A Blameworthiness Test of Collateral Estoppel
As we have seen, a daunting challenge for a same-offense theory is to explain the disparate ways in which two offenses can be the same offense. In chapter 6 I argued that a blameworthiness account unifies the same-offense question, whether it arises in multiple violations of a single statute or in violations of different statutes and whether it arises in one trial or two trials. The next task is to explain the collateral estoppel dimension of the same-offense problem. No one has attempted to unify collateral estoppel with the traditional same-offense issues. That is my goal in this chapter.
The issues are superficially different: double jeopardy collateral estoppel, like its civil-law sibling, is narrowly focused on issue preclusion. A comparison of offense descriptions or an analysis of the scope of the prohibited act-type is quite beside the point. Indeed, if the defendant prevails on a same-offense claim, collateral estoppel is literally beside the point. Because collateral estoppel is a separate, supplemental protection, it could be located in the Due Process Clause, as Amar and Marcus contend.1 But collateral estoppel is conceptually part of a blameworthiness account of double jeopardy. It is a way of showing singular blameworthiness and thus can be viewed as implicit in the same-offense language of the Double Jeopardy Clause.
The collateral estoppel claim is that an earlier proceeding has resolved in the defendant’s favor an issue that is necessary for the second prosecution. Although offense 1 and offense 2 do not share the quality of offense sameness, an acquittal of offense 1 has resolved an issue that makes the defendant not guilty of offense 2. The parallel to autrefois acquit is obvious. Indeed, Friedland has noted that issue preclusion is often implicit in the way autrefois acquit is defined.2
Res judicata has been part of English law at least since the twelfth century, when it appeared in Glanville’s Treatise: “A concord … is called final because it puts an end to the matter, so that neither litigant may in future depart from it.”3 Imprisonment was one penalty for disregarding a “final concord” in that era.4 The procedure by which a party could claim “former concord” in Glanville’s England focused on the record of the first proceeding: “When the justices have come to court and are in full agreement as to the record, it is necessary, as was said above, to abide by their record, and neither party may object to it.”5
Issue preclusion is an implication of res judicata, one that focuses on issues settled by the earlier judgment. In criminal law, it is a one-way street, benefiting the defendant but not the state. Though some commentators and at least one English court have suggested that a conviction can be used as conclusive evidence against the accused in a later case,6 it is generally accepted that issue preclusion does not operate against criminal defendants in this country. Part of the reason is that defendants have the right to confront witnesses and the right to a jury trial.7
While issue preclusion is unrelated to same offense in a formal sense, it is related to blameworthiness. Assume R claims alibi and mistaken identity in defending against the charge of robbing V. The state then indicts R for robbing Z, who was standing next to V. There is no question that robbery of Z is a different offense from robbing V. Nonetheless, on the facts of R’s particular case, the second robbery is the same offense as the first, in the sense that proof of innocence of one automatically proves innocence of the other.
Collateral estoppel can, therefore, be conceptualized as a case-sensitive same-offense claim. If the jury finding as to offense 1 necessarily means the defendant cannot be guilty of offense 2, then 1 and 2 must come out the same way whether or not the offenses share a definitional identity. The matter is different, of course, if R’s defense to the robbery was that the money he took from V belonged to him. In that case, the acquittal cannot be used to prove innocence of robbing Z, and they are not effectively the same offense.
This understanding of collateral estoppel is implied by a blameworthiness conception of same offense. The legislature creates statutory blameworthiness, some of it criminal and some of it civil. But the legislature does not authorize criminal penalties or civil sanctions in the absence of a finding that the actor was the one who brought about the blameworthy state of affairs. A finding that the actor did not bring about the blameworthy state of affairs means that he should suffer no penalty, civil or criminal.
But there are doctrinal problems. When the border is crossed and re-crossed between civil and criminal, as we saw in chapter 4, the comparisons are often difficult. For example, the same parties are inevitably involved when traditional double jeopardy is invoked—the same criminal defendant8 and the same state or federal government. A same-party requirement partly animates the Court’s dual sovereignty doctrine.9 One reason to permit different governments to prosecute the same offense is the fear that the interests of one sovereign will not be vigorously advanced by the other. This is a concern in civil res judicata as well. Should criminal collateral estoppel generally require the same parties?
