1
LOWER COURT
DISCRETION
PAULINE T. KIM
INTRODUCTION
Court decisionmaking has long been a subject of study for
both legal theorists and political scientists. Despite the common
focus of their efforts, several quite distinct approaches have
developed. The dominant approach in traditional legal
scholarship has been normative, with scholars asking how judges
should decide important questions of law. Even when this work
has taken a positive turn by asking how judges in fact decide
cases, the work has emphasized the importance of doctrine,
focusing on court opinions as evidence of the rea- sons for their
decisions. By contrast, social scientists who study the courts have
usually emphasized positive explanations of court behavior and
focused their attention on case outcomes rather than written
opinions…prominent in recent years are positive political theories
that emphasize strategic interactions among judges and between
judges and other political actors. This approach shares the
attitudinalists’ assumption that judges seek to advance their policy
preferences; however, it posits that in doing so, they act
strategically, taking account of the likely response of other
actors and the institutional context in which they operate.
Although this approach is more closely associated with political
science, a number of legal scholars have also embraced strategic
accounts of judicial decisionmaking.
***
In considering how law impacts lower courts, I take
particular aim at the principal-agent model of the judicial
hierarchy on which most of the existing theoretical and
empirical work on lower courts is based. Positive political
theorists commonly accept that the role of lower courts is best
understood in terms of a principal-agent relation- ship. Such a
model conceives of the Supreme Court as the principal,
enunciating policy but delegating the work of implementation
to its agents, the lower federal courts. Law, in the form of court
precedents, is a signal—the means by which the Supreme Court
communicates its preferences to the lower courts, which are then
expected to follow that doctrine. The principal-agent model
focuses attention on the pos- sibility that the interests of the
principal and the agent will diverge and highlights the resulting
need for mechanisms of supervision and control by the principal.
Theorists commonly assume that the Supreme Court exercises
such control through its power of reversal. This explanation is
insufficient, however, given the extremely small percentage of
cases ultimately subject to Supreme Court review. More
importantly, a purely strategic account of law fails to take seriously
internal perspectives—those of judges and lawyers who
partici- pate in the system—on what law is and how it influences
court decisionmaking. Participants report that the law has
independent normative force: Legal rules influence how cases
2
come out, even though they may not determine the result in all
cases.
The failure to develop a more robust account of the law may
have resulted from positive scholars’ predominant emphasis
on the Supreme Court, which operates in a unique institutional
setting. The relationship of lower federal courts to law and
precedent is quite distinct from that of the Supreme Court
because of their different positions in the judicial hierarchy.
Although the principle of stare decisis constrains the Court, it is
not an “inexorable command,” and the Court has on occasion
overturned its prior precedents. The courts subordinate to the
Supreme Court, however, are subject to an absolute duty to
follow its precedents. Federal district courts are further required
to follow the precedents of the court of appeals of the circuit in
which they sit. Thus, how law impacts a judge’s decisionmaking
process will depend very much upon the particular court on which
she sits and where that court is located in the judicial hierarchy.
Despite the demand of hierarchical precedent, lower federal
courts retain a substantial amount of discretion when deciding
cases. By “discretion” I mean to indicate situations in which a
judge is required to exercise judgment because the outcome of a
case is not fully determined by existing legal materials. Even the
judge who understands legal rules to be obligatory and faithfully
attempts to follow precedent will find that she has the power to
exercise choice in deciding a case. To some extent that
discretion exists because it is unavoidable—legal language is at
some point irredeemably indetermi- nate. But discretion may
reflect certain value trade-offs as well: choosing flexibility over
certainty by selecting a standard rather than a bright-line rule; or
allocating certain powers to trial courts, rather than appellate
courts, by establishing a deferential standard of review.
Recognizing the presence and nature of that discretion is key to
understanding how law shapes the decisionmaking environment of
lower court judges and ultimately to developing better theoretical
models of the judicial hierarchy.
