contributions-to-judges-taints-thesystem-13379208.php, (last visited Sep. 26, 2019).
115
See Satija, supra note __.
116
Id. See also Andrew Weber, Travis County’s Public Defender Office is Officially Funded, Aᴜsᴛɪɴ Mᴏɴɪᴛᴏʀ (
(Aug. 30, 2019), at https://www.austinmonitor.com/stories/2019/08/travis-countys-public-defender-office-isofficially-funded/, (last visited Sep. 26, 2019). In May 2019, Travis County applied for a grant to create a public
defender office. That grant was approved on August 29, 2019.
112

21
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
2. Attorney Donations in Trial Courts Elsewhere
In recent years, other states have become increasingly aware that campaign finance might
affect their trial courts’ decision-making.117 Just across the Red River from Texas in Oklahoma, a
controversy has emerged over attorneys donating to civil trial court judges that they practice before,
with one review showing that the majority of donations to district judge candidates in 2018 came from
thousands of individual attorneys.118 Claiming the system puts judicial candidates in a “weird
position,” Oklahoma defense attorney and 2018 district judicial candidate Misty Fields went so far as
to send a letter to attorneys in the area “asking for their vote, but not for ‘for public support of any
kind.’”119 Her letter notwithstanding, Fields ultimately ended up accepting donations from attorneys,
claiming it was a necessary part of the campaign:
I would be remiss to say money doesn’t matter . . . You can only do so much with free
social media and you can only knock on so many [doors]. It’s a sad part of it, but you
need some money if you want to blanket an area with signs or newspaper ads.120
A recent controversy between a prosecutor and a district court judge has exposed how
Oklahoma’s criminal trial judges might be at risk as well. In September 2019, prosecutor David Prater
publicly requested that District Judge Kendra Coleman recuse herself for failing to disclose her
campaign donors, claiming that the judge repeatedly showed favoritism toward donor attorneys and
More generally, Texas’s indigent defense system has been subject to large-scale critiques for decades. In
2000, the nonprofit Texas Appleseed undertook a statewide survey on indigent defense practices. Its final
report, released in December 2000, contained 27 findings about Texas’ indigent defense system at the time,
including: a “complete absence of uniformity in standards and quality of representation among the indigent
defense systems” in the Texas counties surveyed by Appleseed; the “unfettered discretion given to judges
over attorney selection and compensation creates the potential and appearance of a conflict of interest”; and
“costs per capita for indigent defense were approximately $4.65, making Texas near the bottom of all 50 states
for indigent defense funding.” See TX Aᴘᴘʟᴇsᴇᴇᴅ Rᴇᴘᴏʀᴛ, December 2000 at p. 7-13.
Following the issuance of Appleseed’s report on indigent defense, released in December 2000 on the eve
of the legislative session, the Texas Legislature passed the Fair Defense Act in 2001, which required officials
in all Texas counties to adopt written procedures for delivery of indigent defense services in a timely and fair
manner. The legislation created a state body to administer policies and funding across the state, the Texas
Task Force on Indigent Defense, which was renamed the Texas Indigent Defense Commission in 2011. See
TX Aᴘᴘʟᴇsᴇᴇᴅ Rᴇᴘᴏʀᴛ, December 2000 at p.1; Fabelo at 10.
117
Prior scholarly analysis of the impact of campaign finance on judicial decision-making has focused almost
exclusively on the state Supreme Court level. See supra note __ and accompanying text. See also Sue Bell
Cobb, I Was Alabama’s Top Judge. I’m Ashamed by What I Had to Do to Get There. Pᴏʟɪᴛɪᴄᴏ Mᴀɢ.,
March/April 2015, https://www.politico.com/magazine/story/2015/03/judicial-elections-fundraising-115503
(last visited Sept. 26, 2019) (lamenting the role campaign finance played in her election to the bench, and
admitting, “[T[hose of us seeking judicial office sometimes find ourselves doing things that feel awfully
unsavory.”); see also id. (“Donors want clarity, certainty even, that the judicial candidates they support view
the world as they do and will rule accordingly. To them, the idea of impartial and fair judges is an abstraction.
They want to know that the investments they make by donating money to a candidate will yield favorable
results.”).
118
Trevor Brown, Attorneys Help Bankroll Campaigns of Judges Who Hear Their Cases, Oᴋʟᴀ. Wᴀᴛᴄʜ (Oct.
8, 2018), at https://oklahomawatch.org/2018/10/08/attorneys-help-bankroll-campaigns-of-judges-who-heartheir-cases/, (last visited Sep. 26, 2019).
119
Id.
120
Id. Misty Fields would ultimately lose her district court election to Shawn S. Taylor, who got 55.2% of the
vote compared to her 44.8%.

22
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
their clients in her court.121 Highlighting the need for public trust in the courts, Mr. Prater condemned
the judge’s lack of transparency:
For each day this Court sits in judgment of parties appearing before it, no one—neither
the parties, their counsel, nor the public—can be assured that the decisions being
rendered have not been influenced by campaign contributions or other circumstances
that this Court is actively endeavoring to conceal.122
Despite such stories, many Oklahoman attorneys and the judges they donate to remain
unconvinced that the system needs to change. For example, George Gibbs, an Oklahoma attorney
who donated to a judge before whom he had an open civil case, notes, “Our state judges are more
impartial, have better temperaments and are more fair than federal judges. No judge is going to risk
their career, their standing and their reputation over a donation.”123
Michigan, too, appears to be susceptible to links between campaign finance and trial court
decision-making. A 2016 report on Michigan’s indigent defense system found that court-appointed
attorneys do very little work on their cases, with records indicating that in nearly a quarter of criminal
cases in a three-county area, an indigent criminal defendant met their attorney for the first time at
their first court appearance.124 The same report also found that court-appointed attorneys continued
to donate thousands of dollars to the campaigns of circuit court judges.125 David Carroll, executive
director of the Sixth Amendment Center, described how Michigan’s system impacts indigent criminal
defendants: “[Indigent defendants] may get a lawyer in name only, because that person has so many
cases or has financial conflicts of interest. It’s as if you’re going into court with no lawyer at all.”126
These states aren’t the only ones struggling to manage the impact of campaign finance on
their trial court systems. In 2004, the ABA Standing Committee on Legal Aid and Indigent Defendants
created a comprehensive report on the state of the right to counsel, to commemorate Gideon’s 40th

121

Nolan Clay, DA Says Judge Concealed Donors, Should Step Down from All Her Cases, Tʜᴇ Oᴋʟᴀʜᴏᴍᴀɴ, 5
(Sep. 2019), at thttps://oklahoman.com/article/5640461/da-says-judge-concealed-donors-should-step-downfrom-all-his-cases?, (last visited Sep. 26, 2019).
122
Id.
123
See Brown, supra note __.
124
Justin A. Hinkley and Matt Mencarini, Court-Appointed Attorneys Paid Little, Do Little, Records Show,
Lᴀɴsɪɴɢ Sᴛ. J., 3 (Nov. 2016), at
https://www.lansingstatejournal.com/story/news/local/watchdog/2016/11/03/court-appointed-attorneys-paidlittle-do-little-records-show/91846874/, (last visited Sep. 26, 2019)
125
Id.
126
Max Johnston, Trials and Tribulations of Court-Appointed Attorneys in Michigan, Iɴᴛᴇʀʟᴏᴄʜᴇɴ Pᴜʙ. Rᴀᴅɪᴏ
(Sep. 12, 2019), at https://www.interlochenpublicradio.org/post/trials-and-tribulations-court-appointedattorneys-michigan, (last visited Sep. 26, 2019). Former public defender Eve Primus, who has studied
Michigan’s indigent defense issues, also emphasized how poor defendants bore the brunt of Michigan’s
dysfunctional system:
It isn't the rich white people who are being dragged through this system...If you cared about
'Brown v. Board of Education' and if you cared about the way we treated minorities in this society
and you thought separate but equal was a bad thing, you should care about the criminal justice
system because that's where separate and not equal seems to be the pervasive concept.
See Hinkley & Mencarini, supra note __ .

