Concept
Punitive Damages Limits
Other rights and interests
Punitive Damages Limits
Substantive due process claims over property and economic liberty generally fail under the deferential tier described earlier in this chapter, and the notable exception is the award of punitive damages.1 The leading decision is BMW of North America, Inc. v. Gore, which held that in deciding whether a punitive award is unconstitutionally excessive, due process requires consideration of three guideposts: the reprehensibility of the defendant’s conduct, the ratio between the actual or potential harm and the punitive award, and the difference between the award and the civil or criminal penalties authorized in comparable cases.2 Those guideposts exist to give a defendant fair notice of its potential punitive exposure, and the subsections below take each of them in turn.
Punitive damages differ from ordinary damages in function. Compensatory damages replace a loss the plaintiff has suffered, while a punitive award is added on top, in exceptional cases, to punish a defendant whose behavior was more blameworthy than even gross negligence and to deter that behavior in the future.3 The deterrence logic runs as follows: a defendant who acted with something close to knowledge that harm was highly likely has plainly not been deterred by the prospect of compensatory liability, so the threat of a further award shifts the defendant’s cost-benefit calculation. That function explains why the award needs constitutional supervision. A punitive award is a civil punishment whose amount a jury sets in a suit that lacks the protections of a criminal prosecution, and nothing in the verdict form tells a defendant in advance how large a penalty its conduct could draw. A defendant who cannot predict the penalty cannot plan or comply, and an unbounded figure may simply express the jury’s anger at the defendant. The traditional punitive damages regime was perceived as overly subjective and unpredictable, and the Court has responded by circumscribing awards, relying primarily on the Due Process Clause.4
The Eighth Amendment’s Excessive Fines Clause might seem the natural source of a limit, but in Browning-Ferris Industries of Vermont, Inc. v. Kelco Disposal, Inc. the Court held that the clause does not apply to a jury award of punitive damages in a civil case between private parties.5 The Court reasoned that punishment and deterrence underlie both punitive awards and criminal law, but that the overlap is too small to impose the same constitutional limits. The limits therefore come from the Due Process Clause of the Fourteenth Amendment, which imposes substantive limits on the size of a punitive award and procedural limits on the process by which state courts award it.6 The Court has used the substantive limit to invalidate awards as grossly excessive in relation to the state’s legitimate interest in punishment and deterrence, first in Gore and again in State Farm Mutual Automobile Insurance Co. v. Campbell.7
The sections below begin with Fair Notice and State Interests, which states the governing principle through the facts of Gore and explains the territorial limit on what a state’s interest can justify. That section also takes up the position of some Justices that the Due Process Clause sets no substantive limit on the size of a punitive award, a view that has made the Court’s approach a subject of criticism.8 The three guideposts then follow one at a time, and the last section covers the procedural safeguards that surround them, beginning with the requirement of meaningful judicial review.9
Fair Notice and State Interests
The Due Process Clause of the Fourteenth Amendment prohibits a state from imposing a grossly excessive punishment on a tortfeasor, and BMW of North America, Inc. v. Gore applied that limit to an award of punitive damages.10 The wrongdoing was a national car distributor’s decision not to tell its dealers, and through them its customers, about predelivery damage to new cars when the repair cost less than 3 percent of the suggested retail price, and the question was whether a $2 million punitive award to one purchaser exceeded the constitutional limit.11 The answer turns on fair notice, which means that a person must receive notice of both the conduct that will subject the person to punishment and the severity of the penalty a state may impose.12 The three guideposts introduced above are the method a court uses to test whether a defendant had that notice of its potential punitive exposure.