In the 1766 case of Rex v. Duchess of Kingston,10 the issue was whether the finding of an ecclesiastical court that the duchess was not married was conclusive evidence, in the king’s court, that the duchess was not guilty of bigamy. The Lord Chief Justice of the Court of Common Pleas, Sir William De Grey, noted that the judgment of a court “directly upon the point” was binding on all other courts before whom the same parties raised the same matter, even if raised for a different purpose.11 But the king was not a party to the ecclesiastical proceeding, and the duchess lost her collateral estoppel claim. Supporting its judgment, the court noted that the king had responsibility for protecting the public peace, and ecclesiastical courts lacked jurisdiction in any matter not “altogether spiritual.”
An early Supreme Court opinion was not so concerned with the identity of parties. In Gelston v. Hoyt,12 the Court dealt with the preclusive effect of an acquittal in an in rem forfeiture case. An in rem proceeding is one begun against an object itself, in this case a ship, rather than a person. There is no necessity to have jurisdiction over the owner of the object because the suit is against the object itself. The in rem action against the ship had resulted in an acquittal, and the owners of the ship were suing the seizing officers for tort damages. The Gelston issue was whether the defendants (the seizing officers) could litigate the question of whether the ship had been properly seized.
Perhaps the wrongfulness of the seizure had been settled by the acquittal of the ship in the original forfeiture action. Because the ship was acquitted, the seizure must have been wrongful. The other possibility is that the original acquittal might not be binding, for one reason or another, on the defendants in the new tort action. After all, they were not parties to the original forfeiture action, and they had never themselves litigated the question of the ship’s seizure. Should a court bind them to the earlier judgment even though they did not participate in it?
Yes, the Court held in Gelston. How were the interests of the seizing officers represented in the original action? The officer “virtually identifies himself with the government itself, whose agent he is, from the moment of the seizure up to the termination of the suit.”13 The Court went further, noting in dicta that even if the seizing officer “were a mere stranger, he would still be bound by such sentence, because the decree of a court of competent jurisdiction in rem is, as to the points directly in judgment, conclusive upon the whole world.”14
The English court in Kingston did not have a similarly expansive view of collateral estoppel. In one way, the initial proceeding in the ecclesiastical courts was similar to an in rem proceeding. The duchess initially brought suit against a man who claimed to be her husband, asking the ecclesiastical court to determine that she was “free from all matrimonial contracts or espousals.”15 The court held in her favor. As the issue was the validity of the marriage—a determination of legal status—it might seem indistinguishable from an in rem proceeding, which seeks to determine the legal status of property. On this view, the judgment in favor of the duchess in the ecclesiastical courts would be “conclusive upon the whole world.”
An analogous claim would be insanity. In Hotema v. United States,16 the defendant killed three people on the same day. Two of the three indictments were consolidated for trial, Hotema defended on the ground of insanity, and the jury acquitted. The same defense was presented to the third indictment, but the second jury could not agree on a verdict, and a mistrial was granted. A new jury was impaneled, and this time Hotema was convicted.17
As to collateral estoppel, all three killings took place on the same day, creating a plausible collateral estoppel argument that the first jury had determined Hotema’s status as someone who lacked capacity to commit a crime. It is unlikely that Hotema could have been alternately sane and insane during the same day. The Supreme Court rejected Hotema’s appeal but not the collateral estoppel claim, which Hotema apparently did not raise. He instead argued that the trial for the first two killings was a same-offense bar to the third indictment. This argument is obviously flawed, as we have seen; three killings are three different offenses in all but the most unusual circumstances.
Properly presented, Hotema’s collateral estoppel claim should have won. His claim can be distinguished from the marital status question on which the duchess lost because in Hotema’s case the same parties litigated the same issue in both trials. The king in Kingston was not a party to the first suit, nor was the king on constructive notice of the litigation of the marital status of the duchess, which would be true in an in rem proceeding.
