Deciding When To Decide: How Appellate Procedure Distributes the Costs of Legal Change

Aaron-Andrew P. Bruhl - University of Houston Law Center

Posted in , , , , , , , , , , , , , , , ,

An Introduction to the Problem

Imagine you are a judge on one of the federal courts of appeals.  You have before you a case that requires you to apply a thirty-year-old Supreme Court precedent, A v. B.  Given your role as a lower-court judge, ordinarily you would simply apply A v. B without any hesitation.  But this time you are having second thoughts because the old precedent of A v. B seems hard to reconcile with a case the Supreme Court decided just last year, X v. Y.  The X v. Y opinion harshly criticized the old case’s reasoning.  Nonetheless, the Court stated that “the precise question presented in A v. B is not before us today, and we express no opinion on it.”  The two dissenters in X v. Y wrote that it was unclear to them how A v. B could survive.

Now add to this scenario one more fact:  Today, the Supreme Court has granted certiorari in a case in which the petitioner asks the Court to overrule A v. B.  The Court will issue its decision on the merits in roughly nine months.

What should you do?  A helpful agent might suppose that the best approach is to decide every case by asking, “What would the Supreme Court do?”  If that is the right question to ask, you would be quite confident that you should not follow A v. B.  All the same, you also know that the Supreme Court itself has cautioned that lower courts should not be proactive but should instead let the Supreme Court itself deal the final blow to a staggering precedent.  So, dutifully, you follow A v. B.  Some months later, the Supreme Court inters A v. B.

When commentators discuss the problem described in the preceding paragraphs, they typically frame the issue as whether lower courts should attempt to predict future Supreme Court rulings or, instead, should strictly follow existing precedents.  That is an interesting question, but framing the problem in that way overlooks another dimension of the lower court’s predicament and neglects other ways of handling the situation.  For instance, rather than now ruling in either direction—following ailing precedent or anticipating its demise—the lower court could just delay its decision until after the Supreme Court rules.  To be sure, waiting would not always be proper: there might be an alternative basis on which to decide the case, or the particular circumstances might demand swift action despite the risks.  The point is that the lower court has to choose not just which way to decide but, at least implicitly, when to decide.

This same general problem—the problem of deciding when to decide—arises in various contexts.  Sometimes it can be a matter of life and death, as when a condemned inmate seeks a stay of execution based on the prospect of a forthcoming change in law that would call the inmate’s death sentence into question.  And although the likelihood of a change in law is most obvious and immediate when the Supreme Court has granted certiorari, the prospect of change arises in many other situations too, such as when a concurring opinion signals that a prior case is ripe for overruling, when the legislature is considering a new statute, when an administrative agency has issued notice of a forthcoming regulation, and so on.  It might not be proper for a court to delay based on these events—all of them are less powerful or immediate signals than a grant of certiorari—but that too is a choice.

The full-length Article on which this editorial is based considers how courts make these choices about timing and how they could make them better.  To put the problem of timing in its proper context, I begin by describing our system’s doctrines and institutional arrangements for managing legal change.  I then present a case study based on the experiences of appellate courts in the tumultuous six-month period between two major United States Supreme Court criminal sentencing decisions: Blakely v. Washington and United States v. Booker.  One lesson from that otherwise much-studied period is that even courts that reached the same substantive outcomes made very different procedural choices, and those choices distributed costs quite differently.  Although that episode is unusual in terms of the size and salience of the case-management problem, the same general phenomena routinely manifest themselves on a smaller scale.  Many people assume that lower courts would almost always defer their decision when the Supreme Court has a dispositive issue pending before it.  The reality is more complicated: sometimes courts wait, but sometimes they do not, and it is often hard to explain why they choose one course rather than the other. 

Having described what courts actually do, I then suggest a framework for evaluating, and perhaps improving, how courts process cases during transitional periods.  Case-management decisions are highly context specific, which makes it difficult and perhaps undesirable to formulate general rules, but we might be able to improve courts’ handling of such matters by altering the institutional environment and modifying incentives.