***
By contrast, the principal-agent model highlights the
importance of the institutional context but neglects the
discretionary spaces created by law. Scholars utilizing a
principal-agent model have not examined closely the concept
of discretion or how it fits with various theories of law. To the
extent that they write about discretion at all, they tend to use the
term as a pejorative, suggesting that the exercise of discretion
by lower courts is a form of shirking or that the existence of any
judicial discretion evidences a failure of law. In doing so, they
implicitly embrace formalist understandings, assuming that law
can and should determine the outcome in every case. This
approach ignores much recent jurisprudential thought and the
widespread understanding that discretion is inherent in the
task of applying the law in particular cases.
***
3
I
LOWER COURT DECISIONMAKING
A. The Judicial Hierarchy
…certain legal norms govern the work of judges at every level of
the hierarchy. By “norms,” I refer to rules—both formal and
informal—which are understood by actors within the system as
creating obligations to act or not act in a certain way. Thus broadly
defined, legal norms include the substantive legal rules found in the
precedential opinions, statutes, and regulatory provisions that judges
are expected to apply to the facts of the cases before them. They
also include a wide variety of both formal rules and informal
practices that govern how judges should go about the task of deciding
cases. These norms tell judges such things as which materials are relevant
to consider, how they should treat the opinions of judicial colleagues
above and below them in the hierarchy, when reasons are
required to explain a decision, and what kinds of reasons are authoritative.
Some of these norms—such as the rule of strict hierarchical
precedent—are well established; others are highly contested—for
example, the norm regarding the use of foreign legal materials.
Because at least some of the relevant legal norms depend upon a
judge’s place in the judicial hierarchy, taking account of institutional
context is crucial to understanding how judges decide.
Positive political theorists commonly use principal-agent models
to describe the judicial hierarchy, assuming that the Supreme Court
is the principal and the lower federal courts are the agents. These
models posit that judges have policy preferences and are motivated to
decide cases in a manner consistent with those preferences. When the
preferences of lower court judges diverge from those of the Supreme
Court, value conflicts arise, raising the possibility of noncompliance by
lower courts. Songer, Segal, and Cameron describe the model:
The Supreme Court is the principal, whose subordinates, the
courts of appeals, are the agents. If the circuit courts
consisted of faithful agents, they would obediently follow the
policy dictates set down by the Supreme Court. But utility
maximizing appeals court judges also have their own policy
preferences, which they may seek to follow to the extent
possible.
The existence of divergent policy preferences means that
lower court judges will have an incentive not to follow the
directives laid down by the Supreme Court—to “shirk” their
duty as agents. Principal-agent theory is thus concerned with
how hierarchical control is maintained: How and to what extent
can the Supreme Court control the behavior of lower federal
courts to ensure that its policy dictates are implemented?
The principal-agent model of the judicial hierarchy draws on
theories of economic organization. In the context of a firm, the
principal hires the agent to act on its behalf in order to reduce the
costs of coordination and increase efficiency. However, the agent
has her own interests, which may conflict with those of the
principal. The central challenge for the principal is to design an
incentive structure to ensure that the agent pursues the
principal’s objectives—a goal made more difficult by the fact
that it typically lacks complete information about the agent’s
4
efforts and the context in which she acts. Common
mechanisms of control in hierarchical organizations include
monitoring agent behavior and the use of sanctions and rewards
to induce compliance. The principal-agent model has been
extended to the political context as well, leading to theories of
how Congress and the Executive maintain control over
bureaucratic agencies to which responsibility for policy
implementation has been delegated. Regardless of the context,
the principal-agent model focuses attention on “issues of
hierarchical control in the context of information asymmetry and
conflict of interest.”
Applying the principal-agent model to the judiciary usefully
high- lights the possibility that judges may have conflicting
goals. As discussed in the next section, considerable empirical
evidence supports the notion that judges in fact are influenced
to some degree by their policy preferences and that these
preferences often diverge along ideological lines. The principalagent
model focuses attention on the issue of how value
conflicts will be resolved in the context of a hierarchical
organization and the extent to which superior courts can effectively
direct and control the actions of lower courts. It also asks
how the institutional structure of the judiciary will affect the
possibilities for and limitations on monitoring and sanctioning
lower court behavior—for example, by focusing attention on the
Supreme Court’s lack of power over traditional economic
incentives such as the salaries, job security, or promotion
prospects of its subordinates.
