Volume 14 Number 2, Fall 1994

A SURVEY, CRITIQUE, AND NEW DEFENSE OF TERM LIMITS

Alexander Tabarrok

Term limits are an important aspect of the U.S. political system. Since 1990, 21 states have passed congressional term limit laws. [1] Term limits on governors have been common since the founding of the United States. The president was limited by constitutional amendment to two terms in 1951 and by tradition prior to Franklin Roosevelt. Many states limit the terms of state politicians and many mayors and other local officials are also term limited.

Arguments about term limits have advanced far less quickly than term limits themselves. Most of the arguments concerning term limits are weak and the best arguments in favor of term limits are not used by term limit proponents. I survey the arguments for term limits, demonstrate their weaknesses, and propose a new theory.

Incumbency Advantage

All of the arguments for term limits are premised on the costs of incumbency advantage. There is little disagreement about the importance of incumbency advantage and its increase over the past several decades. Gary King and Andrew Gelman (1991), for example, find that incumbency increases a representative's expected vote share by 11 percent. There are many possible explanations for why incumbents are advantaged but none are entirely satisfactory. Gerrymandering (redistricting) has undoubtedly benefited some representatives (perhaps at the expense of others) but incumbent senators also have an advantage over challengers and this cannot be explained by the redistricting hypothesis (Reed and Schansberg 1992). Political Action Committees (PACs) give nearly eight times as much money to incumbents than challengers. Partly as a result, incumbents spend more than two and half times as much as challengers (Maisel 1990, Abramowitz 1991). But do PACs give money to incumbents because they know the incumbent is more likely to win or are incumbents more likely to win because PACs give them more money? Some degree of co-causation is undoubtedly involved but its precise nature is difficult to identify.

Although the causes of incumbency advantage are uncertain there is a large body of evidence which shows that incumbency advantage is important and has been increasing over time. Incumbents are more likely to be reelected than challengers, they have higher margins of victory, and they participate in more unopposed elections than do challengers (Mayhew 1974, Maisel 1990, Reed and Schansberg 1992).

What are the costs of incumbency advantage and consequent lengthy terms? Term limit proponents allege incumbency advantage creates politicians who shirk their responsibilities and deviate from the wishes of the voters. I refer to this as the shirking-deviation theory. The hypothesis is that incumbency advantage serves as a relaxation of the election constraint: As reelection becomes more likely, politicians can afford to ignore the wishes of the voters. A second and related hypothesis states that even a superior challenger cannot replace an inferior incumbent. This hypothesis does not require that the incumbent serve the public less well in the ninth term as in the 1st term; even if quality of service remains constant, incumbency advantage implies that it is difficult to replace an incumbent with a superior challenger.

A central problem with all of these arguments is that the people who are being implored to vote for term limits are the same people who reelect their representatives. Ultimately, it is the voters who are responsible for incumbency advantage. Term limit proponents treat incumbency advantage as if it were an exogenous force imposed upon the voters by a nefarious power. They often fail to ask, "If incumbency advantage creates unresponsive politicians why do voters continue to reelect their representatives?" And if the question is asked at all, the answers given are unconvincing. Most arguments for term limits gloss over this issue and for the moment I shall follow this practice. After surveying a broad spectrum of arguments for term limits I shall return to this problem and analyze it in detail. Surprisingly, a proper understanding of the problem leads to a new conceptualization of term limits and their benefits.

With these preliminaries on incumbency advantage kept in mind let me turn to the arguments for term limits.

A New Political Man?

Edward Crane (1990) and James Coyne and John Fund (1992) argue that term limits would remove the professional politician from Washington and bring back the superior "citizen-statesman" and the "citizen legislature." They argue that term limits would make politicians more representative and more responsive.

The main argument that suggests term limits will make politicians more representative is that a term-limited political system will attract or "select" different types of politicians than an unlimited system. In an unlimited system most politicians who run for Congress are looking for a career. In a term limit system most politicians who run for Congress will be looking for, at most, a 12-year job. [2] Career politicians, it is assumed, have different preferences than non-career or citizen-politicians. This statement is rarely justified but the reasoning seems to be that a non-career politician will be more aware of the results of his actions on the private sector. Coyne and Fund (1992: 17) approvingly quote George McGovern, formerly the 1972 Democratic presidential candidate and now a hotel owner, who says, "I wish I had known a little more about the problems of the private sector . . . I have to pay taxes, meet a payroll--I wish I had a better sense of what it took to do that when I was in Washington." Reduced to essentials, the selection effect argument says that term limits will create a more diverse legislature and, in particular, a legislature with more business experience.

Historically the House was not more diverse when turnover was high. During the 19th century, when turnover rates were four to six times greater than they are today, lawyers made up over 60 percent of the House, more than at any other time before or since. The occupational background of representatives has remained remarkably stable since the founding of the United States. Lawyers have always made up about 50 to 60 percent of the House and since the 1850s those with business backgrounds, the largest group after lawyers, have constituted about 20 percent of the House (Bogue et al. 1976). Observed changes in turnover are much greater than those which would be created by term limits. No argument has been made which explains why changes created by term limits should lead to greater effects on the diversity of the House than have been observed from natural tenure changes in the past.

Most characteristics of representatives have been as stable across historical tenure changes as has been their occupational background. Those characteristics which have changed have done so in pace with changes in the suffrage. If a citizen-legislator is a representative who shares similar characteristics with her constituents then it is clear that we have more citizen-legislators today than ever before. Not only are women and minorities better represented in the late 20th century but representatives are more likely to come from the same "class" as voters (as measured, for example, by education levels; Bogue et al. 1976). The political class of the 19th century had fewer citizen-legislators than today and the process as a whole was more restrictive. Perhaps these factors made the politics of the 19th century preferable to those of today but few people, term limit proponents or otherwise, appear willing to make this argument.