The Law of Legal Change

Any regime in which the law can change must have a set of rules to implement the changes.  These rules address which cases are governed by new law, who is responsible for implementing new law, and a host of other matters.  We can call this set of norms and institutional arrangements the law of legal change.  A system’s law of legal change determines how costly and disruptive legal transitions will be.

Our system has a law of legal change, even if we have typically not used that term to describe it.  Retroactivity doctrine is perhaps the best-known part of our law of legal change.  The usual rule today is that newly announced judicial precedents apply to all cases that are still pending on appeal.  (New statutes, in contrast, sometimes apply to pending cases but usually do not.)  Thus, a case that is tried under one rule can later become “wrong” in light of a new Supreme Court decision.  If the case is still pending on appeal when the law changes, it can be brought into conformity with the new law.  Once the appellate process has run its course, however, the decision is considered final and can rarely be reopened.

Other aspects of our law of legal change are less appreciated but no less important.  One of those less understood aspects—and the one that is our concern here—is the institutional problem of how to process cases when the law is about to change.  In fact, timing is especially critical in our system because of the combined effect of other aspects of our law of legal change.  In particular, our retroactivity doctrine means that the law governing a pending case is always liable to shift abruptly with a new development, even though the events underlying the lawsuit occurred years ago.  And because of the rule against prediction, the law does not change until the moment the Supreme Court says it does.  The combination of retroactivity and non-prediction generates a regime in which lower courts apply one rule today but the opposite rule tomorrow.  Therefore, the timing of lower court decisions matters a great deal.  A lower court’s decision to defer adjudication (or not to defer) determines which rule the court applies.

And what determines timing?  If the timing of a decision resulted only from happenstance and the vagaries of scheduling, we might find that state of affairs troublingly arbitrary.  In a sense, though, the problem is that timing is not the product of mere chance but instead depends in large part on deliberate choices.  Cases do not come up for decision on a conveyor belt that the courts cannot control.  On the contrary, if a change in law appears to be in the offing (perhaps due to a grant of certiorari), a lower court can hold potentially affected cases in abeyance or, less formally, just fail to act until the Supreme Court rules.  Alternatively, a court could issue a decision but suspend the decision’s finality so as to keep the case open in that court.

It is virtually impossible to say what percentage of the time courts hold cases in abeyance or otherwise alter their procedures in anticipation of a potential change in law.  The nature of timing decisions is that they have low visibility, if they are visible at all.  Based on anecdotal experience and searches of available sources, all one can say with confidence is that sometimes courts wait and sometimes they do not and that both outcomes occur with appreciable frequency.  Only rarely does one find extended discussion or public disagreements over how to proceed. Courts do not act as if there exists some law of abeyances to which they must demonstrate their faithful conformity.

Yet, to say that these choices about case handling dwell largely in zones of discretion is not to say that they lie beyond evaluation.  One notable way in which decisions about timing can be better or worse, as we explore next, is in terms of the costs of legal change.

A Case Study in Appellate Procedure and the Costs of Legal Change

In any system in which law changes, the transitions bring costs.  Depending on the details of a system’s law of legal change, the magnitude and distribution of the costs varies.  This is fairly obvious when it comes to retroactivity: a system in which changes apply retroactively will tend to have more disruptive transitions as compared to a regime of nonretroactivity.  (It might have fewer transitions too, precisely because they are disruptive.)  What I aim to show here is that the less celebrated part of the law of legal change, in particular the choice of when to decide a case, also profoundly affects the size and distribution of costs during transitional periods.  My primary example is the roughly six-month period between two major United States Supreme Court sentencing decisions: Blakely v. Washington1 and United States v. Booker.2