In January 1990, Dr. Ira Gore, Jr. bought a black BMW sedan for $40,750.88 from an authorized dealer in Birmingham, Alabama.13 About nine months later an independent detailer, asked to make the car look better than usual, found evidence that it had been repainted, and Dr. Gore, convinced he had been cheated, sued the American distributor, alleging among other things that the failure to disclose the repainting was suppression of a material fact.14 At trial BMW acknowledged a nationwide policy it had adopted in 1983. If the cost of repairing damage done in manufacture or shipping exceeded 3 percent of the car’s suggested retail price, the car was placed in company service and later sold as used, but if the cost fell below that line the car was sold as new and the dealer was told nothing about the repair.15 Dr. Gore’s repainting cost $601.37, about 1.5 percent of the price, so BMW disclosed nothing.16
To prove actual damages of $4,000, Dr. Gore relied on a former BMW dealer’s estimate that a repainted BMW is worth about 10 percent less than an undamaged new one.17 To support punitive damages he added evidence that since 1983 BMW had sold 983 refinished cars as new, 14 of them in Alabama, without disclosing the repainting, and he argued that multiplying $4,000 by roughly 1,000 cars would produce a fitting penalty of $4 million.18 The jury returned $4,000 in compensatory damages and $4 million in punitive damages, based on a finding that the nondisclosure policy was “gross, oppressive or malicious” fraud.19 The Alabama Supreme Court then ruled for BMW on one point, finding that the jury had improperly computed the punitive award by multiplying the compensatory damages by the number of similar sales in other jurisdictions.20 Having found the verdict tainted, that court held that $2 million was a constitutionally reasonable punitive award and ordered a remittitur, which is an order cutting an excessive verdict down to the amount a court will allow to stand.21
Each half of the notice requirement protects something different. Notice of the conduct is rarely the weak point in a punitive damages case, and it was not here, because no one doubts that a state may prohibit deceptive trade practices and may require automobile distributors to disclose presale repairs that affect the value of a new car.22 Notice of the severity of the penalty carries more weight in this area, because a defendant that cannot foresee how large a sanction its conduct could draw has no way to weigh that cost in advance or to arrange its affairs to avoid it, and an unforeseeable figure may express nothing more than the jury’s reaction to the defendant. The resulting standard is a threshold of gross disproportion. Punitive damages may properly further a state’s legitimate interests in punishing unlawful conduct and deterring its repetition, states have considerable flexibility in setting the level, and only an award that is grossly excessive in relation to those interests enters the zone of arbitrariness that violates the Due Process Clause.23 An award that a reviewing court would have set lower is not unconstitutional for that reason alone. In Gore the Court concluded that the guideposts showed BMW had not received adequate notice of the magnitude of the sanction Alabama might impose, so the $2 million award was grossly excessive.24
Because the test measures the award against the state’s legitimate interests, the inquiry begins by asking what those interests are.25 In Gore the answer depended on how Alabama’s regulation compared with the rest of the country’s. Some states leave presale-repair disclosure to courts applying contract and tort principles, and others have enacted legislation defining the disclosure obligations of manufacturers, distributors, and dealers, so the result is a patchwork reflecting the policy judgments of lawmakers in 50 states.26 Reasonable people may disagree about full disclosure: one legislature may trust sellers’ interest in customer goodwill to produce voluntary disclosure, and another may exempt minor repairs to hold down the cost of regulation or to give sellers a safe harbor against lawsuits over minor repairs.27 That variety is the premise of the territorial limit that comes next.
Dr. Gore argued before the Court that a large award was needed to induce BMW to change the nationwide policy it adopted in 1983.28 The Court assumed that universal disclosure of presale repairs might be wise and observed that Congress has ample authority to adopt that policy for the whole Nation, but that no single state could do so or impose its own policy choice on neighboring states.29 From those principles of state sovereignty and comity it follows that a state may not impose economic sanctions on violators of its laws with the intent of changing the tortfeasor’s lawful conduct in other states, whether the sanction takes the form of a legislatively authorized fine or a judicially imposed punitive award.30 The sanction must instead rest on the state’s interest in protecting its own consumers and its own economy, so Alabama lacked the power “to punish BMW for conduct that was lawful where it occurred and that had no impact on Alabama or its residents.”31
Two separate reasons support the limit. The first comes from how the excessiveness inquiry is built: the award is compared with the state’s legitimate interests, those interests reach the state’s own consumers and economy, and so any portion of an award that responds to conduct beyond the state’s borders adds nothing to the interest side of the comparison. The second comes from federalism. If one state’s juries could set the penalty for a company’s nationwide policy, the most aggressive state’s view of disclosure would become the rule in states whose legislatures drew the line elsewhere, as the states that exempt minor repairs did.
