25
The principal–agent problem
An institution may have members or employees. Members can also be employees, as in workers’ cooperatives. Members interact horizontally, through the processes of arguing, bargaining, and voting that I discussed in the previous chapter. The vertical relations between employees and their superiors have a different character. To simplify, assume that the organization has a single executive (“the principal”) and many employees (“the agents”). In a cooperative, the employees collectively are the principal and individually act as agents. A principal-agent problem arises when, as is often the case, the principal and the agents have different interests. Workers may have an interest in a moderate work pace, whereas the manager may wish them to make a stronger effort.
Similarly, the head of a state agency has an interest in employees’ being honest, in the sense of not taking or demanding bribes from the public. She also has an interest in efficiency, so that the size of the public sector can be kept to a minimum. Employees can have opposite interests in both respects. If they are motivated only by their economic self-interest, they will take bribes when they can get away with it. As a consequence of their interest in power, agents also have an incentive to swell the size of their departments and to multiply the number of subordinates. Again, monitoring may be difficult. The principal may sometimes catch an agent taking a bribe, but in general this is not a method she can rely on. She may try to reduce opportunities for corruption, for instance, by having competitive bids for public contracts, but this precaution does not help if agents tailor the contracts so as to favor particular suppliers. Because agents often have a near-monopoly on information, principals may be unable to tell which requests for more hirings are justified and which are not. During the Cold War, wildly exaggerated estimates about the economic and military power of the Soviet Union produced by the American defense establishment justified a massive build-up of weaponry.
Subordinates are not the only ones that may have incentives out of line with the organization. American university presidents have had to step down when it turned out that they were awarding themselves huge salaries or had their homes redecorated at the organization's expense. One American vice president (Spiro Agnew) had to resign when he was exposed for corruption. Kleptocracy – the rule of thieves – has become an exceedingly familiar phenomenon across the world. Although they are principals in one sense, such leaders may also be subject to monitoring. Yet overseers (voters, boards of trustees, shareholders, the World Bank, or the International Monetary Fund) have often been conspicuously unsuccessful in regulating the behavior of chief executive officers or heads of state. As in other cases, they may lack either the information needed to correct excesses or the incentive to do so.
In workers’ cooperatives, a conflict may arise between workers as principals and workers as agents. Speaking to the Social Science Congress in 1863 Sir James Kay Shuttleworth said, with respect to the cooperative Lancashire cotton mills:
[Then] arose the formidable question – What benefits should the shareholders have in this mill beyond the ordinary profits? The first claim that was practically put forward in such societies was, that a preference should be given to the families of the shareholders in selecting the workers in the mill … He had witnessed on his own property the failure of one of these concerns. There was a desire to introduce into the concern the principle of cooperation to this extent, that the shareholders should have the advantage of the employment of their families in the mills. The immediate effect of that was this, that instead of producing a stricter discipline and that close attention to the working of machinery, which was so necessary in cotton mills (and he might mention that the discipline of a regiment was inferior in strictness to that of a cotton mill), at their quarterly or half-yearly meetings most vexatious complaints were made by the workers against the overlookers, and an overlooker who had dared to discharge a worker who was a shareholder, was in extreme danger of being dismissed at the next meeting.
Another frequently occurring problem in cooperatives arises from their reluctance to lay off their members in times of low demand. Commenting on the downfall of the Wolverhampton Plate-Locksmiths in 1878, a contemporary wrote that
if the business had been carried on by a private manufacturer, he would probably have discharged the workmen for whom, from the falling off of the demand for plate-locks, he could not find profitable employment, and applied himself to develop the trade that remained. But this would have involved on the body of workers who formed the society an amount of self-sacrifice for which they were not prepared. Instead, they worked for stock, in the hope that the demand would revive. As it did not revive before their resources were exhausted, they inevitably came to grief. Debts multiplied upon them; the best workers fell away.
There are three main ways in which a principal can respond to opportunistic or otherwise undesirable behavior by the agents: by limiting their opportunities to misbehave, by monitoring their behavior, and by providing incentives that will align their behavior with his.
The first solution is hard to implement. To be effective, an agent needs some independence and freedom of action. Slaves have rarely been used in occupations that demand application and care. It may prove impossible to structure the situation so that the agent is free to pursue the aims of the principal while literally unable to pursue his own. One may try to approximate this goal by having decision makers serve for such a short period that it is hard to bribe them. The jury system and the American electoral college (in its original conception) have been justified on these grounds (among others). Short tenure of elected officials (often combined with non-reeligibility) and frequent rotation of appointed officials, practiced by both the Roman and Chinese empires, are also supposed to reduce the opportunity for corruption. In two compelling metaphors, John Lilburne told the private soldiers in the New Model Army, “Suffer not one sort of men too long to remaine adjutators [delegates], lest they be corrupted by bribes of offices, or places of preferment, for standing water though never so pure at first, in time putrifies,” and the Jacobite John Byron described co-opted bodies “as stagnant pools, which needed regular elections to clean them.” Short tenure has severe efficiency costs, however. By the time officials have become familiar enough with the job to do it well, they may have to leave it.1 In many modern governments, the actual and expected turnover of ministers is so high that their time horizon is truncated over and above the normal shortening effect produced by the electoral cycle.
Historically, a common method has been to monitor the actions of the agents, not only to prevent shirking, but also to deter theft. Traditionally, that was the task of the foreman, but how can the manager be sure that the foreman is not demanding bribes from the workers or using his authority to promote his own financial or sexual ends? The nineteenth-century workplace was often characterized as “the tyranny of the foreman.” In department stores, floorwalkers exercised a similar power. In such cases, who shall guard the guardians? Also, the principal's monitoring of the agent is implicitly based on suspicions that may turn out be self-fulfilling. Monitoring can easily cause resentment, bad morale, and lower productivity. Finally, monitoring is costly. An article from 1989 reported that 7 percent of workers in the US non-agricultural private sector were employed as supervisors or inspectors. (I was unable to retrieve more recent data.)
Relying on incentives may seem more promising, since they do not require the principal to monitor actions, only outcomes. For this reason, incentive systems – rewarding individuals or groups according to their contribution to the goal of the principal – have often been seen as a panacea. The most direct application is observed in firms and institutions that use piece rates. A soccer coach may reward his players according to how many goals they score, or an administration may reward courts according to how many executions they carry out.2 Norwegian universities at one point gave a bonus to teachers or to their departments for each successful Ph.D. candidate, sometimes with a higher bonus for a female student. Individual teachers may get a salary raise, and departments more positions, as a function of the number of articles published in peer-reviewed journals. The budget of universities may be determined in part by their place on the Shanghai rankings, that of hospitals by the success rate of surgery or the turnover time for patients, and that of police forces by the number of reported crimes or by the percentage of crimes that are solved. Some organizations use tournaments (winner-takes-all competitions) among their members, notably to determine who shall be promoted. Examples could be multiplied indefinitely.