Building on several prior cases, the June 2004 decision in Blakely limited the role of judicial fact-finding in criminal sentencing.  Although Blakely itself concerned the sentencing law of Washington State, its real significance lay in its implications for the federal Sentencing Guidelines.  The dissenting Justices and almost all commentators thought the reasoning of Blakely applied equally to the federal Guidelines, dooming them to invalidation.  Yet Blakely had not actually invalidated the Guidelines because they were not before the Court.  In Blakely’s wake, something approaching chaos descended on the federal courts, as their daily business of criminal sentencing was thrown into uncertainty.  Responding to the mess it had made, the Supreme Court granted certiorari in Booker a few months later.  In January 2005, the Court issued its highly anticipated decision in Booker and held that the Guidelines suffered from the same defect the Court identified in Blakely.  The judge-made factual findings that dictated sentences under the Guidelines were no longer controlling; courts were to treat the Guidelines as advisory only.

Naturally enough, the question that attracted the most attention during the six-month Blakely-to-Booker interval was how various lower courts were deciding the question of the hour: whether the Guidelines remained valid.  Eventually, most courts of appeals held that the Guidelines remained binding until the Supreme Court itself nullified them.  But the question of which outcome to pick—here, thumbs up or thumbs down on the Guidelines—is only one dimension of the problem courts face during a transitional period.  Another question is when to decide: should lower courts rule at all before the Booker decision or instead just wait?  From the point of view of appellate procedure, the major divide was not whether the Guidelines would survive but whether courts would process cases as usual.

Courts differed widely on this score.  Some courts processed cases as usual despite the prospect of a forthcoming change in law.  Some tried to avoid ruling on the Guidelines’ validity by holding cases in abeyance pending the outcome in Booker.  Others took an intermediate approach, deciding cases but suspending their finality by withholding issuance of the appellate mandate.

These case-management decisions had strikingly divergent consequences.  Compare the Second Circuit and the Fifth Circuit, for instance.  Both courts upheld the Guidelines, but the similarity stopped there.  The Fifth Circuit processed cases as usual.  If defendants wanted to obtain the benefit of the potentially favorable upcoming ruling in Booker, their best procedural route was to file petitions for certiorari with the Supreme Court.  Hundreds of defendants did so.  The Supreme Court would then typically hold the petitions on its docket; once it decided Booker, it granted, vacated, and remanded (“GVR’d”) all of the affected petitions for further consideration by the lower court.  Even for defendants who negotiated the procedural hurdles (which not all did), the business-as-usual approach required the cost and hassle of a trip to the Supreme Court.  The Second Circuit, in contrast, withheld the mandates in its Guidelines cases and stored them locally.  Once Booker came down, the Second Circuit reopened the cases and ordered new sentencing proceedings as appropriate.  There was no need for a trip to the Supreme Court in the interim.  The two approaches thus had quite different effects on the implementation of the legal transition, certainly in terms of costs and sometimes in terms of ultimate case outcomes.

To be sure, the Blakely-to-Booker interval was an exceptional period.  That is one reason it is so interesting.  In more routine episodes of legal change, fewer cases are involved and systematic patterns, if they exist, are harder to detect.  Nonetheless, the Blakely-to-Booker episode is symptomatic of a general phenomenon: whenever the Supreme Court has granted certiorari on an issue that affects many cases, one will observe a variety of disparate approaches to appellate case management.  The everyday version of the problem largely escapes notice because no single episode is very momentous on its own.  An event like Blakely is valuable because it can draw attention to the more general problem of managing all legal transitions, large and small, civil and criminal.  Now that our attention has been drawn, let us see if we can develop a systematic account of how courts should deal with this problem.

What Should Courts Do?

Any attempt to evaluate courts’ performance during transitional periods, let alone to recommend reforms, needs to work from an accurate understanding of the problem.  This turns out to be surprisingly complicated.  Accordingly, the first step is just to define the nature of the decision-making problem.  Having done that, we can then consider, with due caution, some possible improvements.

