Concept
Three Tests for Punishment
Bills of Attainder
Three Tests for Punishment
Article I, section 9 forbids Congress to pass a bill of attainder, and section 10 forbids the states to do so, yet neither provision says what counts as the punishment that makes a legislative act one.1 The category reaches well beyond death and prison, as the historical list below shows, and that breadth creates a problem, because nearly every statute leaves someone worse off. Taxes, licensing rules, and eligibility conditions all impose burdens. The mere fact that a law imposes burdensome consequences does not make them forbidden punishment,2 and the severity of a sanction does not settle its character either.3 A court therefore needs a method for separating punishment from the ordinary burdens of lawmaking. In Selective Service System v. Minnesota Public Interest Research Group, the Court supplied one: in deciding whether a statute inflicts forbidden punishment, it has recognized three necessary inquiries.4
The first inquiry asks whether the challenged statute falls within the historical meaning of legislative punishment. The second asks whether the statute, viewed in terms of the type and severity of the burdens it imposes, reasonably can be said to further nonpunitive legislative purposes. The third asks whether the legislative record shows an intent to punish.5 The Court drew this framework from Nixon v. Administrator of General Services and, applying it to section 12(f), the provision that conditioned federal student aid on draft registration, concluded that all three criteria showed it was not a punitive bill of attainder.6
Each inquiry looks in a different place for evidence of punishment, which is why the three work as a set. The first looks to history and asks whether the sanction resembles what legislatures traditionally inflicted on people they wished to condemn. The second looks to the design of the law and asks whether a legitimate goal explains the burden. The third looks to the record and asks what the lawmakers said they were doing. A statute that matches no historical sanction can still be punitive if its burdens cannot reasonably be explained by a nonpunitive goal, so a court does not stop at the first inquiry.7
Historical Punishments
The constitutional text names no punishments, so the Court fills the gap with history.8 In Nixon v. Administrator of General Services it described the substantial English and American experience with attainders as “a ready checklist of deprivations and disabilities so disproportionately severe and so inappropriate to nonpunitive ends that they unquestionably have been held to fall within the proscription of Art. I, § 9.”9 A statute that imposes any of those sanctions on named or identifiable individuals, the Court added, “would be immediately constitutionally suspect.”10 This is the first of the three inquiries. It asks whether the burden belongs to the family of penalties that legislatures have inflicted on people they wished to condemn.
A checklist works because of how its entries got there. Each sanction was used by legislatures that wanted to condemn particular people, and each is so severe, and so poorly suited to any regulatory goal, that a legislature imposing it on identifiable persons can hardly claim to be managing public affairs. History also gives a court a fixed reference point. Asking whether a law imposes what Parliament once imposed on a rebel is easier than asking what the legislators privately wanted, and the answer does not depend on any admission of motive.
The English entries came in two groups. A bill of attainder originally meant a parliamentary act sentencing a named person, or identifiable members of a group, to death, as in the 1685 attainder of the Duke of Monmouth, then in open rebellion, who was declared convicted of high treason and made to suffer death and the forfeiture of all his property.11 Chief Justice Marshall's description in Fletcher v. Peck, that such a bill may affect life, may confiscate property, or may do both, captures the same pairing.12 Article I also proscribes the milder bills of pains and penalties, which punished without taking life and were generally aimed at persons considered disloyal to the Crown or state.13 Those acts commonly imposed imprisonment, banishment, and the punitive confiscation of property by the sovereign. The opinion's examples show each in operation: a 1701 act continued the imprisonment of Counter and others for a plot against the king, Georgia banished named persons it declared guilty of treason, and after the Revolution the states seized the property of suspected Tory sympathizers.14
American legislatures added a fifth entry. A legislative enactment barring designated individuals or groups from specified employments or vocations became a mode of punishment commonly used against those branded disloyal, and the Court's examples are the bar on clergy who would not take a loyalty oath in Cummings, the bar on named individuals from government employment in Lovett, and the bar on Communist Party members from union office in Brown.15 A list built only from English practice would miss the form punishment actually took in this country, which explains why the employment bar sits beside death, imprisonment, banishment, and confiscation. Because the list is fixed, the inquiry is best run as a matching exercise followed by a check on substance.