The same-parties issue puts two principles directly in conflict. The first, argued by the duchess in Kingston, is that once an issue is determined by a court of competent jurisdiction, the matter is at an end. No reason exists to prefer the finding of a subsequent court to the initial court. Thus, the initial finding can be accepted as the truth and, therefore, as binding on the whole world. The second principle, inherent in the requirement of same parties, is that fairness should give every person who has standing to raise an issue the opportunity to litigate it. The in rem proceeding in Gelston is arguably a substitute for same parties; the requirement that the property be served with notice presumably puts on notice everyone with standing to defend the claim.
The English court followed the second principle. As a matter of double jeopardy, rather than fairness, the Kingston result is suspect. Perhaps it is only because bigamy is not a criminal offense that deeply disturbs the modern conscience; perhaps it is that the reliance of the duchess on the ecclesiastical judgment seems justified on the facts of the particular case—what court could better determine what was a matter of church law? Whatever the reason, a determination by a court that specializes in questions of marriage that the duchess was not married should have been an end to the matter.
A few years after Hotema, the Supreme Court accepted a version of the duchess’s argument. In United States v. Mason,18 the defendants were charged with a federal conspiracy that provided enhanced punishment if state crimes were committed during the course of the conspiracy. Because the defendants had been acquitted of the murder that the government wanted to use for enhanced sentencing, the Court upheld the defendant’s plea in bar with an analysis that could have been used by the duchess:
The murder in question, if committed at all, was, as a distinct offense, a crime only against the state; and after the defendants were acquitted of that crime by the only tribunal that had jurisdiction of it as an offense against the state, it is to be taken that no such crime of murder as charged in the indictment was in fact committed by them.19
To be sure, Mason can be distinguished on the technical ground that the king’s court had concurrent jurisdiction over determining marital status, at least where bigamy was the issue, while the federal court had no jurisdiction to determine whether state crimes had been committed. This is a formalistic response to the question of whether “same parties” is always a prerequisite to a valid collateral estoppel claim. The same parties were not involved in Mason, and the Court upheld what is in essence a collateral estoppel claim.
Kingston could happen today. Suppose W sues H for alimony in a state where alimony can be awarded only if marriage is proven, and a civil court determines that W and H were never married. Then the state charges H with the criminal offense of bigamy, based on a claim that H was married to W. This is, again, a status determination, which is somewhat similar to the in rem proceeding in Gelston. Modern collateral estoppel would not protect H, though it is difficult to see why the state should have a chance to prove a different legal status than W was able to prove.
There is, of course, the fear that H and W will collude to obtain a judgment that is a fraud upon the court. This fear raises an accuracy concern about the first finding and would be a reason to prefer another principle if the fear were well founded. The king argued in Kingston that the duchess might be guilty of collusion concerning the judgment in the ecclesiastical court. Should we deny preclusive effect to all judgments because a small percent might be flawed?
One problem with any attempt to extend the Gelston in rem principle to status judgments is that it threatens to abolish entirely the same-parties requirement. Suppose the parents of V sue X for the tort of contributing to the delinquency of V, a minor. In the tort case, X defends by seeking to prove that, when the relevant events occurred, V was no longer a minor. The jury returns a verdict for X. This verdict determines V’s legal status at the critical point in time and, on the analysis just given, could be seen as a bar to any future effort to determine that particular legal status by a criminal charge. But what if X defended on the grounds that, though V was a minor, his acts did not constitute contributing to the delinquency of a minor? If the jury acquits, it is not a determination of a legal status but purely a finding of fact in X’s favor.
Why should we distinguish between determining a legal status, which requires findings of fact, and a pure finding of fact? As no reason suggests itself, we must either reject any attempt to extend Gelston, or reject the same-parties requirement generally. We return to this question after examining the other collateral estoppel issues. One issue is whether a civil judgment can be used in a criminal case.
There is no double jeopardy harm, indeed there is no double jeopardy, in the absence of two jeopardies of life or limb or two convictions of life or limb offenses. Double jeopardy guarantees a single blameworthiness determination for a single offense. Thus, on the chapter 4 argument, if either of two offenses is not a “life or limb” offense, the state may seek a judgment in both.
The collateral estoppel aspect of double jeopardy is a specialized application of that blameworthiness principle. Here, the protection centers not on same offense but on the process as a whole. On a robust view of this blameworthiness principle, a defendant may offer res judicata from a civil proceeding to bar a criminal trial. From a blameworthiness perspective, a finding in the defendant’s favor on an issue required for “life or limb” blameworthiness necessarily determines that issue for all time, whether or not that determination arose in a “life or limb” proceeding.