Understanding the Problem

Initially, it is tempting to view the problem largely as a matter of correctly forecasting the Supreme Court’s forthcoming rulings.  On this view, the task for the lower court is to (1) predict how the Supreme Court will decide; (2) determine how the lower court itself would currently decide (taking into account constraints of circuit law, Supreme Court precedent, and other considerations that would not bind the Supreme Court); and then (3) either delay or not, depending on whether the two anticipated decisions match.

Although tempting, this prognostication model is inadequate.  In particular, it uses a flawed notion of success that neglects some of the costs of legal change.  To illustrate this point, let us return to the Blakely-to-Booker episode and suppose that things went as well as possible for a court like the Fifth Circuit.  That is, suppose that the Supreme Court surprisingly held that the Guidelines survived perfectly intact.  On the prognostication model, this result would count as a success for the Fifth Circuit.  The Fifth Circuit would look good—accurate and fast—while courts that stayed or held cases would have needlessly delayed.  But that assessment changes when one considers the institutional mechanics of legal change.  The Fifth Circuit’s “success” would not have averted the costs to the litigants and the Supreme Court that we have discussed, at least not completely.

To see why, consider the likely responses of criminal defendants whose sentences the Fifth Circuit affirmed.  If they are well counseled, they will file petitions for certiorari in order to preserve their ability to benefit from a potentially favorable ruling in Booker.  Indeed, hundreds of defendants from the Fifth Circuit filed petitions for certiorari (as did defendants from other circuits that followed the strategy of affirming sentences as usual).  Note that these defendants and their attorneys expend all of this effort even though, in this hypothetical, the Booker decision later turns out to vindicate the Fifth Circuit’s opinion.  It is true that the Fifth Circuit would not have to reconsider all of its cases in this scenario because the Supreme Court would just deny, rather than GVR, the numerous petitions for certiorari.  But a court like the Second Circuit also would not need to reconsider its cases; it could simply issue all the mandates that it had previously withheld.  The difference is that the Second Circuit, by warehousing its cases locally, would not have required litigants to go to the Supreme Court while waiting to see how Booker turned out.

Again, this insight is not limited to the admittedly unusual Blakely-to-Booker mess.  The broader lesson is that deciding a case rather than holding it in abeyance or otherwise delaying it can still impose transition costs on the litigants and the Supreme Court, even if the lower court guesses “right.”  Put differently, wrong guesses are not the only source of bad outcomes.  Any adequate notion of success should include the costs of legal change as well.

Improving the System

Making a correct timing decision requires careful consideration of a number of factors: the accuracy of lower-court predictions, the burden of delay on courts and litigants, the costs to the system of correcting errors, the parties’ expenditures to preserve their options, and so forth.  The relevant factors are likely to balance differently in different cases, such that an across-the-board solution—such as “always wait” or “never wait”—is unlikely to be optimal.  Indeed, giving courts a degree of case-by-case flexibility is desirable because it allows them to take advantage of their informational advantages: a lower court is in the best position to know, for example, whether it can decide a particular case on alternate grounds that will survive any anticipated change in law.

Although some degree of case-specificity is thus appropriate and unavoidable, there is reason to worry that courts might not do a very good job of using their discretion.  One type of impediment would concern information and information processing.  For example, even if lower courts were perfectly aware of which issues the Supreme Court had before it, lower courts’ attempts at prediction might suffer from familiar cognitive biases, such as a tendency to overestimate the chances that the Supreme Court will agree with their view of the law.

It may be, however, that the problems are of a different sort.  We are trying to identify optimal results from the point of view of the system as a whole.  But a single, impartial engineer does not make these case-handling decisions—judges do, and judges are people too.  Like the rest of us, they will be more aware of, and experience more vividly, the burdens that particular decisions impose on themselves as opposed to costs visited upon others.  They will, or at least some of them will, incline naturally toward bureaucratic paths of least resistance rather than toward creative solutions.  They might manage cases with an eye toward whether the anticipated change in the law is ideologically palatable to them.  In other words, the impediments might not be informational as much as motivational.