Lovett supplies a clean match at the second step. Congress cut off the pay of three named federal employees, which the Court itself describes as a bar on named individuals from government employment, the American entry on the list.17 Nixon supplies a clean miss. Congress there directed the General Services Administration to retain control of records that the former President claimed as his property, yet section 105 of the Act provided for an award by the district court of just compensation, and the Court held that this undercut even a colorable claim of punitive confiscation, because an owner paid compensation is put in the same monetary position as if nothing had been taken.18 The Court therefore found that no feature of the Act fell within the historical meaning of legislative punishment.19 The mechanism is that confiscation punishes by forfeiting value, and compensation returns the value, so what remains is a transfer for a price.
The draft-registration aid law shows a second kind of miss. The Court found that its sanction was the “mere denial of a noncontractual governmental benefit,” with no affirmative disability or restraint and nothing approaching the infamous punishment of imprisonment.20 A student refused federal aid has not been imprisoned, banished, stripped of property, or barred from an occupation, so the burden fits none of the five entries.
The labels on the list also depend on context. Banishment is on it, yet in Fong Yue Ting v. United States the Court sustained the deportation of Chinese laborers who lacked certificates of residence and rejected the attainder objection, describing the deportation order as an administrative order and not a punishment or sentence.21 The dissent answered that deportation is punishment, a removal from home, family, business, and property that parallels banishment, defined as forced exile inflicted principally upon political offenders.22 The disagreement shows that a word on the list does not decide the case. A court asks whether the burden is the historical sanction in the setting before it, and the majority there treated the removal of the laborers as an administrative matter and not as banishment inflicted as punishment.
A miss at this stage does not end the analysis. After finding that the Nixon Act imposed no traditionally prohibited punishment, the Court said its inquiry was not ended, because new burdens and deprivations might be legislatively fashioned that are inconsistent with the guarantee, and it went on to ask whether the law reasonably furthers nonpunitive purposes.23 Finding a burden and stopping skips the comparison, and finding no historical match and declaring the law valid stops one inquiry too soon. The functional test, taught next, picks up exactly where this one leaves off.
Functional Test
A law that matches no sanction on the historical list can still violate the bans in sections 9 and 10.25 In Nixon v. Administrator of General Services the Court found that the Act imposed no punishment traditionally judged to be prohibited, then said that this finding did not end the inquiry: its treatment of the clause has never precluded the possibility that legislatures might fashion new burdens and deprivations inconsistent with the guarantee.26 The Court therefore looks past historical experience and applies a functional test, which asks whether the law, “viewed in terms of the type and severity of burdens imposed, reasonably can be said to further nonpunitive legislative purposes.”27 Selective Service System v. Minnesota Public Interest Research Group later listed this as the second of three necessary inquiries, after the historical one and before the search of the legislative record for intent.28
The functional test works by inference, and the inference rests on a plain fact about legislatures: they seldom announce that they mean to punish anyone. A court therefore cannot settle the question by reading labels or waiting for an admission, and it asks instead what explains the burden. If a legitimate goal such as protecting evidence or preserving records accounts for the burden, the law is ordinary governing. If no legitimate goal accounts for it, the Court said, it is reasonable to conclude that punishment of the individuals disadvantaged by the enactment was the purpose of the decisionmakers.29 The conclusion follows because every burden has some motive behind it, and once the legitimate motives are exhausted, the one that remains is a wish to make the targets suffer for who they are or what they have done, which is the legislature acting as judge.
The test's focus on the type and severity of the burden supplies two measures of fit. Type asks whether the kind of burden connects to the goal, as custody of papers connects to preserving evidence while a bar from a job does not. Severity asks whether the burden stays within what the goal needs, because a drastic burden is left unexplained by a goal that a mild one would fully serve. Severity is a measure of fit here and does not itself decide the question, since a sanction's severity is not determinative of its character as punishment.30
Step three does most of the work. A court that stops at step two, satisfied that the legislature offered some respectable goal, has not applied the test, because the question is whether that goal explains this burden on these persons.