This means, of course, that a civil res judicata claim can pose a double jeopardy bar to a criminal prosecution that seeks to reexamine that issue (at least as long as the same-parties requirement is satisfied). The equation does not, however, work the other way. Though other res judicata doctrines may apply, nothing about double jeopardy bars a civil suit unless the “life or limb” threshold is met. An acquittal in a criminal case is irrelevant to a civil trial that falls below the “life or limb” threshold, not because, as the Court has held, the standards of persuasion are different,20 but because there is nothing at risk in the second trial that requires double jeopardy protection. On this “life or limb” account, Gelston would have been wrong to apply a double jeopardy collateral estoppel bar because the second proceeding was a tort suit, which is not a “life or limb” action. (The actual basis for Gelston seems to be more civil res judicata than criminal collateral estoppel.)
As long as the “life or limb” case follows the civil case, however, collateral estoppel is potentially involved. The Court has never addressed this precise issue, but United States v. Oppenheimer21 can be stretched to fit. Oppenheimer contains the famous statement by Holmes that offers a rationale for applying collateral estoppel to a criminal case: “It cannot be that the safeguards of the person, so often and so rightly mentioned with solemn reverence, are less than those that protect from a liability in debt.”22
The Oppenheimer issue was whether a pretrial dismissal on statute of limitations grounds barred another trial. Under the Court’s twice-in-jeopardy doctrine, the second trial can proceed because, as we see in the next chapter, pretrial dismissals occur outside the jeopardy framework and thus cannot constitute a jeopardy outcome. As the case was thus outside the Fifth Amendment, the government argued that collateral estoppel “does not exist for criminal cases except in the modified form of the 5th Amendment.”23 Holmes seemed to concede the premise that the Double Jeopardy Clause would not bar a second trial on the facts of the case and then wrote, “But the 5th Amendment was not intended to do away with what in the civil law is a fundamental principle of justice in order, when a man once has been acquitted on the merits, to enable the government to prosecute him a second time.”24
To be sure, Oppenheimer involved two criminal cases, and it is therefore not authority for the proposition that a civil judgment in favor of a defendant could bar a criminal prosecution. Nonetheless, Holmes referred to collateral estoppel as a “fundamental principle of justice” in “the civil law” and then used it in a context when double jeopardy would not otherwise apply. This suggests that Holmes accepted its cross-over nature between civil and criminal. If a principle of civil justice would bar a criminal trial when jeopardy had not attached in the earlier proceeding because of a timing problem, it should bar a criminal trial when double jeopardy did not “attach” because the prior proceeding was civil in nature.
This reading of Oppenheimer is inconsistent with Hoag v. New Jersey, in which the Court later “entertain[ed] grave doubts whether collateral estoppel can be regarded as a constitutional requirement.”25 But Hoag did not survive Ashe v. Swenson, where the Court held that collateral estoppel was “embodied in the Fifth Amendment guarantee against double jeopardy.”26 In Ashe, the state charged six counts of robbery based on the robbing of six victims. This is a classic example of six act-tokens that manifest six separate Feinberg harms. Even on a lenient act-token view (see chapter 5), there is no same-offense bar to six trials and six convictions for robbery. But the state tried one count first, and the jury acquitted Ashe. The Court held that the state could not try him for any of the other robberies.
As Ashe overruled Hoag, Oppenheimer can be read expansively to permit offering a civil verdict as a collateral estoppel bar in a criminal case. The next section explores whether the different standards of persuasion in civil and criminal cases have any implication for collateral estoppel.
The ability of collateral estoppel to cross over between civil and criminal law raises an issue about different standards of persuasion that does not, of course, exist when two criminal prosecutions (or two civil trials) form the basis of the issue preclusion claim. If the civil case goes first, as we saw in the preceding section, the defendant should be able to offer a collateral estoppel claim against a subsequent criminal prosecution that seeks to reexamine the same blameworthiness (always assuming that the relevant same-parties requirement is met). The civil verdict has established that less than a preponderance of the evidence supported finding blameworthiness, which means that the proof was far below the criminal standard of beyond a reasonable doubt.