Gore’s own record falls on both sides of that line. Dr. Gore’s request for $4 million was built by multiplying his $4,000 in actual damages by roughly 1,000 refinished cars sold since 1983, and the Alabama Supreme Court agreed that a computation reaching sales in other jurisdictions was improper.32 Alabama had no interest in punishing sales elsewhere to the extent the conduct was lawful where it occurred or never touched Alabama or its residents. Its legitimate interest lay in what BMW did within its borders, meaning the refinished cars sold in Alabama, and the state remained free to insist that BMW adhere to a particular disclosure policy there.33 Correcting the computation did not end the case, because the $2 million figure the Alabama court substituted was itself held grossly excessive.34
The Court reaffirmed the territorial limit in State Farm Mutual Automobile Insurance Co. v. Campbell, which arose after an insurer refused to settle a claim against its own policyholder within the policy limit and was later sued for bad faith and fraud.35 The jury’s $145 million punitive award, reinstated by the Utah Supreme Court on full compensatory damages of $1 million, was held excessive by a vote of 6 to 3.36 The Court repeated that a state cannot punish a defendant for conduct that may have been lawful where it occurred and that, as a general rule, a state has no legitimate concern in punishing a defendant for unlawful acts committed outside its jurisdiction.37 The Utah proceeding had been presented to the jury as a platform for holding the insurer accountable for its practices across the country, and the Court responded that a defendant “should be punished for the conduct that harmed the plaintiff, not for being an unsavory individual or business.”38 The limit governs what an award may punish. It leaves room for evidence of misconduct directed at third parties when a jury assesses the reprehensibility of what the defendant did to the plaintiff, but only if that misconduct is sufficiently similar to the misconduct that injured the plaintiff, a use taken up in the reprehensibility guidepost below.39
The usual error is to begin with the size of the award and ask whether it feels too large. Under this framework the size is judged against the state’s interests, so the first task is to define those interests and strip out any part of the award that responds to conduct the state cannot punish. A second error is to treat fair notice as one more factor to weigh beside the guideposts, when the guideposts are the method for testing it.
The Justices did not agree on the source of the limit. Justice Stevens’s opinion for the Court, joined by Justices O’Connor, Kennedy, Souter, and Breyer, rested its conclusion partly on procedural due process, specifically the requirement that a defendant receive fair notice of potential liability.40 Justice Breyer, joined by Justices O’Connor and Souter, wrote separately to explain that the limit can also be seen as substantive due process, because the concern arises from the basic unfairness of taking property through arbitrary coercion in place of law, and requiring law instead of a decisionmaker’s caprice both gives notice and helps assure uniform treatment of similarly situated persons.41 That reasoning explains why the concurrence rests on two features together, the lack of standards that constrain the award and its gross excess.42
Justice Scalia, joined by Justice Thomas, dissented because he did not regard the Due Process Clause as a “secret repository of substantive guarantees against ‘unfairness.’”43 On his account a state procedure that commits the amount to the jury’s discretion, subject to some judicial review for reasonableness, gives a defendant all the process that is due, because the procedural guarantee secures an opportunity to contest an award’s reasonableness in state court and no federal guarantee that the award actually be reasonable.44 He added that the Court’s new rule was constrained by no principle other than the Justices’ subjective assessment of the award’s reasonableness.45 Justice Ginsburg, joined by Chief Justice Rehnquist, also dissented.46 That position has not prevailed, since three Justices dissented again in Campbell, adhering to their views in Gore, and the majority applied the guideposts and held the award excessive.47 An answer should apply the majority’s framework and, if the problem invites it, note that some Justices would recognize no substantive limit.
Reprehensibility Guidepost
The first of the three guideposts in Gore is the degree of reprehensibility of the defendant’s conduct, and the Court called it the first and most important of the three.49 It comes first because of what the guideposts are for. Each one tests whether the defendant received fair notice of the severity of the penalty a state might impose, and what a defendant can foresee about the size of a sanction depends chiefly on how blameworthy its own conduct was.50 A company that knowingly endangers customers, repeats the practice, or covers it with deceit has reason to expect a heavy penalty, whereas a company that took a defensible position in an economic dispute has little reason to. Reprehensibility therefore sets the range of penalties the defendant could have anticipated, and the ratio and comparable-penalty guideposts then ask whether the jury’s figure stayed within that range.
In State Farm Mutual Automobile Insurance Co. v. Campbell the Court turned the guidepost into five questions about the defendant’s conduct, which a court asks in deciding how blameworthy that conduct was.51 The questions are these:
- Was the harm physical rather than economic?
- Did the conduct show indifference to, or a reckless disregard of, the health or safety of others?
- Was the target of the conduct financially vulnerable?
- Did the conduct involve repeated actions rather than an isolated incident?
- Did the harm result from intentional malice, trickery, or deceit rather than mere accident?