Nixon itself passes at step three. Congress had acted against the background of an agreement between the former President and the Administrator of General Services that expressly contemplated destroying some of his materials, and it identified two goals: preserving information needed to complete Watergate-related prosecutions, and safeguarding the public interest in access to materials of general historical significance.34 The Court reasoned that legislation guaranteeing the availability of evidence for criminal trials is a fair exercise of Congress's responsibility to the fair administration of criminal justice, and that Congress's authority to preserve records of historical value is well established, so neither goal supports an inference of punishment.35 The burden fit those goals in type. The government took custody of the materials so they could be kept and screened, and the Act directed the agency to protect the former President's opportunity to assert legal rights, to leave in his sole custody materials unlikely to relate to the stated objectives, and to hear his claims in court on a priority docket.36 Each of those features shows a burden shaped by its purpose and limited to it.
United States v. Brown comes out the other way even though the legislature there also had a respectable goal. Section 504 of the Landrum-Griffin Act made it a crime for a member of the Communist Party to serve as an officer or employee of a labor organization, and the Court held the provision a bill of attainder.37 The Court accepted that Congress may prevent political strikes by rationally related means, and it still held that a blanket decree against everyone on the party's membership rolls was unjustifiable because it determined guilt by legislative fiat.38 The mechanism explains the result. Preventing political strikes depends on what particular officers are likely to do, yet the statute treated membership as an equivalent expression for a list of undesirable characteristics, a suggestion the Court had long found fallacious, and it forfeited the job of every member whether or not that member would ever call a strike.39 The burden also outlasted membership, since former members stayed barred for five years after resigning.40 The government's answer that the ban looked forward instead of avenging past acts failed, because the Court called it archaic to limit punishment to retribution when punishment also serves rehabilitative, deterrent, and preventive purposes.41 The Court repeated the point in the draft-registration case, observing that punishment may involve deprivations inflicted to deter future misconduct and that a legislative intent to encourage compliance with the law does not establish that a statute is merely legitimate regulation.42 Chief Justice Warren's opinion in Brown described the clause as a safeguard against “trial by legislature.”43
Nixon and Brown share a feature: in each, the legislature could point to a legitimate-sounding goal. What separates them is the fit between burden and goal. In Nixon the burden was custody of documents, which is what preserving evidence and history requires, and it came with protections for the person burdened. In Brown the burden was loss of a livelihood, which the goal of stopping political strikes did not require for members with no shown inclination to strike, and it rested on a presumption drawn from membership alone. A stated nonpunitive purpose opens the inquiry, and the court must still test whether that purpose explains the burden actually imposed.
Less burdensome alternatives matter because they show whether the burden was needed. In Nixon the former President argued that Congress could have authorized a civil suit to enjoin disposition of the materials by an unreliable custodian, which would have left the question of his blameworthiness to a court.44 The Court agreed that Congress might have chosen that course, but it reasoned that Congress may have viewed a full judicial inquiry into his conduct and reliability as no less punitive and intrusive, since he had resisted judicial scrutiny of his records only months earlier, and that a rational and fairminded Congress might have preferred the carefully tailored law it enacted.45 The Court added that if the record unambiguously demonstrated legislative punishment, the difficulties of the judicial alternative would be of no constitutional significance.46 A gentler alternative therefore counts as evidence that a heavier burden was unnecessary, and it does not settle the question, because the legislature's choice is judged by whether its judgment was rational and fairminded.47
The usual exam error is to stop once the legislature has named a respectable purpose, since nearly every statute can be described as protecting someone or preventing something. The inquiry continues until the court has asked whether that purpose reasonably explains this burden on these persons, and the preventive character of a purpose does not by itself remove a deprivation from the category of punishment.48 A law that clears this inquiry may still be punitive if its legislative record shows an intent to punish, which is the third inquiry and the subject of the next section.