Principal-agent models thus offer important insights into the
decisionmaking context of lower court judges; however, the way
in which positive political theorists have utilized these models
to describe the judicial hierarchy currently suffers from certain
limitations. The following sections explore these limitations,
arguing that principal-agent models currently lack a
satisfactory account of why lower courts comply with
precedent and fail to take seriously the role of law as a
normative force.
***
C. Law as a Signal
***
Crucial to any principal-agent account is an explanation of
how the Supreme Court exercises control. It is commonly
observed that few of the usual economic incentives are
available to the Supreme Court to influence the behavior of
lower courts. Federal judges are appointed for life, and
Congress, not the Supreme Court, determines their level of
compensation. In the absence of traditional economic carrots
and sticks, principal-agent models have relied heavily on the
reversal power of the Supreme Court to explain lower court
compliance.
In its simplest form, the model posits that the risk of
reversal works as an effective sanction, inducing lower courts to
comply with higher court precedent even when their own political
preferences would lead them to a different result. While the
5
model has a certain facial plausibility, a closer examination
suggests some practical and theoretical difficulties. On a
practical level, the effectiveness of the sanction depends in part
on the actual risk of reversal. For court of appeals judges, the
chance of reversal by the Supreme Court is quite small. Even
though the Supreme Court reverses a majority of the cases it
accepts for review, it only reviews a tiny fraction of the output of
the federal courts of appeals—far less than one percent of the
appellate cases decided in a given year. As a result, the threat of
reversal is quite low and its effectiveness as a sanction
questionable. For district court judges, the analysis is further
complicated by their position at the bottom of the judicial
hierarchy. While they face a significant chance that their
decisions will be reversed by a court of appeals, the likelihood
of reversal, or even review, by the Supreme Court is extremely
low. Thus, their incentives for acting as faithful agents of the
Supreme Court seem quite attenuated. More likely, if fear of
reversal in fact motivates judges, the district court judge should be
primarily concerned with avoiding reversal by the appellate
court directly above it. Determining the preferences of the
circuit court, however, is complicated by the fact that appeals
are heard by panels of three judges whose identity is unknown
at the time that the trial court makes its decision. It is further
complicated by the possibility that the circuit court’s decisions
may be influenced by the panel judges’ own desire to avoid
reversal—either by the entire court of appeals sitting en banc
or by the Supreme Court. Given these uncertainties, it is unclear
to what extent fear of reversal motivates district courts to
comply with Supreme Court precedent.
Aware of the extremely low rate of Supreme Court review,
scholars have elaborated on the basic principal-agent model to
explain how the Court can nevertheless maintain control over
lower federal courts. For example, Songer, Segal, and Cameron
have suggested that litigant policing plays a crucial role in
insuring effective monitoring. They assert that litigants who have
lost in the lower courts are more likely to seek Supreme Court
review of noncomplying decisions, thereby sounding a “fire
alarm” that alerts the Court to cases of “fla- grant doctrinal
shirking.” In their view, this risk is sufficient to induce substantial
compliance with precedent by circuit judges. However, even
assuming that a litigant’s petition for certiorari provides an
accurate signal, the Supreme Court must still sort through
approximately eight thousand petitions each year and only accepts
about ninety of those cases for review. Thus, even when a fire
alarm is sounded, the risk of reversal remains quite low.
Knowing this, cir- cuit judges who are constrained only by the
risk of reversal might respond by increasing their level of
noncompliance, thereby diluting the effectiveness of the fire
alarm and further reducing the reversal risk in any particular
case.
***
6
D. Law as a Normative Force
If the fear of reversal is insufficient to explain judicial
behavior, then the principal-agent model presents a puzzle. In
the absence of any effective sanction, why would lower court
judges—assumed to be motivated by their policy preferences—
choose to follow legal authority rather than pursuing their
own preferred outcomes? The simplest explanation for lower
court compliance is that judges have legal preferences
independent of their political preferences. More precisely, even
if judges care about whether the outcome in a given case
advances their preferred policy, they likely care about whether it
conforms to legal norms as well. Judges may have a variety of
legal preferences regarding matters such as the appropriate
mode of interpreting statutes, or the relevance of foreign legal
materials, and these preferences may vary from judge to judge.