The five questions work as a sliding scale. A single factor pointing toward a large award may not be enough to sustain it, and the absence of every factor makes any award suspect.52 That follows from a presumption that compensatory damages have already made the plaintiff whole, so a further sanction needs the defendant’s culpability to exceed what the compensatory payment already answers.53 Each factor identifies a way conduct can be worse than an ordinary wrong. Physical harm and disregard of safety describe injuries that money repairs poorly and that threaten people beyond the plaintiff. Exploiting a vulnerable target shows the defendant took advantage of someone who could not protect itself. Repetition shows a practice rather than a lapse, and malice or deceit shows the defendant chose the harm, which is the kind of choice a penalty can discourage. The defendant’s wealth appears nowhere on the list, and wealth cannot justify an otherwise unconstitutional award, although a state may let the jury hear evidence of it.54
Gore sits at the weak end of the scale. The conduct under review was the failure to disclose that a car had been repainted before sale, and the Court concluded that it was “not particularly reprehensible.”55 The harm to the buyer was purely economic, because nothing suggested that the refinishing affected the car’s performance or safety, and the record held no evidence of an evil motive or of reckless or callous disregard for the rights of others.56 The nondisclosure also violated no federal statute or regulation, and nothing showed a pattern of egregious conduct.57 The jury’s finding of fraud under Alabama law did not settle the guidepost, since the question was whether the conduct carried the aggravating features the factors identify, and it carried none of them. Because reprehensibility was low, a defendant in BMW’s position had little reason to foresee a severe penalty, and that conclusion was the first of three signals the Court read as showing that BMW lacked fair notice of the sanction’s magnitude.58
Gore itself did not decide a case with the aggravating features, so the opposite pole comes from the Court’s own list of factors and from a decision that applies it. In Wyeth v. Rowatt a state court upheld a punitive award of about $57.8 million, cut from $99 million, against compensatory damages of $23 million, because the harm was physical, with the plaintiffs developing breast cancer after taking the defendant’s hormone replacement drugs, and because the defendant had misrepresented the drugs’ benefits and concealed data showing their risks.59 Physical injury and deceit are two of the features missing in Gore. When both are present, a large award becomes far easier to defend on this guidepost, though the ratio and comparable-penalty guideposts still have to be satisfied.
The factors measure the defendant’s conduct toward the plaintiff, and Campbell added a limit on the evidence that may be used to measure it. State Farm had refused to settle a claim against its policyholders, and the Campbells justified a large punitive award by offering evidence of a nationwide scheme to cap claim payments in order to meet corporate fiscal goals. The trial court allowed extensive expert testimony about State Farm’s practices in numerous states over some twenty years, although most of those practices bore no relation to third-party automobile claims like the one that injured the Campbells.60 The Court held that a defendant’s dissimilar acts, independent of the acts on which liability rests, may not serve as the basis for punitive damages, because a defendant should be punished for the conduct that harmed the plaintiff and not for being a disreputable business.61
Two reasons support the limit. A jury that punishes unrelated misconduct is in effect deciding other people’s hypothetical claims against the defendant under the label of reprehensibility, and because nonparties are not bound by whatever judgment one plaintiff wins, the same conduct could draw punitive awards again and again.62 The notice rationale points the same way, since a defendant can foresee the penalty for what it did to the plaintiff far better than the penalty for conduct that has nothing to do with the plaintiff’s injury. Other misconduct is not shut out altogether. Repeated misconduct is more reprehensible than a single lapse, but in a civil case the court must ensure that the conduct in question replicates the prior transgressions, and evidence of other acts need not be identical to be relevant.63 The Campbells showed scant evidence of repeated misconduct of the sort that injured them, so the conduct toward them was the only conduct relevant to reprehensibility.64 Misconduct directed at third parties therefore counts only if it is sufficiently similar to the misconduct directed at the injured plaintiff, and many lower courts have applied a “strict similarity” standard.65
The other mortgages inform the lender’s culpability toward this borrower, because they bear on whether the fee was a mistake. The same evidence carries a further limit. In Philip Morris USA v. Williams the Court recognized that a jury may use evidence of actual harm to nonparties to show that the defendant’s conduct posed a substantial risk to the general public and so was particularly reprehensible, but may not go further and use the punitive verdict to punish the defendant directly for harms it is said to have caused nonparties.66 A jury shown the hidden-fee pattern may treat it as proof of a practice, and the award must still answer to what the lender did to this borrower. The workplace-safety citation fails at the first step of that analysis, since it describes the lender’s general character without touching the conduct at issue.