23
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
anniversary.127 Relying on the public testimony of attorneys from around the country, the Committee
found, “In many localities, the selection and payment of counsel is still under the control of judges or
other elected officials instead of an independent authority as recommended by national
standards.”128 It continued:
Accordingly, lawyers must depend on judges to approve their compensation claims, as
well as requests for expert or investigative services. Attorneys may be removed from
court-appointed lists if they apply for fees considered by judges to be too high, creating
a disincentive to spend adequate time on a case. In some places, elected judges award
court appointments as favors to attorneys who support their campaigns for reelection.129
There is no reason to believe the situation has improved since then. In late 2014, the ABA
identified 39 states in which judges are elected, either in partisan, non-partisan or retention
elections.130 Thirty of those states ban judges from personally soliciting funds for campaigns, though
many of them allow campaign committees to solicit on their behalf131 and most allow judges to accept
unsolicited funds.
Moreover, many other states that elect trial judges permit judicial candidates to solicit
campaign funds, or have bans with significant loopholes that would allow candidates to indirectly
receive (and know they have received) donations from attorneys who appear before them. These

127

Am. Bar Ass’n Standing Comm. on Legal Aid & Indigent Defendants, Gideon’s Broken Promise: America’s
Continuing
Quest
for
Equal
Justice
(2004),
available
at
http://www.americanbar.org/content/dam/aba/administrative/legal_aid_indigent_defendants/ls_sclaid_def_bp
_right_to_counsel_in_criminal_proceedings.authcheckdam.pdf.
128
Id. at 39.
129
Id. (emphasis added)
130
Brief of Amicus Curiae The American Bar Association in Support of Respondent, Williams-Yulee v. Florida
Bar, No. 13-1499, U.S. Supreme Court (brief in support of the Florida Bar, whose ban on personal solicitation
by judicial candidates had been challenged as violated the First Amendment) (hereinafter, “ABA Brief”).
131
For example, Alaska judges stand for retention election and are prohibited from personal solicitation, though
candidates may create campaign contribution committees, which “are not prohibited from soliciting and
accepting reasonable campaign contributions and public support from lawyers.” See Ala. Canon 5(C)(3) (cited
in ABA Brief, supra note __, at Appendix 2a-3a). Some states indicate that judges should not be informed who
contributed to the campaign committee. See, e.g., Colorado Canon 4.3; Idaho Canon 5(C)(2); Minnesota Rule
4.4(B)(3); New Mexico Rule 21-402(E) (campaign committees, but not judges, may solicit lawyers with matters
pending; committee must not disclose source of campaign funds to the candidate). Cf. Louisiana Canon
7(A)(6); id. Comment [2] to Canon 7 (“A judge or judicial candidate is prohibited from personally soliciting or
personally accepting campaign contributions, but is not prohibited from knowing the identities of his or her
campaign contributors.”).

24
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
states include Alabama,132 California,133 Georgia,134 Kansas,135 Kentucky,136 Maryland,137
Missouri,138 Montana,139 Nevada,140 North Carolina,141 and Ohio.142 Each of these states also relies,

132

Alabama judges are elected and “strongly discouraged [but not prohibited] from personally soliciting
campaign contributions. It is highly recommended that a candidate establish committees of responsible
persons to solicit and accept campaign contributions, to manage the expenditure of funds for the candidate’s
campaign, and to obtain public statements of support for his or her candidacy. Such committees may solicit
and accept campaign contributions and public support from lawyers.” See Ala. Canon 7(B)(4)(a) (cited in ABA
Brief, Appendix 2a).
133
Cal. Canon 5(b)(4) (emphasis added) provides:
In judicial elections, judges may solicit campaign contributions or endorsements for their own
campaigns or for other judges and attorneys who are candidates for judicial office. Judges are
permitted to solicit such contributions and endorsements from anyone, including attorneys and
other judges, except that a judge shall not solicit campaign contributions or endorsements from
California state court commissioners, referees, court-appointed arbitrators, hearing officers, and
retired judges serving in the Assigned Judges Program, or from California state court personnel.
In soliciting campaign contributions or endorsements, a judge shall not use the prestige of
judicial office in a manner that would reasonably be perceived as coercive.
Id. See also ABA Brief, supra note __, at Appendix 4a.
134
“Georgia judges are elected in non-partisan elections[ and] may personally solicit campaign contributions
and publicly stated support.” See ABA Brief, supra note __, at Appendix 6a. But see Weaver v. Bonner, 309
F.3d 1312 (11th Cir. 2002) (encouraging judges to establish campaign committees).
135
A judicial district may choose to have elected trial judges, and judges may solicit campaign contributions.
See ABA Brief, supra note __, at Appendix 8a (noting Kansas’s prohibition on personally soliciting campaign
contributions was struck down in Yost v. Stout, No. 06-4122-JAR, 2008 WL 8906379 (D. Kan. Nov. 16, 2008)).
136
Kentucky elects judges in non-partisan elections, and the Sixth Circuit previously found Kentucky’s
solicitation clause unconstitutional. See ABA Brief, supra note __, at Appendix 8a (citing In Carey v. Wolnitzek,
614 F.3d 189 (6th Cir. 2010)).
137
“Trial court judges are elected in non-partisan elections. The Maryland Code of Judicial Conduct does not
prohibit candidates from personally soliciting or accepting contributions.” See ABA Brief, supra note __, at
Appendix 9a.
138
Missouri judges face retention elections. Missouri Rule 2-4.2(B) states: “[While] candidate[s] shall not solicit
in person campaign funds from persons likely to appear before the judge[, they] may make a written
campaign solicitation for campaign funds of any person or group, including any person or group likely to appear
before the judge.” See also ABA Brief, supra note __, at Appendix 11a.
139
Montana judges are elected in non-partisan elections and not prohibited from personally soliciting campaign
contributions. See ABA Brief, supra note __, at Appendix 12a.
140
Nevada judges are elected in Nevada in nonpartisan elections. Nevada Rule 4.2(B)(4) permits judges who
are opposed in an election to personally solicit and accept campaign contributions. See also ABA Brief, supra
note __, at Appendix 12a.
141
Judges are elected in publicly funded, non-partisan elections in North Carolina. Canon 7(B)(4) states a
judge or candidate may “personally solicit campaign funds and request public support from anyone for his own
campaign or, alternatively, and in addition thereto, authorize or establish committees of responsible persons
to secure and manage the solicitation and expenditure of campaign funds.” See also ABA Brief, supra note __,
at Appendix 14a.
142
Judicial candidates in general elections are placed on a non-partisan ballot in Ohio. The Ohio Code of
Judicial Conduct prohibits candidates and judges from personally soliciting campaign contributions in Rule
4.4(A), but numerous exceptions to this rule would make it easy for attorneys who wish to donate to do so. For
example, Ohio permits a judicial candidate to make “ a general request for campaign contributions when
speaking to an audience of twenty or more individuals,” “sign letters soliciting campaign contributions if the
letters are for distribution by the judicial candidate’s campaign committee and the letters direct contributions to
be sent to the campaign committee and not to the judicial candidate;” and “make a general request for
campaign contributions via an electronic communication that is in text format if contributions are directed to be

25
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
at least to some extent, on assigned counsel systems.143 Hence, when considering this list of states
in addition to Texas, Oklahoma, and Michigan,144 we can see campaign finance might increase the
gap between attorney and defendant interests in judicial systems across multiple states, collectively
home to tens of millions of Americans.

IV.

The Incentive Gap and Pay to Play: An Empirical Analysis of Campaign Finance
and Indigent Defense in Harris County, Texas

In this section, we provide empirical support for the claim that campaign finance exacerbates
the incentive gap in the right to counsel. Using detailed data from Harris County, Texas, we
empirically analyze the relationship between attorney donations to elected judges and appointments
to indigent defense cases. Our results are strongly consistent with the qualitative evidence we
presented above: judges appoint their donors to a disproportionate number of cases, consistent with
a system in which pay to play occurs with regularity. Moreover, if anything, donor attorneys appear
to underperform for their clients relative to non-donors, even after controlling for case and defendant
characteristics.
We begin by explaining in detail how judicial elections and court appointments work in Harris
County. Next, we describe how we constructed our linked dataset, which is the first large-scale
dataset that connects criminal court data with donations to trial court judges. We then jump into the
empirics.

