What's a federal appellate court to do when presented with a question of state law that lacks a definitive answer? Traditionally, the federal appellate court should do its best to predict the correct answer, which is defined as the answer that the highest court of the state in question would give if it were to decide the question. But in many states across the nation, including Pennsylvania, another option is available. The federal appellate court can choose to certify the question to the highest court of the state whose law is unclear or unsettled in the hope that the state court will agree to provide a definitive answer.

While certifying an unresolved, difficult question of state law to the highest court of a state eliminates the need for a federal appellate court to engage in any guesswork, the certification process presents its own significant complications. Chief among those complications are delay, substantial added costs, uncertainty over whether the state court will accept certification, and uncertainty over whether a definitive answer will in fact result at the end of the process.

When a federal appellate court decides to certify an unresolved question of state law to the highest court of a state, the certification comes after briefing has occurred, after the briefs have been considered by the three-judge panel assigned to the case, and often after the appeal has been orally argued. And the federal appellate panel must compose an opinion that sets forth the facts of the case, the question(s) being certified, and the federal court's explanation of why it views the case as appropriate for certification. While a certification request may be less elaborate than a full-blown ruling on the merits, it still takes some not insubstantial time to prepare before issuance.

After a federal appellate court asks the highest court of a state to accept for review one or more certified questions, the state court must decide whether to accept the certification. The state appellate court may allow the parties to file briefs concerning whether the certified question should be accepted for review. But even if briefs are not allowed at that initial juncture, it still takes time and effort for the highest court of a state to decide whether to accept or refuse a federal appellate court's certification request. That decision may depend among other things on the importance of the case, the importance of the certified question(s), and the workload of the state court.

If the state's highest court accepts certification, usually a new round of appellate briefs will be filed to present the case to the state appellate court, and then oral argument will occur in the state court. Thereafter, the state court of last resort will proceed to decide the case, and some of those courts take longer than others to issue decisions. Because certified questions tend to be especially difficult, often such cases defy prompt resolution by a state appellate court.

Once the state court of last resort issues its ruling, the case returns to the federal appellate court that issued the certification request. In most instances, the state court's answer is easy to apply, but sometimes it is not. A colleague has told me that once, in an oral argument before the U.S. Court of Appeals for the Third Circuit, the three-judge panel raised the possibility of certifying a question to the Supreme Court of Pennsylvania, but the presiding Third Circuit judge explained that in an earlier case, Pennsylvania's highest court took an inordinately long time to rule, and then the outcome was an evenly divided decision. You see, the Supreme Court of Pennsylvania issues full-blown opinions even in cases in which that court is evenly divided as to the result.

A recent noteworthy expression of judicial dissatisfaction with the certification of legal questions to the highest court of a state occurred in early 2003 in a case in which a divided three-judge panel of the U.S. Court of Appeals for the Ninth Circuit decided to certify to the Supreme Court of California questions concerning which party owned the rights to the Internet domain name sex.com.

Circuit Judge Alex Kozinski dissented from the certification order, explaining: "When a federal court certifies a case to a state supreme court, it draws from a limited reservoir of comity. Certifying the case shifts the difficult work of deciding it to the state court, which is often so busy keeping its own house in order that it scarcely has time for our overflow laundry. Certification also burdens litigants, forcing them to reargue the case in a different forum -- a process that is costly and full of delay."

Another highly respected federal appellate judge, First Circuit Judge Bruce M. Selya, wrote a law review article published in the Fall 1995 edition of the Suffolk Law Review in which he concluded that "certification often does not provide a means of achieving its anticipated goals, and frequently adds time and expense to litigation that is already overlong and overly expensive."

The main advantage of certifying a difficult, unresolved question of state law to the highest court of the state in question is that the highest court of the state is the definitive expositor of its state's law. Absent certification, the best that any federal court can do is provide its best guess concerning the answer that the highest state court would render.

But there are at least two other strategies that federal courts can invoke when confronting difficult, unresolved questions of state law. The first, and perhaps more controversial, step is that federal courts can and already sometimes do resolve closely contested uncertain state law questions against the party that is responsible for the case's being in federal court.

For example, a plaintiff who has the option of filing a state law claim in federal court could instead choose to file the claim in state court. If the state law claim in question is one that state courts have not previously recognized, and if the decision whether to recognize the state law claim is one that a federal court finds itself unsure how to resolve, the federal court could permissibly decide to reject the claim on the view that a plaintiff seeking recognition of a new state law claim should have brought suit in state court instead, where a definitive ruling could have been had.

The inverse of this example also can be applied. Assume that a defendant removes a state court suit involving a state law claim from state to federal court. If the defendant asserts a defense the propriety of which is unclear under state law, the federal court if uncertain how to resolve the question could decide to reject the defense on the ground that a defendant wishing to pursue a previously unrecognized state law defense should have kept the suit in state court, where a definitive ruling would have been available.

My third and final point is that federal appellate courts should provide federal trial courts with an added degree of freedom to decide for themselves whether a federal appellate court's prediction of state law remains accurate. Consider the following illustrations.

If a federal appellate court resolves an undecided question of federal law, the federal district courts under its jurisdiction must follow that precedent unless and until either the U.S. Congress changes the law, the U.S. Supreme Court rules on the question, or the federal appellate court itself changes its position on the issue. And, if a federal appellate court predicts the answer to an undecided question of state law but then the highest court of the state reaches a different result, the federal district court must then follow the ruling of the highest state court.

A more complicated question presents itself when a federal appellate court predicts the answer to a question of state law as to which no state court precedent exists but, thereafter, lower state courts reach conclusions that, while internally consistent within the state court system, conflict with the federal appellate court's prediction. In this scenario, the federal appellate court should not rigorously enforce its earlier precedent against the district court, but rather should allow the district court to decide for itself whether the intervening lower state court developments, while failing to provide a definitive resolution, nevertheless provide a reason to conclude that the federal appellate court's prediction was erroneous.

If the federal district court comes to the conclusion that the federal appellate court's prediction of state law probably was erroneous based on intervening lower state court developments, the federal district court should be free to apply its best understanding of state law, without regard to otherwise controlling precedent from its own federal court of appeals.

It is a part of the law's majesty that even in the year 2003 so many important questions of both federal and state law have yet to be definitively resolved and can be intelligently debated on both sides. Moreover, cases governed by state law make up a substantial and important part of the caseload of federal trial and appellate courts, and many federal judges receive great satisfaction from presiding over such diversity cases.

In struggling to predict the correct answer to undecided questions of state law, federal appellate courts should exercise great restraint in deciding whether to certify questions to the highest court of a state. In especially close cases, in the absence of certification it may be appropriate to resolve such questions against the party that caused the case to be present in federal court. Finally, federal appellate courts should encourage the trial courts under their jurisdiction to decide unsettled questions of state law correctly based on all the information currently available when the question arises for decision, instead of applying a rigid form of intra-federal court precedent that remains appropriate in cases governed by federal law.