Ratio Guidepost
The second guidepost in Gore compares the punitive award with the harm the plaintiff suffered or was likely to suffer.69 The jury there awarded $4,000 in compensatory damages and $4 million in punitive damages, which is 1,000 times the compensatory figure, and the Alabama Supreme Court’s reduction to $2 million still left a ratio of 500 to 1.70 The Court set that disparity beside the 4-to-1 ratio in Pacific Mutual Life Insurance Co. v. Haslip and the ratio of roughly 10 to 1, on one calculation, in TXO Production Corp. v. Alliance Resources Corp., and it concluded that although it had declined to impose a bright-line ratio, the disparity was so great as to violate due process.71
A ratio earns a place among the guideposts because it lets a defendant test a penalty in advance against a figure it can foresee. Reprehensibility says how blameworthy the conduct was but produces no number, whereas the compensatory award measures the injury the conduct caused, so a defendant can estimate a proportionate punitive exposure by applying a modest multiple to that injury. The multiples the law has used are modest ones, since the Court has pointed to a legislative history reaching back more than 700 years of sanctions at double, treble, or quadruple damages to deter and punish.72 An award hundreds of times the injury gives a defendant nothing to anticipate and leaves room for a jury’s anger to set the figure, which is the arbitrariness the fair-notice requirement exists to prevent.
State Farm Mutual Automobile Insurance Co. v. Campbell turned that comparison into a working benchmark.73 The Court again declined to impose a bright-line ratio, but it said that in practice few awards exceeding a single-digit ratio of punitive to compensatory damages, to a significant degree, will satisfy due process.74 Single-digit multipliers, it explained, are more likely to comport with due process than ratios in the range of 500 to 1 or, in that case, 145 to 1, and it recalled that Haslip had treated an award of more than four times compensatory damages as possibly close to the line of constitutional impropriety.75 The Court then announced a presumption against the 145-to-1 award, and its reasons show what makes a large ratio hard to defend. The compensatory award was $1 million, which the Court regarded as complete compensation for about a year and a half of emotional distress. The harm arose from an economic transaction and involved no physical injury, and the compensatory sum likely already contained a punitive element, because much of the distress came from the outrage and humiliation of the insurer’s conduct, which is the very thing a punitive award exists to condemn.76 Once a plaintiff has been fully compensated, and partly compensated for the outrage itself, a large multiplier punishes the same wrong twice, while a smaller multiple still serves the State’s goals of deterrence and retribution.
The benchmark bends in both directions because the ratio is a proxy for proportionality, and the compensatory figure sometimes misdescribes how much punishment a wrong deserves. Upward, the Court recognized that ratios greater than those previously upheld may comport with due process where “a particularly egregious act has resulted in only a small amount of economic damages.”77 The reason is incentive: when the compensatory recovery is tiny, the plaintiff may have little reason to sue and the wrongdoer little reason to change, so a larger multiple may be needed for the penalty to deter.78 In Mendez v. County of San Bernardino, a civil rights suit over an illegal arrest and search, the jury awarded $2 in nominal damages, and the Ninth Circuit held that $5,000 in punitive damages was appropriate, a ratio of 2,500 to 1 that was still far below the $250,000 the jury had first awarded.79 A used-car buyer fared similarly in Adeli v. Silverstar Automotive, where the seller failed to disclose fuel-system repairs recommended in a pre-purchase inspection, a defect that could have let harmful gas into the passenger cabin or started a fire. The jury awarded $20,000 in compensatory damages and $5.8 million in punitive damages, the trial court cut the punitive award to $500,000, and the Eighth Circuit upheld the resulting 25-to-1 ratio, rejecting a single-digit limit as constitutionally mandated and reasoning that a precise formula would turn fraud into a calculable business decision.80 Gore and Adeli both involved an undisclosed problem with a car, and what separates them is a safety risk and a far smaller ratio: the repainting in Gore had no effect on performance or safety and produced 500 to 1, while the fuel-system defect in Adeli threatened gas in the cabin or a fire and produced 25 to 1.81
The adjustment runs downward as well. When compensatory damages are substantial, a lesser ratio, perhaps only equal to the compensatory damages, can reach the outermost limit of the due process guarantee.82 Applied to the $1 million award, the Court thought the guideposts likely would justify punitive damages at or near the compensatory amount, yet it left the calculation to the Utah courts.83 On remand the Utah Supreme Court read that suggestion as “words of prediction, not direction” and approved a 9-to-1 award of about $9 million, and the Supreme Court declined further review.84 The California Supreme Court reads Campbell to set a presumption that holds absent special justification, such as extreme reprehensibility or compensatory damages that are unusually small, hard to detect, or hard to measure, and it has treated 1 to 1 as the constitutional maximum where reprehensibility was relatively low and noneconomic damages were substantial.85 The denominator in all of these calculations is the compensatory award to the particular plaintiff, and Campbell made no attempt to count effects on other insureds.86