Institutions can also use negative incentives to induce compliance with the principal's goals. The boot camp is a paradigmatic example. In the classroom, positive as well as negative incentives are used. In France, students are (or were) moved around the classroom every week according to their performance, good students being moved to the front and less good to the back, thus combining the carrot and the stick. In a field experiment, students who earned their grades (“earners”) received positive incentives and those who maintained their grades (“maintainers”) received negative incentives. The earners started the semester with 0 points and added points with each graded assignment, whereas the maintainers were given the maximum number of points available for the course at the beginning of the semester and then subtracted points from this overall total as they lost points on a graded assignment. Consistent with the theory of loss aversion (Chapter 14), the maintainers performed (slightly) better.
An incentive system can succeed in aligning the interests of the principal and the agent or agents, as in the following stylized example. Suppose that the principal has an asset whose value may be high or low depending on whether the effort of the agent is high or low. The principal is a mayor who runs for reelection on the promise of a successful public safety campaign. The likelihood of reelection is high if the agent, the chief of police, supplies a high effort rather than a low effort. Since effort is costly to the agent, it has to be elicited by incentives. Assume that the high and low values of the asset are, respectively, $30,000 and $10,000. The costs to the agent of a high and low effort are, respectively, $8,000 and $4,000 (he has to make some effort to avoid being fired3). The probability of the high-value outcome being realized with high effort is 80 percent; with a low effort it is 30 percent. Suppose the principal proposes a bonus B in case she is reelected. If the agent makes a low effort and the mayor is reelected, his benefit is 0.3B – 4,000. (Since the principal can observe only the outcome, not the action, the agent might get the bonus even if he is slacking.) If he makes a high effort, the benefit is 0.8B – 8,000. A simple calculation shows that the agent's benefit for hard effort exceeds that of low effort when the bonus is greater than $8,000. If the principal pays the bonus, her net benefit is $18,000 (0.8 × $30,000 + 02 × $10,000 – $8,000). If she does not, it is $16, 000 (0.3 × $ 30,000 + 0.7 × $10,000). Hence she has an incentive to pay the bonus, which will give the agent an incentive to work hard. Both parties gain.
In this case, there was only one principal and one agent. In many important cases, a principal has to choose one of several agents. If each of the latter has private information about the facts on which the principal will base his decision, her task is to create incentives for the agents to reveal the information, not verbally, but through the choices they make. An early, if imperfect example is provided by the judgment of Solomon:
Then came there two women, that were harlots, unto the king, and stood before him. And the one woman said, O my lord, I and this woman dwell in one house; and I was delivered of a child with her in the house. And it came to pass the third day after that I was delivered, that this woman was delivered also: and we were together; there was no stranger with us in the house, save we two in the house. And this woman's child died in the night; because she overlaid it. And she arose at midnight, and took my son from beside me, while thine handmaid slept, and laid it in her bosom, and laid her dead child in my bosom. And when I rose in the morning to give my child suck, behold, it was dead: but when I had considered it in the morning, behold, it was not my son, which I did bear. And the other woman said, Nay; but the living is my son, and the dead is thy son. And this said, No; but the dead is thy son, and the living is my son. Thus they spake before the king. Then said the king, The one saith, This is my son that liveth, and thy son is the dead: and the other saith, Nay; but thy son is the dead, and my son is the living. And the king said, Bring me a sword. And they brought a sword before the king. And the king said, Divide the living child in two, and give half to the one, and half to the other. Then spake the woman whose the living child was unto the king, for her bowels yearned upon her son, and she said, O my lord, give her the living child, and in no wise slay it. But the other said, Let it be neither mine nor thine, but divide it. Then the king answered and said, Give her the living child, and in no wise slay it: she is the mother thereof.
(King James Bible, 1 Kings 3)
The example is imperfect, in a technical sense, because the false claimant has an incentive to mimic the true one, leaving Solomon with no useful information.4 The modern theory of incentives tries to overcome this problem, by creating mechanisms that will separate true from false claimants, or “good types” from “bad types.” As shown in one of the first analyses of this issue, employers may use years of education as a signal of productivity – not because education makes people more productive, but because the cost of undertaking an education is lower for more productive individuals. Anecdotally, the British civil service was run on this principle: if a person can acquire the capability in three years to compose Greek and Latin verse, he (very rarely she) was also qualified to run part of the empire. Applicants sort themselves out into “good types” and “bad types” by their willingness to undertake education. Similarly, low-risk drivers can signal their type by choosing insurance policies with higher deductibles but lower premiums, since their risk of having to pay the deductible is less than that of high-risk drivers.
Sometimes, as I shall now discuss, incentive systems fail. There are three main sources of failure: the incentives may induce rational behavior contrary to the intended one; they may change the preferences of the agents, undermining the standard (if usually tacit) assumption that preferences will be unaffected by the scheme; and they may induce irrational behavior that undermines the equally standard assumption that people respond rationally to incentives.
Individual incentives may come at the expense of group performance. If individual soccer players are rewarded by how many goals they score, they may prefer to make a shot from a long distance or at a sharp angle rather than passing the ball to someone who is in a better position to score. Since the group is small, the members may monitor one another. They might punish a player who does not pass the ball to players who are in a position to score by not passing it to him when he is in that position. The use of team bonuses might also, as in the case of waiters who pool their tips (Chapter 23), create an incentive to monitor players who do not pull their full weight. In other cases, neither of these corrective mechanisms may be available. In an incentive-based competitive environment, agents may not be willing to share their information with each other even though the principal would benefit if they did.
Incentive systems can backfire if they create a fear of a moving target. If a system succeeds in making employees work harder, they may fear that the higher-level effort will be the new benchmark. As noted in Chapter 23, this may lead to the emergence of norms against rate busting. In France, research fellows in the social sciences and the humanities at the National Council of Scientific Research (CNRS) have reacted in similar ways against what they claim to be an excessively individualistic and result-oriented approach. Although they have life tenure and no teaching obligations, the outcome of their grant applications may depend on the evaluation of their research; hence high standards are a threat. As is the case in the same disciplines in the Italian academic system, these sections of the CNRS are permeated by a norm against excellence.