A. Background on Harris County
1. Courts and Judicial Elections
The State of Texas has 254 counties, more than any other state. This translates into 252
County Courts of Law that primarily handle misdemeanor offenses in 92 counties145 and 450 District
Courts for felony offenses.146 Texas’s court system is said to be “designed to discourage the

sent to the campaign committee and not to the judicial candidate.” See also ABA Brief, supra note __, at
Appendix 15a.
143
For instance, consider California. See, e.g., Gidenonat50.org, California, at: http://gideonat50.org/in-yourstate/california/#state-independence (“California provides very little funding and almost no oversight over right
to counsel services at all levels: trial and appellate . . . . Most smaller, rural counties provide services through
flat-fee contracts with private firms or with individual attorneys for primary and for conflict services. There is no
oversight board for any trial-level system in any California county. With a few notable exceptions, most right to
counsel systems are overseen by the county’s legislative board.”).
144
Interestingly, both Oklahoma and Michigan appear to have judicial ethics codes that prohibit judges from
personally soliciting campaign contributions. See Oklahoma Rule 4.1(A)(8); Michigan Canon 7(B)(2)(a);
compare with Texas Code of Judicial Conduct Canon 4(D)(1) (permits judges to solicit “funds for appropriate
campaign or officeholder expenses as permitted by law.”). That media reports suggest that judges, in fact, do
receive campaign contributions from attorneys who practice in their court calls into question the efficacy of
such bans in terms of preventing pay to play.
145
Counties with County Courts at Law, TX Oғғɪᴄᴇ ᴏғ Cᴏᴜʀᴛ Aᴅᴍɪɴɪsᴛʀᴀᴛɪᴏɴ, available at
http://www.txcourts.gov/media/1444575/countycourtsatlaw_september2019.pdf
146
Id.

26
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
accumulation of power,”147 resulting in a system with “numerous elected officials at every level, for
every branch, [that has] shaped the state’s unstructured approach to indigent defense.”148
Our data come from Harris County, home to over 4.65 million people and the third most
populous county in the United States.149 The county comprises all of the City of Houston and a few
major suburbs, most notably Pasadena, Texas. Here we focus on the 22 District Criminal Courts
located in Harris County, which handle all state felony cases.
Like all judges in Texas,150 Harris County district judges are elected. They serve four-year
terms, with roughly half elected in November of off-presidential election years (e.g., 2002, 2006, etc.)
and the other half elected in November of presidential election years (e.g., 2004, 2008, etc.). District
judges take office on January 1 of the year following an election.
Typically, judicial elections in Harris County are not uncontested; rather, there is a party
primary followed by a partisan general election, with one Democrat running against one Republican.
In the early 2000s, Republicans controlled most of the judicial seats in Harris County. Beginning in
2008, however, Democrats found more electoral success in judicial elections as the county shifted
politically toward the left. At any rate, there have been a fair number of close general elections in
Harris County throughout the early part of this century.
Campaign finance has also been an integral part of the judicial landscape in Texas for
decades. In 1980, Texas was the first state with a judicial race that cost over one million dollars, and
the 12 candidates in the 1988 Supreme Court elections raised $12 million in total.151 The seven
winning candidates for the Texas Supreme Court from 1992-1997 each raised an average of $1.3
million, with more than 40% of those funds “contributed by parties or lawyers with cases before the
court or by contributors linked to those parties.”152 As we show below, Harris County district judges
also actively raise money for their campaigns, often from attorneys who practice before them.153

2. Indigent Defense
Like most of Texas and many other states, Harris County relies primarily on assigned counsel
to provide indigent defense, where judges are supposed to distribute cases to attorneys via a “wheel”

147

Fabelo, Harris County Public Defender Survey, at 9.
Fabelo at 9.
149
See
Uɴɪᴛᴇᴅ
Sᴛᴀᴛᴇs
Cᴇɴsᴜs
Bᴜʀᴇᴀᴜ,
American
Fact
Finder,
available
at
https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk (population estimate
for Harris County as of July 1, 2018, is 4,698,619)
150
Texas has had partisan election of judges at all levels of courts since 1876. See Judicial Selection in the
States:
Texas,
Nᴀᴛɪᴏɴᴀʟ
Cᴇɴᴛᴇʀ
ғᴏʀ
Sᴛᴀᴛᴇ
Cᴏᴜʀᴛs,
available
at
http://www.judicialselection.us/judicial_selection/index.cfm?state=TX
151
Id.
152
Id.
153
See infra notes __ and accompanying text.
148

27
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
that rotates through a list of eligible lawyers.154 Figure 2 below155 shows how prevalent such systems
are in Texas, as counties that are colored blue rely solely on assigned counsel for indigent defense
cases.
Ten counties are colored green, which means they also have public defender’s offices. This
is misleading, however, because public defenders might handle only a small fraction of indigent
defense cases in a county. This is true, for example, in Harris County, which is the large green county
in the southeast corner of the state. Although the county created a public defender’s office in 2012,
the Harris County Public Defender handles less than 3% of the cases in our sample.156 Moreover,
these public defenders are simply treated as additional attorneys to be added to the “wheel.”157
Hence for all practical purposes, Harris County relies almost exclusively on assigned private
attorneys for indigent defense.158

154

See supra notes __ and accompanying text. Currently, to be added to the master list of eligible attorneys,
licensed Texas attorneys who are in good standing must have at least two years of criminal law practice
experience, pass a certification test (or be board certified in criminal law), and satisfy certain continuing legal
education requirements. To be assigned third-degree felonies the attorney must have “[t]ried to verdict at least
three (3) felony jury trials as lead counsel.” To be assigned second-degree felonies, s/he must have “[p]racticed
criminal law for at least four (4) years and tried to verdict at least four (4) felony jury trials as lead counsel.”
And to be assigned first-degree felonies, the attorney must have “[p]racticed criminal law for at least five (5)
years, tried to verdict at least eight (8) felony jury trials as lead counsel and been accepted as competent to
receive first-degree appointments by a majority of the district judges.” The attorney must also “ [e]xhibit
proficiency and commitment to providing quality representation to criminal defendants [and] [d]emonstrate
professionalism and reliability when providing representation to criminal defendants.” Finally, the attorney must
be voted on to the master list in by a majority of district court judges. See Quick Guide, Minimum
Requirements—Harris
County
District
Courts,
at
https://www.justex.net/JustexDocuments/0/FDAMS/2017/quickGuide.pdf (last visited Sept. 29, 2019).
According to a local defense attorney, testing requirements were not always present and were added after
counsel assignments in Harris County came under increased scrutiny. See Murray Newman, Too Much for
Too Few, Life at the Harris County Criminal Justice Center (Jan. 26, 2013), at
http://harriscountycriminaljustice.blogspot.com/2013/01/too-much-for-too-few.html (last visited Sept. 27,
2019).
155
Source: Mapping American Public Defense Project, available at: https://safe-stream85568.herokuapp.com/ (last visited Sept. 1, 2019)
156
This number has been increasing over time; our data show that the Harris County Public Defender handled
just over 5% of the felony cases in 2017.
157
See supra notes __ and accompanying text.
158
See, e.g., Gidenonat50.org, Texas, at: http://gideonat50.org/in-your-state/texas/#state-independence (“The
vast majority of Texas indigent defense representation, even in those jurisdictions that have a public defender
office, is provided by appointed attorneys.”).

28
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

Source: Mapping American Public Defense Project, at: https://safe-stream-85568.herokuapp.com/

Figure 2

B. Construction of the Data
To analyze how campaign contributions might relate to indigent defense case assignments,
we connect a number of different datasets. Specifically, we combine six types of data from Harris
County: (1) criminal cases and their outcomes; (2) judicial elections; (3) campaign contributions to
judicial candidates; (4) annual revenue earned by attorneys for indigent defense appointments; (5)
attorney eligibility for indigent defense appointments; and (6) attorney bar information. Figure 3 below
summarizes the relevant information in these datasets and how we link them together.

29
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

Figure 3
Our central dataset is our criminal cases/outcomes dataset, obtained from the Harris County
District Clerk (HCC). We focus on felony cases in the dataset, which comprise the universe of all
such cases disposed in Harris County District Court from January 2005 through May 2018. The
dataset includes information on all indigent defense counsel appointments made by district judges
during that time period, as well as the characteristics of the cases to which they were assigned. In
particular, the HCC dataset includes the full name and an identifier for the appointed attorney159 as
well as the appointment type (Public Defender, Appointed Attorney, Hired Attorney); the defendant’s
full name, sex, age, race, and address; and case characteristics like the offense level and degree,
sentence type and sentence duration, and filing and disposition dates.
We combine these data with a dataset on Harris County judicial elections, provided by the
Texas Secretary of State. While the HCC case data do not include the name of the presiding district
judge, we can identify this information by looking at electoral outcomes for district court races and

159

The identifier is a System Person Number (SPN), used to identify the individual throughout the dataset. We
were able to convert the SPN number into the corresponding attorney bar number, which allowed us to merge
with attorney-related data in other datasets.