Working the arithmetic on real facts shows the benchmark operating as a ceiling. A federal court of appeals reviewed a $5 million punitive award against ARCO for racial discrimination that cost a corporate contractor, Flying B, its contract, and the court took $50,000 as the harm because the contract loss and the discrimination were intertwined.87 The case was not one of the rare small-economic-damages cases, since $50,000 is substantial, so the court read the controlling authority to imply a ceiling of at most $450,000, nine times the compensatory damages and far below the 100-to-1 award the jury had returned.88 Flying B’s argument that ARCO’s corporate wealth justified more failed, because a defendant’s wealth cannot justify an otherwise unconstitutional award.89 The court also gave no weight to potential harm, since that idea concerns what would have happened had a wrongful plan succeeded, and the wrongful conduct there had succeeded.90
The single digit is therefore a starting point and offers no safe harbor to an award merely because it falls under ten. An award more than nine times compensatory damages is, in effect, presumptively unconstitutional, which is a fair shorthand for the starting point so long as the exceptions stay in view.91 Courts nevertheless sustain much larger ratios on unusual facts, and there is no one-size-fits-all formula.92
A separate 1-to-1 limit applies in maritime cases. Exxon Shipping Co. v. Baker capped punitive damages at the $507.5 million compensatory award, but the Court rested that limit on federal maritime common law and not on due process, so an answer should treat it as its own regime and not as the due process benchmark.93
The arithmetic in the second step is the easy part. The third and fourth steps decide whether the benchmark applies at full strength, and an answer that stops at single digit, so constitutional, or double digit, so unconstitutional, has skipped them.
Comparable Penalties Guidepost
The third guidepost in Gore compares the punitive award with the civil penalties authorized or imposed in comparable cases.95 Alabama’s Deceptive Trade Practices Act set a maximum civil penalty of $2,000, and comparable statutes in other states imposed similarly modest penalties, so the $2 million award against BMW was 1,000 times the most the legislature had authorized for the nondisclosure.96 The Court read that gap as further support for its conclusion that BMW lacked adequate notice of the magnitude of the penalty it might face.97 This is the quickest of the three guideposts to apply, because once the right statute is found the work is a comparison of two numbers.
The comparison belongs among the guideposts because a legislature announces how seriously the community views a wrong through the penalty it attaches to that wrong, and a defendant can look that penalty up before it acts. A seller who reads Alabama’s statute learns that this kind of nondisclosure can cost at most $2,000, and it has no comparable way to learn what a jury will add. When the jury’s figure dwarfs the legislated figure, the jury’s reaction to the defendant has set the sanction, and the legislature’s public judgment about seriousness has dropped out of the calculation. For that reason a court reviewing an award gives substantial deference to legislative judgments about the appropriate sanctions for the conduct at issue, and the penalty the legislature chose becomes the reference point.98
State Farm Mutual Automobile Insurance Co. v. Campbell applied the same comparison, and the Court said it need not dwell long on the point.99 The most relevant Utah civil sanction for the insurer’s fraud was a $10,000 fine, and the $145 million award dwarfed it, a gap of 14,500 to 1.100 The Utah Supreme Court had pointed to the possible loss of State Farm’s business license, disgorgement of profits, and imprisonment, but those references drew on the broad fraudulent scheme proved through out-of-state and dissimilar conduct, and the Court held that the analysis could not justify the award.101 The two cases differ in how the comparison was built. Gore set the award beside the penalty in the statute that defined the wrong, while the Utah court tried to enlarge the comparison by listing every sanction the insurer might have faced for a whole course of conduct, and that move failed for the same reason that evidence of dissimilar acts failed under the reprehensibility guidepost. A fine written for an ordinary violation is also usually far smaller than an award sized to deter a defendant for whom the wrong pays, so the comparison tends to work against awards built around deterrence, which is one reason it is weighed together with the other two guideposts and does not decide a case alone.102
Criminal penalties raise a separate question. The existence of a criminal penalty bears on how seriously a state views the wrongful act, and courts may look to criminal fines for similar misconduct as a guide when they review an award for excessiveness.103 Used to fix the dollar amount of a civil award, however, a criminal penalty has less utility, because that sanction can be imposed only after the heightened protections of a criminal trial, including a higher standard of proof, and great care must be taken to avoid using the civil process to assess it.104 The mechanism is straightforward. A plaintiff who could point to a statute authorizing years of imprisonment and ask a civil jury for a matching sum would obtain a criminal-scale punishment from a proceeding that lacks those safeguards. Punitive damages are no substitute for the criminal process, and the remote possibility of a criminal sanction does not by itself sustain an award.105