A more general and more important problem is that many – some would say all – incentive schemes create an opportunity for gaming the system, in ways that range from cheating to exploiting loopholes.5 Let me cite some examples.
A famous caricature in the Soviet magazine Krokodil showed a nail factory that fulfilled its quota (specified in weight) by producing one gigantic nail.
If you cannot win the promotion tournament by your own effort, you can always trip up the competition.
A transplantation center that is judged by the success of its grafts can increase its budget allocation by refusing to accept patients with a bad antigen match.
In a British hospital, patients who should have waited no longer than four hours before they were seen were left sitting in ambulances because the clock did not start ticking until they entered the building.
Some British hospital managers inquire when patients intend to go on holiday and then offer an appointment during this period. Few patients cancel their holiday for medical reasons, preferring to postpone their appointment. Since the patients initiate these delays, their wait is not recorded.
“Coding creep”: if a major risk factor is recorded in a higher proportion of patients before surgery, the unit's predicted mortality will increase, as will the likelihood that the unit's actual mortality falls within or below the expected range.
When British schools were evaluated by the number of students who obtain good results in their exit exams, they responded by excluding more students, which led to more local petty crime.
The system of rewarding scholars by their peer-reviewed publications has created the concept of the “smallest publishable unit,” that is, the smallest part of an article that can be taken out and published separately.
When school boards use student test scores to punish schools, the effect may be, as one study found in Chicago, that teachers falsify the scores they report rather than making an effort to improve the performance of students.
When teachers or departments are rewarded for how many Ph.D. degrees they confer, academic standards, like water seeking the lowest level, tend to fall. The No Child Left Behind Act had similar perverse effects.
The New York police department reclassified some felonies as misdemeanors because the latter did not count in the crime statistics.
These examples, which again could be multiplied indefinitely, confirm the saying that performance targets are good servants, but bad masters.6
A much-debated question is whether the use of positive incentives by the principal can change preferences by undermining the intrinsic motivation of the agents. Although this effect plausibly operates in some contexts – one should not pay one's children to make their beds or do their homework – it is not clear whether it matters in an institutional setting. Negative incentives may certainly change the preferences of the targeted individuals. This effect is a small-scale version of the “psychology of tyranny” that I discussed in Chapter 2. People can resent negative incentives for the same reasons for which they resent monitoring. In the short run, compliance is increased; in the long run, initiative suffers.
Finally, incentive schemes may fail if the agent does not respond rationally, for one of the many reasons uncovered by behavioral economics. The self-sorting of individuals into high-risk and low-risk drivers is undermined if, as seems to be the case, a large majority of people believe they drive better than the average person. Although pay-for-performance incentives are widely used in the health sector, they have had little impact. The reason may be that programs are poorly designed and do not reflect what is known about the psychology of how people – including doctors – respond to incentives. One study examined a program designed to increase the number of women who receive a mammogram, and identified seven psychological mechanisms that may have contributed to its poor success rate. One failure was that the incentive payment was an incremental increase of the usual reimbursement, e.g. increasing the reimbursement per visit from $100 to $106. The designers of the scheme neglected the minimal psychological effect of this change, due to an irrational tendency to underestimate the difference between $100 and $106.7 The authors also make the ironic observation that attempts to incorporate elements from behavioral economics could actually backfire. As I noted in the discussion of the use of negative and positive incentives in the classroom, the former may be more effective because of loss aversion. Since incentive payments to doctors can also be structured either as a withholding (a perceived loss in income) or as a bonus (a perceived gain), performance should improve by using the first scheme. As in other cases of negative incentives there is a risk, however, that this advantage might be outweighed by the risk of angering physicians.
Constitutions
Constitutions can be studied from two perspectives, to understand their upstream causes or their downstream effects. I begin with some comments on the former.
The making of constitutions can be a collective action problem (Chapter 23), notably if it requires different social groups to agree on a common tax policy. In the United States before 1787 and in France before 1789, each state or each estate wanted to benefit from the public goods – infrastructure, law and order, national defense – funded by taxes, while contributing as little as possible. In America, many states refused calls from the Continental Congress to pay contributions to the common cause; in France, the nobility and many members of the Third Estate sought and obtained tax exemptions. After considerable turmoil, the constitutions that came into effect in, respectively, 1789 and 1791 imposed centralized tax structures.
The Federal Convention overcame another collective action problem, by establishing in the constitution that “Senators and Representatives shall receive a Compensation for their Services, to be ascertained by Law, and paid out of the Treasury of the United States.” Under the Articles of Confederation, travel and accommodation expenses of representatives to the Continental Congress were paid by the individual states, which often sent either no representatives or two (the minimum size of a delegation) to save money. Since a two-member delegation could not cast a vote if the two disagreed, the result was often paralysis, as Thomas Jefferson wrote to George Washington on March 15, 1784: “I suppose the crippled state of Congress is not new to you. We have only 9 states present, 8 of whom are represented by two members each, and of course, on all great questions not only an unanimity of States but of members is necessary. An unanimity which never can be obtained on a matter of any importance. The consequence is that we are wasting our time & labour in vain efforts to do business. – Nothing less than the presence of 13 States, represented by an odd number of delegates will enable us to get forward a single capital point.”
Constitution making also illustrates the variety of motivations that can animate social agents, and notably the interplay among interest, passion, and reason (Chapter 4).
Passions arise from the fact that constitutions tend to be written in times of crisis and turbulence. The conditions include war (Norway and France 1814, Germany, Czechoslovakia, and Poland after WWI, Germany, Italy, Japan, and France after WWII), revolution (France and Germany 1848), the fall of a dictatorship (Greece 1974, Portugal 1974, Spain 1976, South Africa 1996, many Latin American countries in the 1980s), the fear of a coup (France 1958), regime implosion (Eastern Europe after 1989), and financial crises (US 1787, France 1789, Hungary 2010, Iceland 2011). As I noted in Chapter 8, the emotions that are triggered include fear and enthusiasm. In Iceland, the predominant emotion was anger at the banks, which were perceived to be responsible for the financial crisis.