30
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
identifying each judge’s electoral history. Hence, by merging these data, we can precisely identify
which judge appointed counsel in an indigent defense case.160
Next, we connect the data to campaign contributions data, provided by the Texas Ethics
Commission (TEC). This dataset encompasses all political contributions to district court candidates
in Texas from 2004 to 2016.161 The TEC dataset includes information on contribution dates and
amounts, along with self-reported identifying information from contributors such as occupation,
employer, and zip code.162 This information lets us supplement the cases database with identifiers
for attorneys who have contributed to the appointing judge, which might happen before, during, or
after the judge’s tenure.163
The fourth dataset we connect are payments made by the State of Texas to appointed
attorneys. This information is tracked by the Texas Indigent Defense Commission (TIDC), and is
comprehensive and accurate for 2014 to 2018.164 The TIDC data include information on how much
each attorney earned in total from each court (i.e., each judge) during each fiscal year 2014-2018.

160

We also know which judicial candidates received donations from eligible attorneys but failed to be elected.
In related work, we are conducting a regression discontinuity design to measure the effect of donations on
appointments. See David S. Lee and Thomas Lemieux, Regression Discontinuity Designs in Economics, 48
J. Eᴄᴏɴ Lɪᴛ. 281 (2010). In particular, we can causally estimate the impact of donation by comparing case
appointments for attorneys who donated to judicial candidates who barely won an election versus case
assignments for attorneys who donated to candidates who barely lost.
161
Contributions data prior to 2004 are not particularly accurate and are hence excluded from our data. As a
consequence, we might underestimate the number of contributions received by judges who held office prior to
2005.
162
Some attorney contributors list themselves using the name of their private practice (e.g., John Doe might
donate as “Law Offices of John Doe”). If a law firm has just one named partner, we assign that contribution to
the named partner. When a law firm has more than one named partner, or it is unclear who the main partner
is, we do not assign such contribution to any attorney (i.e. contributions under the name of “The Law Firm of
John Doe” were attributed to John Doe, while contributions from “Do, Re, Mi and Associates” are not
attributable to anyone). Since we cannot capture these donations, we likely underestimate the number of
donations from attorneys to judges in our sample.
In addition, in 81 instances, matched attorneys contributed jointly with their spouses or someone else. In
all of these cases, only one of the names in the pair matched to someone in the list of appointed attorneys.
We drop the unmatched contributor name off our list of contributors and attribute the whole donation amount
to the appointed attorney.
163
Matching the contributors in the TEC dataset to the attorneys in the HCC dataset is not easy. This is in part
because the TEC lacks any unique identifier for political contributors, and does provide other identifying data
like mailing address and employer. We take advantage of the name variation in the HCC dataset to identify
different ways that attorneys might list their names for contribution purposes. We use these names to look for
political contributions made by attorneys who have been appointed at least once in any district court in Harris
County. For attorneys unmatched by their exact listed name, we do a second round of matching using fuzzy
matches between unmatched attorney names and contributor names using the Stata function matchit to
compare each name with the neighborhood of the subsequent 5 unmatched names in alphabetical order. After
that we look at the matches suggested and keep the suggestion only if the similarity is above a certain threshold
(.85). Finally, we manually audit the data to confirm the matches we have made are not false positives and to
search for matches we may have missed.
164
Note we could obtain reliably accurate revenue only for fiscal years 2014-2018, but our donation data
stretch from 2004-2018. Hence, the actual return on investment might be significantly higher than what we
estimate here; our estimate is likely only a floor on the real return rate.

31
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
The TIDC data do not provide granular detail as to specific attorney payments (e.g., payments
made based on the number of days an attorney appeared in court or worked on a case outside of
court). The data do, however, include attorney bar numbers, which allows us to accurately merge
the revenue data with our cases database. Hence, the HCC case data can be connected with both
attorney contributions data and attorney revenue data. This allows us to estimate a “return on
investment” for donors. That is, if pay to play exists, we can estimate the revenue that donor attorneys
received for each dollar they contributed.
Fifth, we connect our data to information on attorneys who are appointed to cases in the
district court, as compiled by the State Bar of Texas.165 This dataset includes demographic
information on attorneys, including when the attorney obtained their legal license in Texas and what
law school they attended.166 We also obtain this information for all judges in our dataset, allowing us
to test whether certain social networks (e.g., an attorney or judge attended the same law school or
are approximately the same age) might be driving results.
Finally, we combine all these data with information on when attorneys applied for approval
as indigent defense counsel, obtained from the Harris County District Court. As discussed in more
detail below, we can see whether attorney donations were correlated with their addition to the list of
eligible indigent defense counsel. For example, if pay to play exists, we might expect to see an
attorney would donate to a judge when she applies for or is approved to be added to the list of
attorneys eligible to receive indigent defense appointments.

C. Summary Statistics
1. Cases and Contributions
Together, these datasets enable us to shed light on the relationships between Harris County
district court judges and indigent defense counsel. Table 1 provides some summary statistics on our
final data.167 We cover the universe of 290,633 felony cases168 in which indigent defense counsel
were assigned and ascertainable in Harris County between January 2005 and May 2018, spread
over all 22 district courts. There were 45 different judges who assigned counsel in these cases. The
165

We thank Kyle Rozema for providing us with the Texas bar data. See Kyle Rozema, Lawyer Misconduct in
America (working paper, 2019).
166
We also merge this dataset with data from U.S. News and World Report, which provides an annually
updated, widely followed list of law school rankings. This allows us to compare the law schools attended by
donor and non-donor attorneys.
167
Our final dataset does not include information on 90 cases for which we were unable to match the attorney
to a lawyer listed in state bar attorney rolls. It also does not include 35 cases in which the listed defense
attorney was identical to the judge in the case. These discrepancies are likely due to minor clerical errors in
the cases dataset.
168
Harris County assigns a unique case number to each charged offense, even if they are actually part of the
same case. This might occur, for example, if a prosecutor charges a defendant for one crime, then dismisses
those charges completely and charges for a different crime later. To account for such scenarios, we treated
different case numbers as part of the same case if the two case numbers involved the same attorney, court,
and defendant identifier, as well as sharing either the same filing date or the same disposition date. Our results
remain robust to different aggregation methods, or if we instead conduct the same analysis at the charge-level
rather than the case-level. Moreover, the vast majority of cases (87.47%) in our sample involved just one
charged offense.

32
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
vast majority of the cases (97.30%) were assigned to private attorneys; the remainder were assigned
to a public defender.169
Attorneys who practiced before these judges also donated money to those judges. In
particular, we see that from 2004-2018,170 Harris County district court judges received a total of
$622,917 over 1,841 donations from attorneys who practiced at some point before those judges.
These donations were relatively modest in size: the average donation was $338.36, with donations
rarely exceeding $1,500.171

Next, we examine case and attorney characteristics, both in the aggregate as well as split
out by donor and non-donor attorneys. Table 2 summarizes these data below. Of the 772 attorneys
who were assigned at least one indigent defense case between 2004-2018, 25.65% (198) donated
to at least one Harris County district court judge (or candidate who was elected judge) during our
sample time frame.

169

Because public defenders are also assigned to the “wheel” system and are treated as attorneys to be
appointed in cases, we include them in our analysis below. However, our results are substantially the same if
we exclude them instead.
170
While our case data begin in 2005, we include donations data from 2004 onward since judges who began
their terms in 2005 ran for election in 2004.
171
The smallest donation in the sample was $25; the largest was $5,000.

33
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

Donors and non-donors were assigned similar types of cases, though donors on average
handled slightly more serious felonies.172 For example, first-degree felonies make up a slightly higher
share of donor cases (6.88%) as compared to non-donor cases (6.13%). And state jail felonies—
which are the least severe class of felony in Texas—make up 52.35% and 53.67% of all cases for
donors and non-donors, respectively.173
The demographic characteristics of defendants assigned to donors and non-donors are also
quite similar. Donors were slightly less likely to represent white defendants relative to non-donors
(43.72% versus 46.33%), and both groups were very similar in terms of their clients’ gender (just
under 80% male for both) and age (average age of about 33 for both).
Finally, donor and non-donor attorneys are relatively similar in terms of their educational and
experiential background. The average law school ranking of an assigned attorney in our sample is
116.35; donor attorneys attended slightly better-ranked law schools (average of 107.43) relative to
non-donors (average of 118.60). They also have more experience, as measured by years between
the attorney’s admission to the Texas bar and the filing date of a case (25.36 versus 21.09 years).