Penalty comparisons can also support an award. The Court has pointed to a legislative history reaching back more than 700 years of sanctions at double, treble, or quadruple damages to deter and punish, and it described those multiples as instructive although not binding.106 Their value for notice is that a defendant can anticipate an award in a multiple the legislature has long used, so an award at double or treble the compensatory damages can point to that practice in its defense. Current statutes repeat the pattern: the Uniform Trade Secrets Act and the Defend Trade Secrets Act allow punitive damages for willful and malicious misappropriation in an amount up to twice the compensatory damages.107 An award inside that cap must still satisfy due process, because a court may reduce an award that is grossly excessive even when it falls within a statutory limit.108
Courts outside the Gore and Campbell setting use the same method, and the results run in both directions. In a Fair Housing Act case, a federal appeals court found a $250,000 punitive award against a landlord who repeatedly harassed a tenant excessive, because it was more than eighteen times the $13,685 compensatory award. It considered the statute’s civil penalty of up to $55,000 for a first violation under the third guidepost and approved $54,750, four times the compensatory damages, which was higher than the 1.5 multiple the trial court had allowed and fell just under the statutory figure.109 The ARCO discrimination decision from the ratio discussion looked to a statute that did not even govern the claim. After the single-digit ratio produced a ceiling of $450,000 on $50,000 of harm, the court treated the $300,000 limit on punitive damages in Title VII cases as an appropriate benchmark for race-discrimination claims under § 1981, reading that limit as Congress’s view of the highest appropriate amount in somewhat comparable cases.110
The second step usually decides the guidepost, since the statute that defines the wrong supplies the most relevant figure. The usual errors are to treat the existence of a criminal penalty as a license for an award sized to it, and to compare the award with sanctions the defendant faced for conduct other than the wrong done to this plaintiff.
Procedural Safeguards and Review
The three guideposts from Gore (the reprehensibility of the conduct, the ratio of harm to award, and the comparison with civil penalties) exist to test whether a defendant had fair notice of its potential punitive exposure. Each guidepost asks a question of degree, and a question of degree protects no one unless some decisionmaker with authority applies it to the actual figure and can change the result. The Due Process Clause therefore limits punitive awards on two fronts. States keep discretion over punitive damages, yet the Constitution sets procedural as well as substantive limits on how an award is reached and how large it may be.113 Three procedural safeguards do the practical work: meaningful judicial review of the award, de novo appellate review of whether it is excessive, and protection against a jury that punishes the defendant for harm to people who are not parties to the case.
The need for review comes from how punitive awards are made. Compensatory damages replace a measurable loss, while a punitive award rests on deterrence and retribution, and the jury usually receives wide discretion in choosing the amount.114 The Court has warned that punitive damages pose an “acute danger of arbitrary deprivation of property,” because loosely worded instructions leave the amount to the jury and evidence of a defendant’s net worth can invite verdicts that express bias against big businesses.115 Civil defendants also lack the protections of a criminal proceeding even though the award serves purposes close to those of a criminal penalty, which raises the cost of an imprecise process.116 In Honda Motor Co. v. Oberg the Court held that due process requires meaningful judicial review of punitive awards, so that an excessive award does not become an arbitrary deprivation of property.117 Review happens twice, first when the trial judge rules on post-verdict challenges and again on appeal, and each stage checks the jury’s discretion.118
Review after the verdict has limits, which is why guidance before the verdict matters. A federal court of appeals has emphasized how narrow the grounds are on which a judge may set aside a jury’s verdict, so punitive awards are best controlled mainly through careful instructions given to the jury in the first place.119 Due process requires that the jury receive some guidance about the purpose of punitive damages as retribution and deterrence, perhaps with a statement that an award is not required.120 Vague instructions, or those that merely tell jurors to avoid “passion or prejudice,” do little to help them separate relevant evidence from evidence that is only inflammatory.121