Legislative Intent
Neither section 9 nor section 10 says how a court should discover whether a legislature meant to punish the people it burdened.50 The answer for the third inquiry is a motivational one: the court asks whether the legislative record evinces a congressional intent to punish.51 Selective Service System v. Minnesota Public Interest Research Group lists the same question third among the necessary inquiries, after the historical and functional tests already covered.52 Where those two tests examine the sanction and its fit to a purpose, this one examines what the lawmakers said they were doing, drawing on committee reports, floor debates, and the features of the statute itself.53
The record matters because of the danger the clause was written to contain. Among the concerns behind the ban, the Court has identified the fear that a legislature, in seeking to “pander to an inflamed popular constituency,” will find it expedient to take up the role of judge or, worse, of lynch mob.54 A legislature acting in that spirit will usually explain itself by describing the targets as blameworthy, because that description is what the aroused constituency wants to hear. A legislature solving a public problem explains itself by describing the problem. The record therefore works as a diagnostic, since it shows whether the measure grew out of a verdict about particular people or out of a concern that would exist whoever those people were.
The record in United States v. Lovett shows a legislature reaching a verdict. Congress forbade payment of any salary to three named government employees, the Court held the statute an unconstitutional bill of attainder, and its finding of punishment rested in large measure on the specific congressional history.55 That history included a House Report that expressly characterized the individuals as subversive and unfit to continue in government employment.56 Congress thus announced a conclusion about the employees' character and attached the sanction to it, so anyone reading the record could watch the legislature perform the judge's task of deciding who was blameworthy.
The record in Nixon v. Administrator of General Services looks entirely different. The Senate and House committee reports, in formally explaining why they urged passage, expressed no interest in punishing the former President and justified the Act by two objectives within Congress's competence: preserving the availability of judicial evidence and of historically relevant materials.57 The Court found that both Houses acted predominantly out of a resolve to undo the agreement the former President had recently negotiated with the Administrator of General Services, which expressly contemplated destroying some Presidential materials, so the reports dwelt on the meaning and effect of that agreement and cast no aspersions on his personal conduct.58 The floor debates pointed the same way. Senator Ervin said that the bill “does not contain a word to the effect that Mr. Nixon is guilty of any violation of the law” and that it “does not inflict any punishment on him.”59 The two records differ in what the legislators said about the people burdened. The Lovett report named a character flaw in the employees and the sanction followed from it, while the Nixon reports named a problem with a particular agreement and the sanction followed from that.
A record like Lovett's is evidence, and the doctrine does not make it a requirement. The Court in Nixon expressly declined to suggest that a formal legislative announcement of moral blameworthiness or punishment is necessary to an unlawful bill of attainder.60 The reason follows from how legislatures behave. If the clause applied only when a legislature admitted its purpose, any legislature could avoid the ban by keeping its reports polite and its debates silent while imposing the same sanction. The absence of punitive sentiment still carries weight, because it is probative of nonpunitive intentions and largely undercuts the fear that the legislature is playing judge to an inflamed constituency.61 A clean record is therefore a point for the legislature on this inquiry that does not by itself end the case.
The statute's own terms are part of the record, and in Nixon they corroborated the committee reports. The Act placed the materials with the General Services Administration, the same agency the earlier agreement had named as depository, guaranteed the former President and anyone he designated access at all times, and provided for just compensation if his property had been confiscated.62 It also directed the agency to protect any party's opportunity to assert legal and constitutional rights, and it assured district court jurisdiction and judicial review of his claims, with priority on the docket.63 Each feature shows a burden shaped by the preservation goals the reports announced. A court reading the words of the members and the design of the statute together can see whether the two tell one story.
The materials a court may consult are limited. The former President urged the Court to account for the political turbulence of 1974, and the Court answered that it is not free to invalidate Acts of Congress on inferences drawn from a personalized reading of the contemporary scene or recent history.64 It may look only to the statute's terms, to the intent expressed by the members who voted for it, and to the existence or nonexistence of legitimate explanations for its apparent effect.65 Those three sources can be read, quoted, and contested by both sides. An argument from atmosphere has no such anchor, since nearly every law that burdens a controversial figure passes amid political heat, and a test built on mood would let a court attribute a punitive motive to almost any such law.