172

Even though the magnitude of most differences between donor and non-donor groups in Table 2 is small,
they are statistically significant at the 5% level, with the exception of defendant age and gender. This is due to
the large sample size we have here, which lets us detect even small differences across groups.
173
Offense level was missing in 4,092 cases (1.41% of the total sample). Those cases have been excluded
from the data when tabulating the percentages in Table 2.

34
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
2. Donations, Appointments, and Revenues
As described above, there are only slight measurable differences in the types of cases and
types of defendants assigned to donor and non-donor indigent defense counsel. Donor and nondonor defense counsel also appear largely similar in terms of their educational and experiential
background.
By contrast, there are stark differences between donor and non-donor attorneys in terms of
the number of cases assigned to them by donee judges, as well as the fees they earned from such
cases. This is clear from Table 3 below, which shows summary statistics on appointed cases and
received revenues across all attorneys, as well as split out between donors and non-donors.

As noted previously, over one-quarter of assigned attorneys contributed to at least one judge
within our sample period. But these donor attorneys were assigned over half (52.37%) of the cases
in our sample. Moreover, 20.16% of all cases in which a judge assigned indigent defense counsel
involved an attorney who donated at some point to that specific judge.
The disparity in case assignment between donor and non-donor attorneys becomes clearer
when we look at attorney-judge pairings—that is, unique pairings between judges and attorneys who
were assigned at least one indigent defense case in our sample period. Out of 10,723 attorney-judge
pairings in our sample, 1,107 (10.32%) involved an attorney who at some point donated to that judge.
On average, an attorney in an attorney-judge pairing was assigned 27.10 indigent defense
cases from that judge and earned $16,300 in indigent defense fees for those cases, which yields an
average of $601.48 per case. This shows us how indigent defense cases can be surprisingly lucrative
if, as often occurs, an attorney spends just one or two days in total on each case she is assigned.174
We can also see that an average non-donor attorney was assigned slightly fewer cases—
24.13 cases, with earnings of $13,992 (an average per case of $579.86). By contrast, the average
donor was assigned more than twice the number of cases by that judge, and earned more than twice
174

To illustrate, an attorney who handles one such case each day and works 200 days a year would earn
$120,296.

35
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
the revenue on those cases relative to the average non-donor—52.93 cases and $31,081 in
revenues. Donors earn approximately the same as non-donors per case ($587.21), which suggests
their additional revenue comes from the additional cases they are assigned, not from additional
revenue per case assigned.
We can more easily show these differences between donors and non-donors graphically, as
depicted in the histograms in Figure 4 below. As before, these graphs look at unique attorney-judge
pairs. The left panel of Figure 4 shows the natural log of cases175 assigned to attorneys who donated
to the appointing judge. The gray bars show the distribution of cases assigned to non-donor
attorneys; the white bars show the distribution of cases assigned to donors. As is apparent, the white
distribution is shifted to the right relative to the gray distribution. This means that donor attorneys
were assigned more cases than non-donor attorneys.176

Figure 4: Appointments and Revenues for Donor v. Non-Donor Attorneys
This relationship is even more apparent when you look at revenue. The right panel of Figure
4 below shows log revenue for donor attorneys (in white) versus log revenue for non-donors (in gray).
Once again, we can see the distribution for donor attorneys is clearly shifted to the right—that is,
attorneys who donated to a judge received more revenue than attorneys who did not donate to any
judge.

175

It is common to take the natural log of variables that are positive and highly skewed (i.e., have a few entries
that are very large relative to others). Often this includes revenue or income, or count variables such as number
of cases. For display purposes, logging variables also compresses the histogram into a more reasonable
range. At any rate, the results are very similar in graphs in which the standard count rather than the natural log
is used.
176
Note that when we compare how many cases a judge assigned to donor attorneys versus non-donor
attorneys, we are comparing among the class of attorneys who appeared at least once before that judge.
Hence, our results are not just a product of the judge simply being unaware of the non-donor attorneys; the
judge did in fact appoint each of the non-donor attorneys at least once during our sample period, and hence,
presumably knew who they were and could have appointed them more times.

36
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
D. Pay to Play: Regressions and Graphical Evidence
1. Donations, Appointments, and Revenues
The summary data show clear differences in assignment patterns to donors and non-donors:
in particular, donee judges assign twice the number of cases to their donors relative to non-donors.
Still, we might believe such results are explained by differences, either observable or unobservable,
across attorneys or judges.
To illustrate how this might happen, suppose that donors are just better attorneys. Then
maybe it makes sense for judges to award them more cases, not because they are donors but
because they are simply higher quality lawyers. Indeed, perhaps attorneys who can afford to give
donations are implicitly demonstrating, through their wealth, their legal prowess. If this were true, we
might expect the donor group to comprise more successful attorneys who “deserve” more
appointments.
We can show this is an unlikely explanation in a few different ways. To begin, we can use
ordinary least squares (OLS) regression analysis,177 which allows us to control for attorney- or judgespecific factors.178 Tables 4 and 5 present these results below, which show observations for all
attorney-judge pairs in which a judge assigned at least one case to an attorney during our sample
period.

177

We obtain similar results if instead we use logistic regressions for specifications with binary outcome
variables.
178
We cannot rule out that time-varying, unobservable differences across donor and non-donor cases are
driving the differences in case assignment or revenues that we see here. But we doubt this is the case, given
the size of these differences and how closely they hew to donation practices, as we discuss in more detail
below.

37
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
In Table 4, the key regressor of interest in all columns is “Donor”, a dummy variable (i.e., 0/1
variable) that =1 when the attorney was a contributor to the judge and =0 when the attorney was
never a contributor. The outcome variable in columns (1)-(3) is the total number of indigent defense
cases assigned by a particular judge to a particular attorney. In columns (4)-(6), the outcome variable
is total fees earned by an attorney across all indigent defense cases assigned by a particular
judge.179
Column (1) presents the most parsimonious specification, with no controls. The coefficient
for “Donor” is 28.79, and it is statistically significant at the 1% level. This means that donors on
average were assigned over 28 more cases by their donee judges relative to non-donors. Column
(4) translates this into dollars, as it presents the same specification but with total revenues as the
outcome variable. Here, we can see donors to an assigning judge earn on average about $17,089
more than non-donor attorneys.
The remaining columns introduce a variety of controls for attorney or judge characteristics.
Columns (2) and (5) capture observable differences across attorneys and judges, as they include
separate controls for number of years since admission to the Texas bar for the attorney and the
judge;180 controls for law schools attended by the attorney and the judge;181 a dummy variable for
whether the attorney ever served as a public defender; and a dummy variable for whether the
attorney and the judge attended the same law school.182
Columns (3) and (6) capture unobservable, time-invariant differences across attorneys and
judges. They accomplish this by including fixed effects for attorneys (i.e., controls that should capture
time-invariant attorney-specific behavior) and fixed effects for judges (i.e., controls that should
capture time-invariant judge-specific behavior).
Notably, controlling for these differences does not change the results in any substantial way.
Whether we add controls for observable differences in attorneys and judges (as in columns (2) and
(5)), or we add in judge- and attorney-fixed effects (as in columns (3) and (6)), the coefficient remains
roughly similar in magnitude and maintains its statistical significance.
We now turn to Table 5, which is identical to Table 4 except now the regressor of interest is
“Total Donated,” which is the total dollar amount that an attorney has ever donated to a judge over
179

Note that we have fewer judge-attorney pairs in these columns because we only have revenue data from
TIDC for 2014 to 2018. See supra note __ and accompanying text. If anything, that suggests our estimates of
revenues earned are underestimated, since we only have an estimate of the fees earned by attorneys during
those years and cannot measure fees earned prior to 2014. So indigent defense work is likely significantly
more profitable than we show here.
180
In addition to proxying for a judge’s legal experience, this variable indirectly captures whether the judge is
of a similar age as an attorney. This might matter, for example, if we think that judges and attorneys who are
similar in age are more likely to be friends, and that it is this friendship rather than campaign donations that
are driving the results here.
181
These are “fixed effects” for law school (i.e., separate dummy variables added for each separate law school
attended by an attorney in our sample). We obtain similar results if we instead include law school ranking as a
control.
182
Along with the experience variable, the “same law school” dummy might capture a social “network effect”—
for example, if a judge who went to The University of Texas at Austin is biased toward UT graduates, then this
dummy variable would capture that effect.