Reason – the impartial and rational concern for the long-term public interest – is both facilitated and hindered by passion. In Chapter 4 I cited La Bruyère, “Nothing is easier for passion than to overcome reason; its greatest triumph is to conquer interest.” One could add, as a corollary, that nothing is easier for interest than to overcome reason, except when reason allies itself with passion. As Kant noted, enthusiasm can produce this alliance. As he also observed, however, passion can be an obstacle to rational belief formation (see Chapter 8). By the mechanisms of wishful thinking and urgency, enthusiasm can lead agents to ignore obstacles to their goals that cool and deliberate reflection would have made obvious. In Iceland in 2011, the urge to do something in response to the financial crisis preempted the question whether making a wholly new constitution was the best thing to do. Framers may even ignore their own reason-based decisions, as when both the French framers in 1789 and the Norwegian framers in 1814 ignored their own rule that proposals should be subjected to several successive votes before being adopted. As Gibbon says about the measures taken to enforce the residence of elected bishops in the reign of Constantine, “the same passions which made those regulations necessary, rendered them ineffectual.”
Even though the alliance of passion and reason may override some interests, they are never completely eliminated. In modern constitution making, party interest is often decisive in shaping electoral laws. In earlier periods, class interest often shaped the limitations on suffrage and eligibility. At the Federal Convention, the economic interests of the Southern and Northern states shaped at least a dozen clauses in the final document. In a few cases, personal interests have played a role. The creation of a Senate in the Czech constitution of 1992 was not made on grounds of principle, but to create a place for the Czech deputies to the Senate of the dissolved Czechoslovak Confederation.
I now consider the structure and effects of constitutional texts. A constitution establishes the separation of powers. It prevents any single political actor from concentrating all power in its hands. The single actor may be an individual, a small group, or the people as a whole. The classical terms for their unconstrained power are tyranny (absolute monarchy), oligarchy, and mob rule.8 When constrained by the separation of powers, the regimes turn into constitutional monarchy, aristocracy, and democracy. Here, I consider only modern democratic constitutions. Typically, these have four parts. First, they determine and regulate the machinery of government. Second, they specify the rights and sometimes the duties of the citizens. Third, they lay down rules for amending the constitution. Finally, they stipulate procedures for suspending the constitution, or specified parts of it, in times of an emergency.
The machinery of government has many nuts and bolts, as well as cogs and wheels. The core institutions are the legislative, executive, and judicial organs. As Bentham argued, one should also include the electorate as an organ. Other important institutions include the national auditing office and the Central Bank. Many provisions relative to these organs are simply lists of their functions: the power to coin money, to raise taxes, to increase the money supply, to sign treaties, to enact legislation, to decide in civil and criminal cases, to vote in elections, and so on.
Other provisions involve relations to other organs. One may, perhaps, distinguish two bundles of such relations. One bundle aims at preventing one organ from trespassing on the domain of another. In Sweden, the government cannot instruct the Central Bank in matters of monetary policy. In some countries it is also prohibited from instructing the public prosecutor. A ban on bills of attainder prevents the legislative power from encroaching on the judiciary. Conversely, in France the fear of a “government of judges” blocked the judicial review of legislation for two centuries. The other bundle consists of mutual checks among these organs. An organ can check another to prevent it from making formally unconstitutional decisions, or members of one organ can check another to prevent it from making what they think (or claim) to be substantively bad decisions. Judicial review and bicameralism illustrate these two forms.9
All modern constitutions (except Australia's) include a bill of rights, or at least an enumeration of rights. Civil and political rights are usually defined vertically: they forbid the government from taking certain actions against the citizens. Only in South Africa is there an express commitment to the horizontal application of the constitution. In addition to these “first-generation rights,” modern constitutions increasingly include “second-generation rights” (economic, social, and cultural rights) and the more diffuse “third-generation rights” (e.g. the right to development). Whereas constitutional articles related to the machinery of government are (as a general rule) formulated so sharply that it is unambiguous what can, what cannot, and what must be done, articles affirming rights acquire (again as a general rule) implications for action only when filtered through statutory law or constitutional jurisprudence. I shall return to this question.
Virtually all constitutions provide procedures for their own amendment. Tocqueville observed that the French Charter of 1830 did not contain an amendment clause, from which fact he drew the conclusion that judicial review – “le gouvernement des juges” – would be too dangerous: “If courts in France could disobey laws on the grounds that they found them unconstitutional, constituent power would really be in their hands, since they alone would have the right to interpret a constitution whose terms no one else could change.” There may have been other wholly unamendable constitutions, but I doubt they are important. That being said, some constitutions contain individual articles that are unamendable.
If common sense suggests the need to be able to amend the constitution, it also suggests the need to make amendment relatively difficult. If the constitution were as easy to amend as ordinary laws, one might adopt an unconstitutional law by first changing the constitution, to make it constitutional, and then adopt the law. Generally speaking, however, legal systems do not enable agents to do in two steps what they are forbidden to do in one.10 With insignificant exceptions, it is in fact always more difficult to amend a constitution than to change ordinary laws. Delays and requirements of a supermajority are the most important devices. A delay means that the minimum time between the proposal of an amendment and its adoption is longer than in the case of ordinary legislation. To achieve this end, the constitution may require that the amendment be debated no earlier than one month from its introduction (Bulgaria), that it be subject to two readings and two votes (Brazil), that it be proposed in one parliament and adopted in the next one after a general election (Belgium, Norway), or that it be passed by two successive parliaments (Denmark, Estonia, Finland, Iceland, Sweden). Supermajorities range from three-fifths to three-quarters, with a two-thirds requirement being perhaps the most common. In Finland, Estonia, and Bulgaria, there is a trade-off between the length of the delay and the size of the supermajority. In the first two countries, there exist fast-track procedures by which the normal time-consuming process can be bypassed if a large majority (five-sixths in Finland and four-fifths in Estonia) declares the need to revise the constitution urgently and another majority (two-thirds in both countries) then votes to amend it. In Bulgaria, there exists a slow-motion procedure by which the normal supermajority of three-quarters can be reduced to two-thirds but with a longer delay. The Norwegian constitution may also be seen as expressing a trade-off, since ordinary revisions require both a delay and a supermajority of two-thirds, whereas the delegation of certain powers to an international organization requires a supermajority of three-quarters but no delay.
Just as constitutions regulate their own amendments, they sometimes regulate their own partial suspension during emergencies. The suspension may concern one or more of the other three parts of the document, regulating the machinery of government, individual rights, and the amendment process. In addition to determining the scope of the suspension, the constitution may also identify the grounds for suspension, the organ that decides on the suspension, and the procedures for ending the suspension. An important difference between amendments and suspensions of the constitution is that the latter may lack a basis in the constitution itself. The difference is not absolute: in 1962 de Gaulle tacitly amended the French constitution by the unconstitutional means of a referendum. Yet in cases of force majeure, bypassing the constitution is more common. The suspension of the French constitution of 1793 immediately upon its enactment, in favor of a revolutionary government, had no basis in the document. Saint-Just argued that “In the circumstances in which the Republic finds itself, one cannot establish the constitution; one would destroy it through itself.” Another famous example is Lincoln's suspension of habeas corpus in 1862, a decision that the constitution vests in Congress. His response to criticism – “Are all laws but one to go unexecuted, and the Government itself go to pieces lest that one be violated?” – was an echo of Saint-Just and in turn was echoed in Justice Robert Jackson's dictum, “The bill of rights is not a suicide pact.”