But their decisions are also guided by a set of widely shared
norms—some of which are formulated as legal rules—
regarding their role in the judicial hierarchy. One fundamental
and widely accepted norm requires that lower federal court
judges follow precedent established by a court directly in line
above them in the judicial hierarchy. Adherence to this norm
offers a straightforward explanation of why lower courts comply
with superior court precedent, even that with which they
disagree.
***
An account of legal preferences does not require that the
out- come in every case is determined in advance by existing
legal mater- ials. Rather, a judge’s preferences regarding legal
outcomes might be understood as an “attitude” in much the
same way that Segal and Spaeth model political preferences.
They state that “an attitude is nothing more than a set of
interrelated beliefs about at least one object and the situation
in which it is encountered.” They assume that the relevant
“attitudes” are policy goals, which explain judges’ behavior in
response to a set of cases that present similar “objects” or
“situations,” such as criminal procedure or First Amendment
issues. Judges might also be assumed to have “attitudes” toward
sets of legal materials—such as statutes or judicial opinions by
superior courts— that influence their behavior in cases that
present those sorts of materials. Even the most conservative
judge will not necessarily decide every case involving issues of
criminal procedure in favor of the government’s position; so too,
legal preferences should not be assumed to predetermine an
outcome in every case. Like political preferences, legal
preferences will produce a behavioral tendency, a correlation,
rather than an exact correspondence between attitude and
behavior.
***
Although no definitive explanation exists for the
formation of legal preferences, several plausible theories have
been advanced in the literature. Some scholars argue that the
socialization process involved in professional training or the
role perceptions of judges shape their legal preferences. Others
contend that judges have self- interested reasons for following
7
precedent, such as ensuring respect for their own decisions or
for the judiciary more generally. Judge Posner has suggested
that judges gain inherent utility from following precedent,
analogizing doctrine to the rules of a game to which they must
adhere to make the game meaningful. In any case, the assumption
that judges have legal preferences is at least as plausible as
the theory that they have policy preferences. Judges who decide
cases in ways that advance their policy preferences do not
benefit directly from those decisions, except to the extent that
they derive utility from generating outcomes consistent with their
attitudes, just as they might gain utility from decisions that
comport with legal norms.
***
CONCLUSION
Principal-agent theories have become a common method of
mod- eling relationships between courts in the judicial
hierarchy. They improve on the traditional jurisprudential
approach—focused on the abstract judge wrestling only with
legal materials—by highlighting the importance of institutional
context and the influence of policy goals. They also represent
an advance from simple attitudinal models that relied solely on
a psychological model of judging, without recognizing the
possibility that interactions with other actors in the system
might influence behavior. Thus, principal-agent models usefully
focus attention on interactions between courts within a judicial
hierarchy, emphasizing the potential value conflicts that can
occur and the challenge of aligning judges’ incentives with the
needs of a legitimate and well-functioning judicial system.
Those models fall short, however, in their failure to account for
law and legal norms. Principal-agent models typically rely on fear of
reversal to explain compliance with superior court precedent, yet both
practical and theoretical considerations suggest that such an explanation
is inadequate. As I have argued here, a more straightforward
explanation recognizes that judges likely have preferences for complying
with legal norms, as well as preferences regarding policy. Positive
political theorists have been skeptical that legal norms in
themselves motivate judges, in part because they find formalistic
accounts of the law implausible. The law, however, need not be fully
determinative in order to have a binding quality. Jurisprudential
accounts recognize that law can be both binding on judges and permit
them to exercise discretion in certain contexts. That discretion exists
not only because legal rules will inevitably be indeterminate at some
point, but also because social needs demand some measure of flexibility
in the application of legal rules, and because institutional values
argue for allocating different types of power between different levels
of the judiciary.
Of course, all models necessarily simplify a complex reality. That
simplification, however, creates the risk that some essential aspect of
the process or phenomenon under study will be lost. In the case of
principal-agent models, the failure to account for the nature of law
and legal institutions obscures important aspects of the interaction
between upper and lower courts. Principal-agent accounts emphasize
control by the superior court and evasion by “shirking” lower courts.
8
What they overlook is that lower courts sometimes have power over
law development as well as case outcomes, and, depending upon the
type of power afforded in a particular case, their goals in deciding may
include publicity and persuasion rather than evasion.