38
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
time. Column (1) again presents the most parsimonious specification, with no controls. The
coefficient in column (1) is 0.064, which is statistically significant at the 1% level. This means that
every dollar donated by an attorney to a judge is associated with 0.064 additional cases assigned to
that attorney. If we interpret this within a “pay-to-play” context, then $1/0.064 = $15.63 is the price
for an attorney to “buy” a single case from a judge.

Similarly, column (4) presents the same specification but with total revenues as the outcome
variable. Here, we can see that each dollar contributed from an attorney to a judge on average results
in $27.95 in fees from that judge. Within a pay-to-play context, this can be interpreted as a “return
on investment”—a $1 donation yields by an attorney yields $27.95 in revenues.
Like Table 4, the addition of attorney- or judge-level controls in columns (2) and (5) or fixed
effects in columns (3) and (6) do not substantially impact the results. They remain approximately the
same magnitude and are highly statistically significant. Once again, this suggests that systematic
differences between donor and non-donor attorneys, such as differences in their experience or
education, are not driving our results here.

2. Comparisons Within the Donor Class
Still, one might wonder whether our results can be explained by some difference between
donors and non-donors that cannot be easily captured by regression controls. One way to address
this claim is to focus only on the sample of donors and see whether our results remain.
Specifically, we can exploit the fact that a donor typically only contributes to some, but not
all, of the judges she appears before. Therefore, if we limit our sample only to attorneys who donated
to at least one judge, a natural comparison is to see whether donors received more case
assignments from their donee judges relative to judges to whom they did not donate. If so, that would
suggest there is something specific in the judge-donor relationship, rather than something special
about the donor herself, that is causing the judge to assign more cases to that attorney.
39
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
Once again, graphs can dramatically illustrate our results. Figure 5 below limits our sample
only to attorneys who donated to at least one judge during our sample period of 2004-2018. Like
Figure 4, the histogram on the left shows the natural log of cases, and the histogram on the right
shows the shows the natural log of revenues. The white bars again represent attorney-judge pairs
in which the attorney donated to that judge. The gray bars now show attorney-judge pairs in which
the attorney did not donate to that judge, but s/he did donate to some other judge.
As is plain, the white distribution is once again shifted to the right for both the cases and
revenues graphs. In other words, attorneys in the “donor class” received significantly more cases
(and earned significantly more revenue) from their donee judges relative to other judges to whom
they did not donate. This implies there is nothing “special” about donor attorneys themselves that
earned them additional case assignments. Rather, there is something specific to the donor-judge
relationship that is driving that result, since only judges to whom attorneys donated seem to assign
them a disproportionate number of cases.183

Figure 5: Donor Appointments and Revenues from Donee Judges v. Other Judges

183

We can also run the same regression analysis as we did earlier, but this time limited to attorneys who
donated to at least one judge. Appendix Tables A.1 and A.2 present this analysis, with Table A.1 presenting
the same specifications as Table 4, and Table A.2 presenting the same specifications as Table 5. We can see
that all of the regression coefficients remain highly significant and are similar in magnitude as in the previous
tables. This once again shows that attorney- or judge-level differences do not seem to be driving our results.
In addition, we can compare attorneys who donated more than one time to a judge (“multiple donors”) with
attorneys who never donated or donated just once to the judge. If pay to play is occurring, one might expect
that multiple donors would be assigned more cases and earn more revenues as compared to non-donors and
single-time donors. This is, in fact, what we find. In Appendix Table A.3, we see that multiple donors are on
average assigned between 35.86 and 40.03 more cases, and earn between $17,361 and $23,979 more
revenue, than non-donors and single-time donors to a judge. Appendix Table A.4 presents similar results when
we limit our sample just to attorneys who donated to at least one judge in our sample.

40
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
3. Donation Timing and “Entry Fees”
So far, we have shown that donors attorneys are assigned more cases (and earn more
revenue) than non-donors, and that attorneys who donate receive more cases (and earn more
revenue) from their donee judges relative to other judges to whom they did not donate. Given our
controls, we also know these donation patterns cannot be explained by observable differences
across attorneys or judges, such as where they attended law school, the ranking of the law school,
or the years of practice experience they might have. It also cannot be easily explained by some
“social network” arguments, such as an alumni connection to a particular school that both judges
and attorneys attended, or similarities in age or years of experience. Moreover, it cannot be explained
by unobservable differences across attorneys and judges that remain fixed over time. All of these
results are consistent with a system in which pay to play is rampant, as our anecdotal evidence
suggested.
Still, we can go further. Another approach we can use to get more evidence of pay to play is
by exploiting the timing of donations relative to appointments. As noted above, a prerequisite for
appointment as indigent defense counsel is getting added to a list of eligible attorneys. If pay to play
is occurring, one might expect donors to give their donations either just before or soon after they
become eligible to receive cases. One could view this type of donation as an “entry fee” that attorneys
must pay judges in order to receive appointments.
Since we know when counsel became eligible to be assigned cases, and we know when they
gave their first donation to a judge, we can test whether in fact this is true. Figure 6 below graphically
displays our results as a histogram. It tracks the difference in years between when an attorney first
donated to a judge, and when she first became eligible to be appointed as indigent defense
counsel.184 A negative value means that the attorney donated prior to becoming eligible to receive
cases; a positive value means that she donated after gaining eligibility. A zero value means the
attorney donated in precisely the same year she became eligible to receive cases. If an attorney
donated in an election year, then a value of 0, 1, 2, or 3 means that the attorney received her first
appointment from that judge in the four-year term following that donation.

184

In Figure 6, we limit our sample of attorney-judge pairs to judges who were elected in 2008 or later and
attorneys who became eligible for appointment in 2008 or later. This prevents data censoring issues that might
skew the graph, since we do not have donation data prior to 2004, and we do not know the exact year when
attorneys who were eligible for appointment in 2005 actually became eligible in the first place. The same
general trends typically hold, however, even if we limit our sample in other ways.

41
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

Figure 6
If donations and appointments were completely unrelated, we might expect to see a uniform
distribution—that is, a relatively flat graph from left to right. But that is clearly not what we see here.
This graph is peaked at 0, and the majority of the distribution is just to the right. This means most
attorneys who donated to a judge first donated soon after they became eligible to receive
appointments from that judge. In fact, in over half (54.7%) of all attorney-judge pairs in which the
attorney donated to the judge, the first donation to the judge occurred between 0 and 3 years
following the year the attorney became eligible to be assigned indigent defense counsel cases.
If we also include donations that occurred in the year prior to eligibility—which would be a
plausible time to donate if one wants to pay to play, since an attorney must be voted in by a majority
of judges before she is placed on the eligible list185—then the percentage of relevant attorney-judge
pairs jumps to 62.6%. These facts are again consistent with pay-to-play, where newly-eligible or
soon-to-be-eligible attorneys pay an “entry fee” to start receiving cases from that judge.

4. Case Outcomes
We might be less troubled by the differences in revenue and case assignment that we
observe if donors consistently achieve better outcomes for their clients relative to non-donors.
Indeed, this story is not completely implausible—if “pay to play” is occurring, then perhaps the
attorneys who pay are buying better results for their clients.186

185

See supra note __.
See supra note __ (noting accusation from Oklahoma district attorney that a judge was partial toward
defense attorneys who donated to her campaign).
186

42
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
The data, however, provide no evidence to suggest this is true. Table 6 below shows
summary statistics that compare whether donors and non-donors achieve various positive outcomes
for their clients. We define an outcome as “good” (Good outcome = 1) if a charge187 is dismissed or
reduced,188 or the defendant is acquitted; and an outcome as “not good” (Good outcome = 0) if the
defendant is convicted or pleads guilty or no contest.189 By this metric, non-donor attorneys do slightly
better than donors (37.45% v. 37.01%, respectively). Defendants represented by non-donors are
also less likely to end up in the Texas Department of Corrections (16.53% and 17.97% for nondonors and donors, respectively) and receive shorter prison or jail sentences on average (663.24
and 767.74 days for non-donors and donors, respectively).

A reasonable concern is that these slight differences in outcomes might be driven by
differences in case or defendant characteristics. Based on what we can observe, however, this does
not seem to be true either. We can control for various case and defendant characteristics through
an ordinary least squares regression analysis, as we present in Table 7 below.