This overview provides the background for the question that is most relevant for the purposes of the present book: how does a constitution acquire causal efficacy? Would it not simply be a “rope of sand” (Cromwell) or a “parchment barrier” (Madison)? Why would the government behave like the proverbial chicken that stays inside the chalk circle it could easily transgress? According to the authors or inspirators of the 1799 French constitution, “Il faut qu'une constitution soit courte et obscure. Elle doit être faite de manière à ne pas gêner l'action du gouvernement.” (“A constitution should be short and obscure. It should be written so as not to interfere with the action of the government.”) In a modern perspective, of course, a main function of a constitution is to constrain the government. The question is how it can do so.
Actually, the question is too narrowly framed. A constitution does not merely serve the end of disabling the government in some respects, but also that of enabling it. A government in a country without a rigid (hard-to-amend) constitution will not be able to make credible promises. This was a major problem for the British parliament in the eighteenth century, when it was unable to make credible promises to the American colonies that it would not tax them in the future. In 1776, James Cannon wrote that “I call upon you to prove that Great-Britain can offer any plan of constitutional dependence which will not leave the future enjoyment of our liberties to hope, hazard, and uncertainty … By the constitution [sic] of Great-Britain the present Parliament can make no law which shall bind any future one … Is it wisdom, then, or is there safety, in entering upon terms of accommodation with a power which cannot stipulate for the performance of its engagement?” A rigid constitution can also enable the citizens to engage in long-term economic planning by removing the chilling fear that the government might confiscate their gains. Yet these observations, while correct, presuppose that the constitution is not only hard to amend, but also enforceable and credible. The question is how it can acquire these properties.
Before the near-universal adoption of judicial review to prevent unconstitutional actions by the legislature or the executive, various other methods were proposed, adopted, or observed. Cromwell proposed to give the executive the right to veto encroachments on its powers. Some deputies to the French constituent assembly of 1789–91, notably Mounier and Mirabeau, proposed to give the executive the right to veto all laws. In the nineteenth century, France (de facto) and Sweden (de jure) gave the president of the legislatures the right to prevent proposals they judged unconstitutional from coming to a vote. Each of these solutions amounts to making one part judge in its own cause. The British legal theorist Dicey, commenting on the lack of control of the constitutionality of laws in France around 1900, wrote that the “restrictions placed on the action of the legislature under the French constitution [of the Third Republic] are not in reality laws, since they are not rules which in the last resort will be enforced by the courts. Their true character is that of maxims of political morality, which derive whatever strength they possess from being formally inscribed in the constitution and from the resulting support of public opinion.”
That strength, however, may not be very great. There is evidence that citizens worry little about unconstitutional proposals if they approve of their substance. Commenting on the use of royal prerogatives that were widely seen as contrary to the (unwritten) British constitution, David Hume observed that:
In 1662, Charles, pleading both the rights of his supremacy and his suspending power, had granted a general indulgence or toleration; and, in 1672, he renewed the same edict: though the remonstrances of his parliament obliged him, on both occasions, to retract; and, in the last instance, the triumph of law over prerogative was deemed very great and memorable. In general, we may remark that, where the exercise of the suspending power was agreeable and useful, the power itself was little questioned: where the exercise was thought liable to exceptions, men not only opposed it, but proceeded to deny altogether the legality of the prerogative on which it was founded.
The statement is confirmed by de Gaulle's unconstitutional procedure in 1962, when he changed the constitution by referendum to make the president elected directly by the people. A large majority of the voters voted for the proposal rather than voting against it to punish him for the choice of procedure. Those who voted No mostly did so on grounds of substance, not of procedure. Another case of unconstitutionality in French politics occurred on May 29, 1849, when President Louis Bonaparte gave orders for French troops to march on Rome for the purpose of defeating the Roman Republic and restoring the papacy. The action was clearly a violation of the constitution of 1848, which says that “The French Republic … never uses its troops against the freedom of any people.” In the National Assembly, the radical deputy Ledru-Rollin stated on June 11 that “The constitution has been violated; we shall defend it by all means, even by arms,” to which the President of the Assembly Dupin replied, “The constitution cannot be violated in a more scandalous manner than when in a legislative assembly one talks about defending it by arms.” When Ledru-Rollin called for an armed demonstration on June 13, only a few thousand unarmed persons showed up and were quickly dispersed by force.
The modern solution to this problem is to assess the constitutionality of laws by judicial review entrusted to constitutional courts or supreme courts. The existence of a court with the uncontested power to exercise judicial review does not, however, resolve the question of the causal efficacy of the written constitution. Like the constitution itself, a decision by the court is just a piece of paper. Courts exercising judicial review usually do not have a police force at their disposal to enforce their decisions. (Nor, for that matter, do they have a separate budget to fund decisions with important economic consequences.) Although the US Supreme Court can ask the President to send federal marshals to enforce its decisions, he can refuse, as did President Jackson in Worcester v. Georgia (1833). In an apocryphal story, he is reported to have said, “Well, John Marshall [the Chief Justice] has made his decision, now let him enforce it.” He did say, though, that the decision was “still born.” The decision in Brown v. Board of Education (1954) would not have been enforced had President Eisenhower not sent federal troops to Little Rock. Undercompliance with decisions by the constitutional court has also been documented in South Africa and in Russia.
In the American cases, opposition to the Court was rooted in the conflict between the states, particularly the Southern states, and the federal government. It might seem hard to imagine similar attempts to sabotage the decisions in, say, contemporary France or Norway. In the case of a priori review of legislation, that is, review of legislation before it is promulgated, they are virtually unthinkable. In the case of a posteriori review, that is, review that arises out of a specific legal case, decisions that entail large expenditures by the authorities might be met with the answer, made in good or bad faith, “We don't have the money.” Although the enforcement of economic, social, and cultural rights might be particularly expensive, the enforcement of civil and political rights can also require substantial outlays. For this reason, the government might, for instance, refuse to comply with court-ordered measures to reduce prison overcrowding.