But the standard of persuasion suggests the opposite rule if the criminal case goes first. A defendant is entitled to an acquittal if the state’s proof exceeded a preponderance but was less than beyond a reasonable doubt. The acquittal thus does not necessarily establish that the civil plaintiff would have failed and the Court has held that the civil case can proceed.27
This issue rarely arises. Civil offenses do not authorize incarceration and thus are not “life or limb” offenses. In chapter 4 I acknowledged that a prosecutor can abuse a civil offense in a way that transforms it into a “life or limb” proceeding, as in Halper,28 but these are exceedingly rare occurrences. When the second proceeding is civil in form and mode, defendants cannot assert a collateral estoppel claim on the account offered in this book, because of the “life or limb” requirement, and on the Court’s doctrine, because of the differing standards of persuasion.
For the truly rare case in which the civil trial is a “life or limb” proceeding, the Court’s standard of persuasion bar should be ignored. The precision that justifies this bar is ephemeral. Yes, the acquittal in the criminal case could have been based on evidence greater than a preponderance but less than proof beyond a reasonable doubt. The acquittal was roughly as likely, however, to have been based on evidence less than preponderance.29 In the latter case, the defendant is denied the issue preclusion to which he would have been entitled had the Court known the precise nature of the initial acquittal.
The Court’s current view of this issue implicitly rejects Gelston’s holding that the acquittal in the in rem forfeiture is binding in the subsequent civil case. The Court has also overruled Coffey v. United States,30 which had explicitly rejected the standard of persuasion argument. Coffey was acquitted of the criminal charge of operating an illegal distillery. The Court acknowledged the persuasion argument but held that the civil forfeiture of the distillery apparatus could not follow because it was “substantially the same thing” as a criminal case.31 The Court held, in effect, that the fact of whether Coffey operated an illegal distillery had been determined for all time and for all purposes.
A blameworthiness account is fully consistent with Coffey on those rare occasions when the civil proceeding rises to the level of a “life or limb” proceeding. It is too bad that the Court abandoned Coffey, presumably to reduce the scope of double jeopardy protection, rather than take the more textually justified position that a civil forfeiture is not a “life or limb” offense. Now that the Court has essentially adopted the latter position,32 perhaps it is not too late to resurrect Coffey for those rare cases when the civil proceeding involves a “life or limb” offense.
My account of double jeopardy matches singular blameworthiness with single outcomes. Collateral estoppel depends on (assumes) the accuracy of the first proceeding, but determinations of blameworthiness are at best rough approximations. The Court made this clear in Harris v. Washington.33 The state court held that the failure of the trial court to permit the jury to hear admissible evidence in the first trial should allow the state a second chance; the theory here is that the first acquittal is somehow not accurate enough if the state’s case has not been fully presented through no fault of the state. The Court reversed, holding a second trial to be barred “irrespective of whether the jury considered all relevant evidence [in the first prosecution] and irrespective of the good faith of the State in bringing successive prosecutions.”34
Similarly, there is no “reverse collateral estoppel” that resurrects an earlier outcome that was unfavorable to the defendant. Consider the following scenario: a defendant is convicted, retried after appeal and acquitted, and tried a third time and convicted. The state argued that collateral estoppel should not operate because the first trial ended in a conviction. The Court rejected the state’s argument; the key point in time is the acquittal. Once acquitted, the defendant can rely on collateral estoppel regardless of what preceded the acquittal.35
These cases manifest a sort of “rough” justice: if the defendant gets a favorable ruling, the Court will apply it broadly, without regard to whether it is precisely accurate. This common-sense approach suggests that we should not permit formal differences in the standard of persuasion to permit reprosecution of an issue determined in the defendant’s favor, but that issue comes up so rarely on a “life or limb” understanding of jeopardy that its resolution hardly matters.