187

We conduct our analysis here at the charge level, since whether a defendant is convicted or acquitted and
what the sentence he receives is determined at that level. Nonetheless, our results remain similar if we
aggregate up to the case level instead.
188
We can determine whether charges are reduced by comparing the offense level in the complaint (e.g., 1st
degree felony, 2nd degree felony, etc.) and the offense level when the case was disposed.
189
Some outcomes are ambiguous whether they are good or bad; we exclude those charges when defining
this variable. For example, it is ambiguous whether an outcome is good when the case is dismissed because
the defendant died or when the defendant was convicted on another charge.

43
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

The outcome variable in columns (1)-(2) is a dummy variable whether the attorney achieves
a “good” outcome as defined earlier. In columns (3)-(4), the outcome is a dummy variable for whether
the defendant receives a term of imprisonment in the Texas Department of Corrections, which is
typically where most substantial prison sentences are served. And in columns (5)-(6), the outcome
variable is the sentence of jail or imprisonment received for a charge. Odd-numbered columns
include controls for whether a defense attorney is a public defender and for the type of crime
charged;190 even-numbered columns also include dummy variables for whether a defendant was
white or female; a control for the defendant’s age; and fixed effects for the judge in the case.
We can see that generally speaking, attorneys who have donated to some judge perform
worse than those who are non-donors. If anything, donor attorneys appear between 0.05 and 0.44
190

Because outcomes depend so heavily on the type of crime that is charged, we include as a control the type
of offense charged even in our base specifications. We can distinguish between 631 different charged crimes,
often split quite finely. For example, within the narrow category of “trademark counterfeiting”, we can distinguish
between 5 different classes based on the amount of the offense.

44
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
percentage points less likely to obtain a good outcome, though the results are not always statistically
significant.191 Donors are between 0.57 and 0.75 percentage points more likely to have a client who
is sentenced to a term in the Texas Department of Corrections. And defendants represented by
donors receive jail or prison sentences that are between 18.32 and 33.71 days longer than nondonor defendants.
Of course these results should be taken with a grain of salt. It is possible that unobservable
differences across cases are driving the slightly worse outcomes we see for donor attorneys.
Nonetheless, it would be hard to conclude, based on this evidence, that the opposite holds true—
that donor attorneys are in fact doing significantly better on the whole than their non-donor
counterparts. If anything, donors appear to be underachieving for their clients relative to non-donors.
This is what we should probably expect, given the significant additional caseload that donors choose
to take on.192

V.

The Pervasiveness of the Incentive Gap: Campaign Finance and Other Reforms

The apparent prevalence of pay to play, as revealed by our quantitative and qualitative
analysis, is troubling. In part this is because a system in which attorneys buy cases from judges runs
counter to common notions of legal ethics and justice. And in part it is because, as we have shown,
donor attorneys seem to underperform for their clients relative to non-donors.
But perhaps most discomforting is that pay to play lays bare the core dysfunction in the right
to counsel: that many of those who are assigned to represent indigent defendants really do not share
their clients’ interests. More concisely, pay to play is just a symptom of the incentive gap.
The centrality and pervasiveness of the incentive gap suggests that minor changes to
campaign finance laws or case assignment rules will not eliminate pay to play. And even more
importantly, such minimal reforms are unlikely to significantly improve the quality of representation
for indigent defendants. This is because reforms that solely target pay to play do not necessarily
reduce assigned counsel’s core incentives to seek out as many cases as possible, and to work
minimally on each case they are assigned. So long as the incentive gap remains, attorneys and
judges are likely to find ways around such reforms.
These problems become apparent when we start looking more closely at some plausible
reforms that target pay to play. For example, consider a reform that would prohibit donee judges
from appointing their donors as counsel,193 if not permanently then for at least some time period
following a contribution (e.g., until an intervening election cycle has passed). Such a rule might
reduce the incidence of pay to play, as judicial donations might now cost attorneys potential revenue
191

Since about 37% of all defendants receive a good outcome, this translates into between about 0.14% to
1.19% difference between donors and non-donors.
192
One possible counterargument is that since the outcome differential is relatively small between donors and
non-donors, perhaps donors would actually do better than non-donors if they were not assigned so many more
cases. Even if this is true, however, and donor attorneys are “better” in this sense, the current system in which
donors receive a disproportionate number of cases seems likely to be suboptimal.
193
Alternatively, this same rule could be configured from the perspective of the assigned attorney: ethics rules
could prohibit him from seeking or accepting a case assignment from a judge to whom he has donated.

45
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
in future indigent defense cases.194 But it would not address situations in which attorney
appointments precede donations, which is something we observe in our data and which might
become more prevalent if such a reform is enacted. Hence, attorneys and judges might still conduct
pay to play by tacitly agreeing to transact in cases first and in donations later.
More substantial reforms could further reduce judges’ discretion to assign indigent defense
cases. Mandatory rotation policies, in which judges are supposed to assign attorneys from a rotating
“wheel,” are an example of such an approach.195 Putting aside whether such a system is optimal or
best serves defendants’ interests—there is no reason to believe, for example, that an even split of
cases across all eligible attorneys is necessarily ideal—our empirical results suggest such systems
are honored more in the breach. Harris County criminal court judges are supposed to follow a wheel
assignment system, but as we have seen, they often distribute cases highly unevenly across eligible
attorneys.
One might go further and remove the case assignment power from judges altogether. Some
jurisdictions already have such managed assigned counsel systems, which place the assignment
power with an independent commission or entity.196 But these systems face numerous political197
and practical obstacles that can limit their effectiveness in practice. For example, Travis County,
Texas (home of the City of Austin), created an independent agency to oversee indigent defense in
in 2015. But cases there have continued to be assigned in a grossly uneven manner across assigned
counsel.198
Moreover, even if trial judges are divested of the power to assign indigent defense cases,
they could find other ways to reward their donors. For example, if attorney requests for out-of-court
case preparation expenses and expert opinions must ultimately be approved by the judge—as is
often the case—then a judge would still have the opportunity to be more generous with her donors
than with non-donors in granting such requests.
Instead of focusing on attorney appointments, policymakers might instead target pay to play
through campaign finance reform. For example, policymakers could prohibit judges from personally
194

Relatedly, it would have the positive secondary effect of disincentivizing attorneys from donating to a judge
before whom they might want to practice in the future.
195
See supra notes __ and accompanying text.
196
See supra notes __ and accompanying text.
197
Texas provides a case in point. In 1999, the Texas Legislature passed a bill to create a statewide managed
assigned counsel system, but the bill was subsequently vetoed by then-Governor George W. Bush, who stated:
Senate Bill No. 247 proposes a drastic change in the way indigent criminal defendants are
assigned counsel. While well-intentioned, the effect of the bill is likely to be neither better
representation for indigents nor a more efficient administration of justice. The bill
inappropriately takes appointment authority away from judges, who are better able to assess
the quality of legal representation, and gives it to county officials. The bill creates the potential
for counties to set up a new layer of bureaucracy that could result in increased backlogs and
decreased court efficiency.
Gov. George W. Bush Veto of SB 247, June 20, 1999. Bush’s statement points out some of the political
challenges of taking away the appointment power from state judges.
198
See Neena Satija, Travis County Overhauled Legal Representation for the Poor, But Lawyers Are Still
Overwhelmed, Tᴇxᴀs Tʀɪʙ. (Apr. 26, 2018) (noting that one attorney was assigned 349 felony and 434
misdemeanor cases in 2017).