The last example can be used to introduce another reason why the constitution may lack causal efficacy. Reacting to the arbitrary practices of the courts of the ancien régime, the parlements, Montesquieu said that judges “are no more than the mouth that pronounces the word of law (la bouche de la loi), mere passive beings, incapable of moderating either its force or rigor.” One interpretation of his statement is that the law, including the constitution, always has a unique meaning that is to be determined by the judge. Some constitutional provisions no doubt come close to having this character, notably those that specify the machinery of government. These clauses provide hard constraints on governmental action. Provisions stating individual rights, by contrast, are notoriously ambiguous. The Eighth Amendment to the American constitution prohibits the use of “cruel and unusual punishment.” One might ask whether “double-celling” in prisons – two prisoners sharing one cell – is banned under this clause. In Rhodes v. Chapman (1981), a majority on the US Supreme Court found that it was not, Justice Marshall dissenting.
It seems crystal clear to me that the Eighth Amendment by itself provides no unambiguous resolution of the “double-celling” issue. The majority and the minority formed their opinions on the basis of a vast body of prior jurisprudence, most of which has an equally tenuous relation to the text of the constitution, and on their personal ideas of what constitutes unusual and cruel punishment. Although both the majority and the minority cited an earlier decision affirming that the state cannot impose punishment that violates “the evolving standards of decency that mark the progress of a maturing society,” they drew opposite conclusions from that shared premise. In Roe v. Wade (1971), the Supreme Court very dubiously found a right to privacy in the constitution and a right to abortion in the right to privacy. The idea of “substantive due process” has equally poor foundations in the text of the constitution, as do the Supreme Court decisions about what counts as “speech” in the First Amendment and about exactly which right to bear weapons follows from the Second Amendment (see Chapter 9).
Many similarly creative decisions could be cited from other countries. Here, I consider two pairs of decisions by the Colombian Constitutional Court. First, consider the contrast between decision C-221/1994 allowing personal drug consumption and decision C-309/1997 allowing the mandatory use of safety belts. The question in both cases was whether the law could restrict personal autonomy for the purpose of protecting individuals against themselves. In the first decision the Court affirmed that it could not, in the second that it could. The argument in the first decision was that if the state wanted to reduce drug consumption, it should use the less restrictive means of education. This idea is so removed from reality that one must conclude that the Court was in the grip of an ideology. Could not drivers also be “educated” to use safety belts?
Second, consider the contrast between decision C-1040/2005, which authorized a second term for President Uribe, and decision C-141/2010, which denied him a third term. In the first decision, the Colombian Court distinguished between “amendment” of the constitution and “substitution” of one constitution for another. While it denied Congress the right to substitution, it accepted its right to amendment and judged that the extension of the one-term limit was only an amendment. In its 2010 decision, it struck down a law calling for a referendum on an amendment to the constitution for a two-term extension. On the Court's English website, the two decisions are contrasted in these terms: “The [2005] decision … reviews, in great detail, all the procedural aspects of the amendment's transit through Congress. But its most important feature is the way in which it applied, to a specific case, the ‘substitution theory’. Five years later, confronted with another constitutional amendment that allowed for a third consecutive term, the Court decided against it. Two terms is not a substitution, but three terms is.” The distinction seems arbitrary. The motivation behind the decision was probably the desire of the judges to keep Uribe out of power.
As these examples illustrate, because of the abstract and often vague language of many rights provisions in the constitution, they can indeed be “adapted to our concerns,” as Montaigne said (Chapter 9). The conclusion generates the premises. To the (unknowable) extent that this is the case, the constitution has no independent causal efficacy.11 Even when it is not the case, and judges do their best to decide cases on their merits rather than twist the premises to reach a predetermined conclusion, they are much more constrained by prior constitutional jurisprudence than by the constitution itself. Moreover, as I noted, that prior jurisprudence itself had often very tenuous links with the original text. We can easily imagine a counterfactual world in which many key decisions were made differently, because of the death and replacement of swing judges, with the cumulative effect of creating a constitutional jurisprudence wholly different from the one under which we live today.12 The constitution itself, though, would be the same.
So far I have focused on the causal efficacy of the rights provisions in the constitution. Some provisions regulating the machinery of government definitely have causal efficacy, certainly by affecting opportunities and perhaps by affecting desires (see Chapter 10 for this distinction).
Regarding opportunities, the Twenty-Seventh Amendment, proposed in 1789 and adopted in 1992, states that “No law, varying the compensation for the services of the Senators and Representatives, shall take effect, until an election of Representatives shall have intervened.” Before the adoption of that amendment, self-dealing by members of congress was kept in check by their desire to be reelected, together with the belief that voting themselves a salary increase might thwart that desire.13 If the constitution does not allow the executive to dissolve parliament and call for new elections, it removes an arrow from its quiver.
Regarding desires, it is a cliché that a bicameral system will favor wise decisions, by virtue either of the superior virtue and ability of the senators or by the sheer fact of slowing down the process to let emotions cool off. I do not know of any empirical analyses of this alleged effect, but some momentous American examples suggest doubt. In 1798, the Sedition Acts passed the Senate by a wide margin, but obtained a bare majority of 44 to 41 in the House of Representatives, and then only after trial by jury was substituted for trial by judge. Many legal scholars think that the Acts would have been found unconstitutional if judicial review had been established at the time. In 1964, the Resolution of the Gulf of Tonkin passed the House by 416 votes to 0, and the Senate by 88 votes to 2. The “Authorization for Use of Military Force Against Iraq Resolution of 2002” passed the House by 297 votes to 133 and the Senate by 77 votes to 23. In none of these cases did the Senate show much resistance to the whipped-up atmosphere of hysteria. These three decisions are usually viewed as embodying the opposite of wisdom. Since the American Supreme Court is also supposed to act as a brake on impetuous decisions, it is worth mentioning that it approved Roosevelt's executive order to intern Japanese Americans by six votes to three.14
Constitutions, like incentive systems, can be gamed, as some examples will show.
Article III.1 of the US constitution states that the “Judges, both of the supreme and inferior Courts, shall hold their Offices during good Behavior, and shall, at stated Times, receive for their Services, a Compensation, which shall not be diminished during their Continuance in Office.” At the Federal Convention, Madison insisted that the salaries of judges should neither be diminished nor increased during their time in office, since Congress could bribe judges as well as threaten them. He overlooked the fact that even with the clause as it stands, Congress can game it by increasing the salaries of some judges less than those of others. In 1964, a Congress hostile to the Warren Court increased the salaries of lower federal judges by $7,500, but those of Supreme Curt Justices by $4,500 only. They were not increased again before Warren's departure in 1969.