Appellate review adds a second layer, and the Court has specified how intense it must be. In Cooper Industries, Inc. v. Leatherman Tool Group, Inc. the Court held that an appellate court should review de novo, meaning without deference to the lower court’s judgment, a district court’s evaluation of whether a punitive award is unconstitutionally excessive.122 The requirement also reaches state courts when they examine the federal constitutionality of an award.123 The reason lies in how the guideposts work. Applying them takes a judgment about how reprehensible the conduct was and how large a ratio is tolerable, and if that judgment were reviewed only for abuse of discretion, the constitutional ceiling would drift with the temperament of whichever judge presided. De novo review gives the guideposts one meaning across courtrooms, which is what the Court had in mind in describing exacting appellate review as ensuring that an award rests on an “application of law, rather than a decisionmaker’s caprice.”124 Review also limits what may be used to defend an award. A state may let the jury hear evidence of the defendant’s wealth, but a reviewing court cannot rely on wealth to rescue an award that the other guideposts show to be unconstitutional.125
The third safeguard comes from Philip Morris USA v. Williams. Mayola Williams sued Philip Morris, the maker of Marlboro cigarettes, over the death of her husband Jesse, a lifelong Marlboro smoker, and an Oregon jury awarded $821,000 in compensatory damages (cut to $521,000 under a state cap) and $79.5 million in punitive damages.126 The defendant had asked the trial court to instruct the jury that it could not punish for the impact of its conduct on third parties, and the trial court refused.127 By a vote of 5 to 4 the Supreme Court vacated the award, holding that the Due Process Clause forbids a state to use a punitive damages award to punish a defendant for injury it inflicts upon nonparties, who are strangers to the litigation.128 It remanded for the Oregon courts to fashion safeguards in any new trial, without deciding whether the award was grossly excessive.129
Two reasons support the bar, and each exposes a different defect in a verdict that counts the injuries of strangers. The first is fairness to the defendant, who, when threatened with punishment for injuring a nonparty, has no opportunity to defend against that charge, for example by showing that the other smoker knew smoking was dangerous or never relied on the defendant’s statements.130 The nonparty is absent and the defenses are particular to each injured person, so a verdict covering strangers resolves claims that no one has litigated. The second reason is arbitrariness. Punishing for injury to nonparties would add a “near standardless dimension” to the award, because the trial will not answer how many other victims exist, how badly they were hurt, or how the injuries occurred, and the jury is left to speculate.131 Aggregation adds a further worry. Strangers keep their own claims, so a defendant punished in one verdict for their injuries may be punished again when they sue, and the aggregate of punishments is a recognized constitutional concern in measuring an award.132
The rule does not wall off evidence of harm to others. A jury may consider actual harm to nonparties in deciding whether the conduct that injured the plaintiff also posed a substantial risk to the general public and so was particularly reprehensible, but it may not go further and use the verdict to punish the defendant directly for those harms.133 The same facts are thus proper for one question and forbidden for another, and the Constitution responds with procedure. Where a significant risk exists that a jury will punish directly for harm to nonparties, the state must adopt procedures to protect against that risk, and the risk can arise from the kinds of evidence admitted or from the arguments plaintiff’s counsel makes to the jury.134 A limiting instruction is the usual form such a procedure takes, and the refusal of one in Philip Morris is what left the award without the assurance due process requires.
A model instruction used in the Ninth Circuit shows what a protective procedure looks like. It tells jurors to consider the degree of reprehensibility, “including whether the conduct that harmed the plaintiff was particularly reprehensible because it also caused actual harm or posed a substantial risk of harm to people who are not parties to this case,” and then adds that they “may not, however, set the amount of any punitive damages in order to punish the defendant for harm to anyone other than the plaintiff in this case.”135 The first clause tells the jury how it may use the evidence and the second tells it how it may not, so the instruction separates the two uses the Court distinguished. Trial courts also police counsel: in Ray v. Allergan, Inc. a federal district court did not permit the plaintiff’s lawyer to ask the jury, in closing, to consider all the other people exposed to the same product.136 Even a well-drafted instruction asks a great deal of jurors, because they must weigh harm to strangers in judging reprehensibility and then leave it out of the amount, and the line is likely to confuse them.137 For that reason the wording of the instruction and the control of closing argument are where the protection actually lives.
The second step carries the most weight on an exam, because the nonparty rule is the one fact patterns tend to bury inside a trial narrative about other victims. Two slips follow from misreading it. One is to say that evidence of harm to nonparties is inadmissible, when it remains relevant to reprehensibility. The other is to treat post-verdict review as a cure for a missing instruction, when review is narrow and the instruction is the main control on the jury.