46
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
soliciting donations from practicing attorneys. Such an approach would likely pass constitutional
muster following the Supreme Court’s recent 5-4 decision in Williams-Yulee v. Florida Bar,199 in which
the Court applied strict scrutiny200 and upheld a Florida ethics rule that banned judicial candidates
from personally soliciting any campaign funds.201 The Court found that the state had a compelling
governmental interest “in preserving public confidence in the integrity of the judiciary”202 and that the
solicitation ban was narrowly tailored toward that interest.203 A restriction that merely prohibited
judicial candidates from soliciting funds from practicing attorneys would be narrower in scope than
the Florida ban204 and would thus be presumptively constitutional.
Yet, such a ban would not necessarily eliminate pay to play. Unless the ban also applied to
judicial campaign committees, attorneys could funnel money indirectly into judges’ coffers. And a
judge would still know which attorneys supported his candidacy by seeing who donated to the
committee—hence the judge could still reward those attorneys with case assignments.205 This
loophole might lead one to wonder to what extent a Florida-style ban is really more about form rather
than substance.206
Of course, one might extend the solicitation ban beyond what was considered in WilliamsYulee, and instead prohibit judges from soliciting practicing attorneys either personally or via their
campaign committees. But given that the Roberts’ Court has otherwise consistently invalidated

199

135 S. Ct. 1656 (2015).
Justices Ginsburg and Breyer argued that a less exacting standard should apply for judicial elections. See
id. at __. (Ginsburg, J., concurring) (citing Republican Party of Minnesota v. White, 536 U.S. 765, 803 (2002)
(Ginsburg, J., dissenting) (dissent from majority opinion holding that Minnesota’s “announce clauses” that
prohibit judicial candidates from stating their views on issues was unconstitutional)). The dissent agreed with
the other justices in the majority that strict scrutiny was the appropriate standard of review. See id.
201
Id.
202
Id.
203
In so finding, the Court relied in part on the fact that the Florida law allowed candidates to set up campaign
committees to solicit donations. See id.
204
Note that Texas does not appear to have any such ban. See Texas Code of Judicial Conduct Canon 6D(1)
(emphasis added):
A judge shall refrain from financial and business dealings that tend to reflect adversely on
the judge's impartiality, interfere with the proper performance of the judicial duties, exploit
his or her judicial position, or involve the judge in frequent transactions with lawyers or
persons likely to come before the court on which the judge serves. This limitation does not
prohibit either a judge or candidate from soliciting funds for appropriate campaign or
officeholder expenses as permitted by state law.
205
As noted earlier, some states allow campaign committees to solicit donations but prohibit the committees
from divulging the names of donors to candidates. See supra note __. Of course, a donor attorney could still
presumably announce to the judge that she in fact did make such a donation, thereby side-stepping such a
rule.
206
See, e.g., Eugene Volokh, Supreme Court Upholds Ban on Judicial Candidates Soliciting Campaign
Contributions—Even Via Mass Mailings, The Volokh Conspiracy, Wᴀsʜ. Pᴏsᴛ (Apr. 29, 2015) (concluding ban
might have an impact in limiting context of face-to-face personal solicitations, which might be different if
conducted by a judicial candidate vis a vis a representative of her campaign committee).
200

47
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
restrictions on campaign finance,207 and that any such regulation would be reviewed under strict
scrutiny,208 it is unclear whether these additional limitations would be found to be constitutional.
Moreover, a ban on solicitations from a campaign committee would not touch unsolicited
contributions from attorneys. The identity of judicial candidates is common knowledge; hence, an
attorney could simply send an unsolicited check directly to a candidate as a campaign donation.
Such a donation would comply with state campaign laws so long as it did not exceed rules regulating
contribution limits.209
These examples all illustrate the practical difficulties of eliminating pay to play in assigned
counsel systems. The fundamental problem is that judges and attorneys want to transact with one
another, even after reforms like those described above are enacted. Put differently, there are still
gains from trade between judges, who can provide income to attorneys, and attorneys, who can
provide financial and political support to judges. As such, these parties have strong incentives to
circumvent restrictions on contributions or case assignments.
More importantly, even if pay to play is eliminated, this will not necessarily improve the quality
of representation for poor defendants. Many indigent defense attorneys maintain outside dockets,
with paying clients. Given what our research reveals about the financial motivations of many
assigned counsel, it is unrealistic to believe these attorneys would devote as much time and effort
to their indigent clients as they would to their paying ones if there are no reputational or financial
gains from doing so. Indeed, a recent empirical analysis suggests that court-appointed attorneys in
San Antonio, Texas, generally obtain worse outcomes for their indigent clients relative to their paying
clients, even after controlling for differences in case characteristics.210 Hence, the incentive gap will
continue to plague indigent defendants so long as assigned counsel have incentives to shirk on the
effort they expend on such cases.

207

Richard L. Hasen, Election Law’s Path in the Roberts Court’s First Decade: A Sharp Right Turn but with
Speed Bumps and Surprising Twists, 68 Sᴛᴀɴ. L. Rᴇᴠ. 1597 (2016); John O. McGinnis, Neutral Principles and
Some Campaign Finance Problems, 57 Wᴍ & Mᴀʀʏ L. Rᴇᴠ. 841 (2016); see also Noah B. Lindell, WilliamsYulee and the Anomaly of Campaign Finance Law, 126 Yᴀʟᴇ L.J. 1577 (2017).
208
See supra note __ and accompanying text.
209
Another more radical approach would be to eliminate popular election of judges altogether. As of 2015, 39
out of 50 states had some form of judicial election (either partisan, non-partisan or retention elections). See
ABA Brief, supra note __. Eliminating popularly elected judges would eliminate the need for campaign funds
and the possibility of pay to play. Still, such an approach would still not eliminate the incentive gap. See infra
Section V.B.
210
See Amanda Agan, Matthew Freedman & Emily Owens, Is Your Lawyer a Lemon? Incentives and Selection
in the Public Provision of Criminal Defense, NBER Working Paper No. 24,579 (May 2018) (finding that case
characteristics and within-attorney differences across cases drive these results); see also id. at 3 (finding that
“attorneys resolve their assigned cases 13% faster than their retained cases, consistent with reduced effort.”).

48
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
VI.

Conclusion

"There can be no equal justice where the kind of trial a man gets depends on the amount of
money he has."211 Despite this pronouncement by the Supreme Court, made in the decade prior to
Gideon v. Wainwright, indigent defendants still face a system in which their interests are regularly
brushed aside by those who are supposed to represent them. As we have shown, this phenomenon
is caused at least in part by poor design: indigent defense systems often create financial incentives
for counsel that are at odds with their client’s best interests.
We illustrate the incentive gap by empirically linking, for the first time, trial court campaign
contributions with criminal court data. We find that attorneys regularly donate to trial court judges
before whom they appear. These contributions often look like “entry fees” from attorneys who have
recently become eligible for indigent defense appointments. Judges, in turn, appoint donor attorneys
to over twice as many indigent defense cases as non-donor attorneys, with donors often earning
tens or even hundreds of thousands of dollars in revenues. Importantly, our results are not driven by
unobservable differences between donors and non-donors: only judges to whom attorneys donated
assign them a disproportionate number of cases.
In addition, we find that donor attorneys in our sample are, if anything, less successful than
non-donor attorneys in achieving charge reductions, dismissals, or acquittals, and in avoiding prison
sentences for their clients. These differences are not caused solely by observable differences in
defendant or case characteristics, or by unobservable time-invariant differences. We suggest donor
attorneys might do worse for their clients simply because they have so many more cases assigned
to them.
Moreover, the phenomenon we identify here is likely widespread. While our data are from
Harris County (Houston), Texas—the nation’s third most populous county—qualitative evidence we
have collected suggests pay to play exists throughout Texas. In fact, pay to play might affect millions
of Americans in other states, such as California, Georgia, Maryland, Missouri, North Carolina, and
Ohio, among others, which also permit attorneys to contribute to judges who control indigent defense
appointments.
The apparent prevalence of pay to play in our criminal courts should be a wake-up call about
the state of indigent defense in America, and the corrosive influence of money in the judiciary. Our
assigned counsel system has failed to align the interests of defense attorneys with those of their
clients. Until we eliminate this incentive gap,212 the promise of Gideon will remain out of reach.

211

See Griffin v. Illinois, 351 U.S. 12, 19 (1956) (plurality opinion).
A complete analysis of how we might eliminate the incentive gap is unfortunately well beyond the space
limitations of the present article. It is, however, something we explore in detail in a companion piece. See Neel
U. Sukhatme & Jay Jenkins, Eliminating the Incentive Gap and Pay to Play in the Right to Counsel: Public
Defenders, Caseload Limits and Contingent Fees in Criminal Defense. There we examine possible ways to
eliminate the incentive gap such as by expanding public defender networks, imposing mandatory caseload
limits on defense attorneys, and adopting contingency fee payments in assigned counsel and contract attorney
cases. We argue that reforms such as these would alter attorney and judicial incentives in meaningful ways
that would help indigent defendants.
212

49
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)
Appendix

50
Electronic copy available at: https://ssrn.com/abstract=3611209

70 DUKE L.J. __ (2020) (forthcoming)

51
Electronic copy available at: https://ssrn.com/abstract=3611209