Until the adoption of the Seventeenth Amendment in 1912, elections to the US Senate were indirect. The people of the several states elected the state legislators, who elected the US senators. Yet, as the British legal scholar James Bryce explains,
In 1904 Oregon provided, by a law passed by the people under the initiative method of legislation contained in the constitution of that state, that the political parties might in the party primaries nominate persons for election as United States senators, and that the people might at the ensuing election of the state legislature select by their votes one of these nominees as their choice for senator. Along with this it was also enacted that a candidate for the state legislature might on his nomination either: (1) declare that he would, if elected, vote for that person as United States senator who had received the largest popular vote and thus become “the people's choice”; or, (2) declare that he would consider the popular vote as merely “a recommendation.” Or he might make no declaration at all. In 1908 a majority of the members elected to the legislature, having made the former declaration, felt bound to carry it out, and the person who had received the highest popular vote was accordingly elected by that majority, although he was a Democrat and they were Republicans. Thus the people got their way and the federal Constitution was not formally transgressed.
The French constitution of 1958 states that “Members’ right to vote [in the National Assembly and the Senate] shall be exercised in person. An Institutional Act may, in exceptional cases, authorize voting by proxy. In that event, no member shall be given more than one proxy.” Since the large majority of deputies and senators have one or more local elective offices in addition to the national one (the cumul des mandats), they were motivated to oppose this restriction and successfully gamed it virtually from the outset. They did so by adding to the initial and precisely formulated exceptions the diffuse exception of “force majeure.” Since it was left to the bureaux of the assembly to decide whether this case obtained, and most representatives had a strong interest in being able to absent themselves from time to time, permissions were readily granted. When consulted in 1961, the Constitutional Council naively or cynically affirmed that the bureaux could be trusted to verify that the constitution would be “strictly applied.” In practice, the exceptions have proven far more numerous than the rule. Like the example from Oregon, this case confirms Bryce's statement that it is hard “to keep even a written and rigid constitution from bending and warping under the actual forces of politics.”
I have discussed how constitutions can fail to produce the effects that their framers wanted them to have. One may also consider the opposite case, the unintended consequences of constitutions. I shall discuss three cases, which turn on the fact that the conjunction of two innocuous articles may have absurd or undesirable consequences.
According to Article I.3.4 of the American constitution, “The Vice President of the United States shall be President of the Senate.” According to Article I.3.6, the Senate “shall have the sole power to try all impeachments.” It follows that if the Vice President were to be impeached, he would preside over his own trial. This situation has not arisen, although it was perhaps not far from happening in the case of Spiro Agnew. According to the French constitution, on leaving office a President becomes automatically member for life of the Constitutional Council. Since nothing prevents a former President from standing for reelection, as Nikolas Sarkozy has announced he will do, and as the constitution does not require him to stand down from the Council if he were to be reelected, he might in theory be in the anomalous position of being head of the executive and also a member of the highest judicial body.
These are theoretical dangers. A vastly more consequential conjunction of two clauses that taken separately seem innocent enough is found in the Weimar constitution. Article 48 said that “In case public safety is seriously threatened or disturbed, the Reich President may take the measures necessary to reestablish law and order, if necessary using armed force … The Reich President has to inform Reichstag immediately about all measures undertaken which are based on paragraphs 1 and 2 of this article. The measures have to be suspended immediately if Reichstag demands so.” The last clause was presumably intended to provide a check on presidential discretion. It was undermined, however, by Article 25: “The Reich president has the right to dissolve the Reichstag, but only once for the same reason. New elections, at the latest, are held 60 days after the dissolution.” Using Article 25, the President could (and did) threaten to dissolve the Reichstag should it vote to annul any measures taken under Article 48. This mechanism opened the way to power for Hitler.
In a more paradoxical case, a constitution that was not intended to “interfere with the actions of the government” suddenly became causally efficacious after a regime change, acquiring, as it were, life only after death. In Czechoslovakia, the 1968 constitution introduced, for the first time in the history of the country, a federal structure with separate assemblies (National Councils) for the Czech and Slovak lands and with strong power for these republics in the bicameral Federal Assembly. Like all Communist constitutions, this one remained a dead letter; the National Councils were not even convened. The post-Communist constitutional debates after 1990 were framed, however, by this document. The strong Slovak autonomy now became a major obstacle to reform. An amendment to the constitution required a three-fifths majority both in the proportionally elected lower house and in each of the two Czech and Slovak sections of the upper house. Although the Czech population outnumbered the Slovaks two to one, each section had 75 seats. The power of thirty-one Slovak deputies in the upper house – representing one-fifth of the voters – to block change was arguably a main cause of the break-up of the Federation in 1992. President Havel could have used his enormous moral authority to elect a unicameral constituent assembly, but he maintained the existing assembly, purged of many Communists but retaining its bicameral structure.
While constitutions may fail to have causal efficacy and may have causal effects that were not intended, these facts should not obscure the real and predictable importance of many constitutional provisions. When constitutions guarantee the overrepresentation of certain regions in parliament, as many do, one can expect these to get more roads and bridges than they would otherwise have received. If they forbid the government from instructing the Central Bank in matters of monetary policy, inflation is likely to be lower than it would otherwise have been. Unemployment, though, may be higher. If they ensure freedom of the press and free elections, the government may be reluctant to misbehave in ways that, when exposed by the media, would make it lose the next election.15 If they authorize patents and ban retroactive laws, there will be more economic growth. Civil and political rights create a hard core of freedoms that no constitutional jurisprudence can undo, although much will remain uncertain and sometimes arbitrary at the margins. Constitutions matter, but less than many constitutional designers and scholars believe or would have us believe.