Another common-sense outcome is to reject collateral estoppel in single trials, as the Court has done. The jury in United States v. Powell,36 for example, found the defendant not guilty of the predicate drug crime but guilty of the compound offense that required proof of the predicate crime. Though this outcome is logically impossible, the Court refused to reverse the conviction on the ground of collateral estoppel. To conclude that collateral estoppel should arise from the acquittal “assumes that the acquittal on the predicate was proper—the one the jury ‘really meant.’”37
This issue has come out differently in England, though Friedland contends that “the better interpretation of these cases is that the conviction was quashed because it was unreasonable [based on insufficient evidence] and not because it was not possible as a matter of issue estoppel.”38 If the government introduces sufficient evidence, nothing in a blameworthiness account suggests reversing a conviction because it was joined with a logically inconsistent acquittal.39 Anne Bowen Poulin argues that in some cases the acquittal is not logically inconsistent with the conviction and that it should be held to bar later trials requiring proof of an issue foreclosed by the acquittal (though it would not result in reversal of the companion conviction).40 This accords with my blameworthiness account; the logically consistent acquittal should end the defendant’s blameworthiness for all issues foreclosed by the acquittal.
The same-parties requirement is also grounded in a common-sense view of the legal process. If a private citizen goes first, a realistic fear of fraud exists, the same concern that led Henry II to limit acquittals by the water ordeal.41 The civil litigants could strike their own bargain, leading to a verdict for the defendant that did not accurately reflect his blameworthiness. This was the argument the king made to reject the verdict in favor of the Duchess of Kingston in the ecclesiastical court.
Moreover, there is the quite separate concern about fairness to civil litigants. In the absence of a same-parties requirement, the government could go first and sometimes deprive the citizen of her chance to prevail in a proceeding with a less onerous standard of persuasion. This is the issue that led to the “year-and-a-day” rule. The common law required the king to wait a year and a day before indictment to give the private suit time to proceed.42 Finding this rule too restrictive, Henry VII promulgated a statute in 1487 that permitted both the king’s indictment and the private suit to proceed in murder cases.43
The separation of the sovereign and the tort system was a good compromise in 1487 and is still a good compromise today. The compromise creates a certain conceptual dissonance when the same issue is reexamined in a second proceeding, which is most clearly seen when the judgment of a speciality court is reexamined in a court of general jurisdiction (Kingston and, by analogy, Mason). Nonetheless, the 1487 compromise can be justified on the grounds of fairness to civil litigants, on the notion that the harm to the private citizen is somehow distinct from the harm manifested by criminal law, and on the fear of fraudulent use of the civil process to avoid criminal liability.44
A related question is whether the same governmental entity has to be bringing the criminal charge in both proceedings—whether there is a dual sovereignty exception to collateral estoppel. Friedland argues that the Canadian federal system should not recognize an exception for dual sovereignty in collateral estoppel issues,45 but the Canadian system is different from ours. In Canada, “there is only one system of criminal courts and generally only one body, the province, charged with administering the criminal law.”46
In the United States, different courts and prosecutors are involved. The concern about fraud arises again, although it is surely less likely that prosecutors and courts will participate in fraud to protect defendants from criminal liability in other jurisdictions. But the resolution of this issue is simple: it should track the resolution of the same-offense issue between sovereigns. If a legislature intended its criminal law to create blameworthiness distinct from that of the other sovereign, then it surely did not intend collateral estoppel to bar its prosecutors. Without reaching the legislative-intent issue that underlies my blameworthiness account, the Supreme Court has held that neither sovereign has to recognize collateral estoppel implications from an acquittal in the courts of the other sovereign.47
The account in chapter 6 presumes that states do not intend to reprosecute the same blameworthiness already prosecuted by the federal government, and vice versa. On that presumption, a collateral estoppel bar against the other sovereign would also arise in acquittals. By rebutting this presumption and making its intent clear to reprosecute the same blameworthiness, a legislature would simultaneously make extra-sovereign collateral inapplicable to its prosecutions. Failing explicit rebuttal, an acquittal should give rise both to a same-offense and collateral estoppel claim against other sovereigns.
My blameworthiness account of collateral estoppel can be summarized as follows: no government can proceed against a defendant in a “life or limb” proceeding if it depends on proof of an act-token found not to have occurred in an earlier proceeding in the courts of the same, or another, government. This is but a variant of the basic same-offense account: a defendant cannot be tried or punished twice for a single commission of the same “life or limb” blameworthy act-type. I have thus attempted to show a unity between collateral estoppel and double jeopardy that Amar and Marcus deny is possible.48