If we begin to think about the judges who are making these timing decisions as normal individuals, then we will quickly notice the divergence between the costs that fall on the lower court and those that fall on other courts and the litigants.  Suppose the lower court decides now and, it turns out, guesses incorrectly.  Obtaining a correct judgment will require effort by the litigants because one of them will (if properly counseled, at least) take steps to take advantage of the change in law; the other litigant will expend some effort in response.  Under current procedures, the litigant seeking relief will probably first have to go to some other court (often the Supreme Court though possibly the district court), which imposes costs on that other court.  And, again, the litigants will undertake some of these steps even in cases where it turns out that the court of appeals has correctly anticipated the Supreme Court’s eventual ruling.  The more general point is that some of the costs fall on the court of appeals but other costs fall on other courts and the litigants.  Likewise, some costs of the wait-and-see approach (such as the burden of delay to the litigants) are also external to the court of appeals.  And therein lies the problem: a divergence between private and social costs is a classic recipe for suboptimal outcomes.

The discussion above emphasizes certain administrative incentives and decision costs, but one could come up with alternative stories that emphasize other factors that could also distort incentives.  Perhaps a judge is eager to express his or her view of the law in the hope that it will sway the Supreme Court—a hope that is probably more rooted in human vanity than in reason.   Or perhaps the judge dislikes the anticipated change in law for ideological reasons and hopes to govern as many cases as possible with the old law, even recognizing that some of them might come back.

To be sure, it would be wrong to believe that judges, in making their case-management decisions, are completely ignorant of, or insensitive to, costs that fall on others.   But it is equally wrong to think that they are perfectly informed saints indifferent to their own workload and other burdens.  Again, judges are people too.  Because their private payoffs differ from social payoffs, we have reason to worry about sound decision making.  That worry is always present, of course, but it is especially troubling in a domain that is highly discretionary and virtually immune from close policing.

Although imposing fixed rules for dealing with this complicated problem is impractical and undesirable, that does not mean we are left with nothing better than an unhelpful admonition to exercise good judgment.  There is another way: arrange institutional circumstances and incentives so as to bring about better decisions.

In particular, I suggest that we give the courts of appeals more responsibility for fixing their own mistakes.  Suppose a court of appeals issues a decision despite the pendency of a Supreme Court case, and that decision quickly becomes wrong and needs fixing.  The court of appeals should be in charge of doing so, without necessitating a trip somewhere else first.  (At the level of technical detail, we might implement this reform in a few different ways, as the full-length Article on which this Editorial is based explains.)  That reform should reduce the cost and complexity of correcting improvident decisions.  In addition, it might tend to discourage the issuance of decisions that soon turn out to be wrong.  That is, ex post responsibility might heighten ex ante prudence.  At the same time, and unlike any general rule requiring the court of appeals to hold all cases, this approach would not delay cases that the court of appeals could decide on alternate grounds that would withstand any change in law.  The court of appeals is free to decide the case if it thinks its decision will hold up—but other actors and institutions will not suffer as much if it guesses wrong.


There is nothing new about legal change.  Nonetheless, I hope to have convinced you that there are new ways we can think about the problem of legal change as well as new, and perhaps better, ways of dealing with it.


Aaron-Andrew P. Bruhl is an Assistant Professor at the University of Houston Law Center.

This Legal Workshop Editorial is based on Mr. Bruhl’s Article:

Aaron-Andrew P. Bruhl, Deciding When to Decide: How Appellate Procedure Distributes the Costs of Legal Change, 96 CORNELL L. REV. 203 (2011).

Copyright © 2011 Cornell Law Review.

  1. 542 U.S. 296 (2004).
  2. 543 U.S. 220 (2005).

Post a Comment (all fields are required)

You must be logged in to post a comment.