Bibliographical note
The problem of aligning individual and organizational incentives is the topic of J.-J. Laffont and J. Tirole, A Theory of Incentives in Procurement and Regulation (Cambridge, MA: MIT Press, 1994). A comprehensive handbook on corruption is A. Heidenheimer, M. Johnston, and V. LeVine (eds.), Political Corruption (New Brunswick, NJ: Transaction Publishers, 1989). The references to nineteenth-century English cooperatives are taken from B. Jones, Co-operative Production (Oxford University Press, 1894; New York: Kelley, 1968). The comment on the high turnover in government is from Chapter 22 of A. King and I. Crewe, The Blunders of our Governments (London: Oneworld Publications, 2013). The relative importance of trust and incentives in firms is discussed in E. Fehr and A. Falk, “Psychological foundation of incentives,” European Economic Review 46 (2002), 687–724. The study of “earners” and “maintainers” is T. Docan, “Positive and negative incentives in the classroom,” Journal of Scholarship of Teaching and Learning 6 (2006), 21–40. The incentive example featuring a mayor and her police chief is taken, with slight modifications, from G. Miller and A. Whitford, “The principal's moral hazard: constraints on the use of incentives in hierarchy,” Journal of Public Administration Research and Theory 17 (2007), 213–33. Two useful studies of gaming of incentive systems are Z. Radnor, “Muddled, massaging, manoeuvring or manipulated? A typology of organisational gaming,” International Journal of Productivity and Performance Management 57 (2008), 316–28, and D. Pitches, A. Burls, and A. Fry-Smith, “Snakes, ladders, and spin,” British Medical Journal 327 (2003), 1436–9. The Italian norm against excellence is illustrated and explained in D. Gambetta and G. Origgi, “The LL game: the curious preference for low quality and its norms,” Politics, Philosophy & Economics 12 (2013), 3–23. The French equivalent is exposed in F. Tagliatesta (a pseudonym for Pascal Engel), Instructions aux académiques (Rouen: Christophe Chomant, 2005). The study of the effects of pay-for-performance schemes is I. Siva, “Using the lessons of behavioral economics to design more effective pay-for-performance programs,” American Journal of Managed Care 16 (2010), 497–503. The discussion of constitutions and constitution making draws on Chapter 4 of my Securities Against Misrule (Cambridge University Press, 2013). For the number of swing votes in Supreme Court decisions, see C. Sunstein, “Unanimity and disagreement in the Supreme Court” (unpublished manuscript, 2014).
1 Both the Roman and the Chinese empires adopted two procedures that were not vulnerable to this problem. In China, the “law of avoidance” forbade officials from serving in their native province. To crush a rebellion in one province the governments used troops from another. In Rome, archers on horseback were recruited from Palmyra, but deployed in the Sahara. On June 4, 1989 the Chinese government used a similar stratagem.
2 According to Gibbon, under the reign of Valens and Valentian judges “easily discovered that the degree of their industry and discernment was estimated, by the Imperial court, according to the number of executions that were furnished from their respective tribunals.” In contemporary societies, quotas seem to be the more common pathology of legal incentive systems.
3 That minimal effort has to be observable. This seems realistic: it is easy to observe whether an agent performs routine duties, but not whether he shows initiative.
4 Similarly, the norm that was prevalent in eighteenth-century America, that seeking political office was a sign of being unworthy of it, left voters with no useful information.
5 Providing agents with information can also have perverse effects. When American officers in Vietnam provided information about the location of Viet Cong soldiers to the South Vietnamese army, they sometimes used it to go where there were no guerrillas.
6 Organizational failures can also arise if officials fall victim to what I called “the younger sibling syndrome” (Chapter 13) and neglect the fact that people respond to incentives. This is a cruder mistake than neglecting the fact that they can respond to incentives in more than one way.
7 A more vivid example: people who would normally weigh carefully the quality difference between stereo players costing $100 and $150 usually do not care about the difference between a $20,000 car with the $100 player and the same car with the $150 player.
8 Objecting to the tendency to blame the sovereign for all disorders, Hume wrote that “As if the turbulence of the great, and madness of the people, were not, equally with the tyranny of princes, evils incident to human society, and no less carefully to be guarded against in every well regulated constitution.”
9 The second bundle cannot be neatly distinguished from the first. For one thing, if organ A checks organ B on the grounds that organ B has acted unconstitutionally, the act may be one that encroached on the constitutional powers of organ A or of some other organ C. For another, the dividing line between checks and encroachments is relative. The right of the American Senate to veto the appointment of high federal officials is usually presented as a desirable check on the executive, whereas the right (sometimes claimed in the past) also to veto their removal has been seen as an inappropriate encroachment. In other systems, even the first right would be seen as an encroachment. Judicial review can be assessed either negatively as an encroachment on the domain of the legislature, or positively as a check on its activities. This ambiguity is pervasive. For those who follow Bentham in thinking that the lower house of parliament should be omnipotent, any check on its power is an encroachment. His view is at one extreme of a continuum. The American constitution with its triple check on the lower house – by the upper house, by the president, and by the Supreme Court – may be at the other extreme.
10 Violations of this principle do occur. “In the twenty-third of Henry VI. a law … was enacted, prohibiting any man from serving in a county as sheriff above a year, and a clause was inserted, by which the king was disabled from granting a dispensation. Plain reason might have taught, that this law, at least, should be exempted from the king's prerogative: But … in the reign of Henry VII. the case was brought to a trial before all the judges in the exchequer-chamber; and it was decreed, that, notwithstanding the strict clause abovementioned, the king might dispense with the statute: He could first, it was alleged, dispense with the prohibitory clause, and then with the statute itself. This opinion of the judges, though seemingly absurd, had ever since passed for undoubted law” (Hume). Two other instances were the decision by Tiberius (as reported by Tacitus) to kill the daughter of his enemy Sejanus and the execution of young women in the Iran of the Ayatollahs (as reported in the Boston Globe of August 19, 2009). In both cases, it was illegal to execute virgins; in both, the problem was circumvented by first raping them.
11 Even the interpretive norm of respecting “the plain meaning of the text” does not always give a clear-cut answer. In twenty-one Supreme Court cases decided with an opinion during the spring of the 1993 term, conflict between majorities and dissents derived at least in part from disagreements over the plain meaning of the statute at issue.
12 Until 1941, swing votes (decided five to four) were rare. After that date, about 16 percent of the cases have been decided with the majority of one vote.
13 In 1816, there was such a high degree of citizen indignation when legislators voted themselves a pay increase that almost two-thirds of them failed to be reelected, even though they had hastily repealed the compensation law in the meantime.
14 In 2001 the Patriot Act passed 98 votes to 1 in the Senate and 357 to 66 in the House of Representatives.
15 The government can, however, game the constitutional guarantee of freedom of the press by rationing paper or printer's ink and allocating them preferentially to newspapers supporting it. In 1793, the British Attorney General tried to limit the impact of an attack on Edmund Burke by telling the author to publish his work in an expensive edition, “so as to confine it to that class of readers who may consider it coolly”; otherwise, it would be his duty to prosecute. The government can also game free elections, by choosing times and places for voting that are inconvenient to groups of voters who are likely to vote against them. I mention an example in Chapter 10.