James Payne (1991b) argues that term limits will make politicians more responsive because long-tenured politicians become "socialized" to big spending. He notes that at congressional hearings witnesses advocating government spending overwhelm their opponents by a ratio of 145 to 1. Special interests provide lots of information about the benefits of spending but little information about the costs. Similarly, the pressure to go along with spending increases is intense, the pressure to cut spending is diffuse. Faced with this imbalance of information and pressure Payne argues that representatives become "socialized" to higher spending and special interest influence. [3] In support of his argument, Payne (1991a) presents evidence that spending increases with seniority. That is, regardless of party affiliation, a senior representative tends to vote for more spending than a junior colleague. [4]

The socialization argument is not the only explanation for a positive relationship between spending and seniority. Political power, for example, increases with seniority, so higher spending could be due to more power rather than more seniority per se. If this explanation of the spending-tenure correlation is correct then term limits would sunder the relationship between seniority and power but would leave unaltered the more fundamental relationship between power and spending. With more politicians coming to power sooner, term limits could increase spending.

Robert Reed and Eric Schansberg (1993) carefully distinguish between the different arguments for a spending-tenure relationship and they look for indications in the data which would suggest one explanation is better than the other. Unfortunately, when the data is split into Republican/Democrat and House/Senate components they find that no single hypothesis can explain the data. One hypothesis works better for Democrats, another for Republicans and so forth. The findings of Reed and Schansberg (1993) call into question the existence of a true relationship between spending and tenure.

Neither of the arguments for term limits and the new political man are persuasive. The occupational composition of the legislature is unlikely to be greatly affected by term limits. More work on the spending-tenure hypothesis would be useful. However, with the evidence "to date," this hypothesis is shaky ground on which to base support for term limits.

Interest Groups and Term Limits

Proponents of term limits argue that term limits will reduce the power of special interests. The citizen-legislator hypothesis for this effect has already been discussed. A more plausible argument is that term limits increase the cost of bargaining and making commitments. As these costs increase the amount of special interest legislation is reduced.

Since political exchange is rarely simultaneous or subject to enforceable contracts there is potential for opportunism. Will the representative exert high effort for the interest? Will the interest fulfill its promises of support? Answering these questions in the affirmative requires that there be mutual trust between interest and representative, which can only be created in a long-term relationship. [5] Similarly, log-rolling requires that representatives know and trust one another. Term limits will increase the turnover rate for committee chairs and indeed for the entire legislature--every 12 years the legislature will be composed of entirely new representatives--thereby making trust difficult to build up. Long tenure also allows committee chairs to develop "property rights" in certain areas of legislation (Weingast and Marshall 1988). With term limits, property rights in policy areas will be more difficult to establish and this will make it difficult for politicians to logroll and trade political support.

Furthermore, even if term limits are long--potentially long enough for trust to be created--the fact that the final period is known can disrupt the evolution of cooperation. In a prisoner's dilemma type game if the players know the final period the cooperative strategy can never be rational. In the final period of the game defection is rational and backwards induction show that this is also true of the first period (Kreps 1990). Many problems that representatives face are like a prisoner's dilemma: it is better for each representative to defect from a political bargain rather than cooperate but when both representatives defect both are worse off. Factors other than an uncertain final period can make cooperation rational so this result does not imply that rational cooperation will disappear with term limits. [6] Nevertheless, it is easy to see that representatives in their final terms will be more tempted to break bargains than those in their first term and that this will make cooperation more difficult.

In addition to making it more difficult for representatives to bargain with interest groups, as well as among themselves, term limits increase the costs of committing the legislature through time. Legislation passed in the 98th Congress is unlikely to be overturned by the 99th Congress if the large majority of representatives are reelected. Legislation is more likely to be overturned when turnover rates are high. Although the possibility of overturning reduces the total demand for special interest legislation the number of bills passed may actually increase. [7]

These factors all work towards reducing the power of special interests. Lengthy terms, however, are only one way of producing long-term bargains. Substitutes exist, and these will be used increasingly in a term limit world. Political parties, for example, can substitute for lengthy terms in the production of long-term bargains. Instead of working through individual representatives special interests could lobby through the parties. Political parties in turn could use their growing power and wealth to discipline politicians. Politicians today have independent power bases and can afford to ignore their parties. Without lengthy terms, independence becomes harder to acquire and the resources of a stable, long-lived, political entity like a political party become more valuable and less easy to ignore.

The same forces which increase the cost of special interest legislation also increase the cost of producing public goods. Public good production requires trust and long-term bargains. A reduction in the power of special interest groups because of an increase in the cost of making bargains is not, therefore, necessarily welfare enhancing. Moreover, since public goods are probably already underprovided (the free rider problem exists in politics just as it does in markets), a reduction in public goods production is likely to be more costly than the benefit from reduced special interest legislation. In a model developed by Tyler Cowen, Amihai Glazer, and Henry McMillan (1994) private (special interest) goods must be tied to public goods in order to provide an incentive for politicians to produce public goods. The public good of national defense, for example, is provided because military contractors have an interest in lobbying for national defense. If this model is correct, a reduction in special interest legislation is not necessarily to be applauded.

Finally note that greater rotation in office increases the incentives of politicians to pass legislation which benefits their own post-Congress careers. This is just the negative spin on the pro term-limit argument that term-limited politicians will be more concerned about the effects of their actions on the private sector.

In summary, term limits increase the cost of lobbying through individual politicians. Special interests have more than one means of lobbying, however, and in a post term limit world they will turn to lobbying through the parties and other more long-lived actors. The final effect on special interest politics is likely to be small. To the extent that special interest politics is reduced we should also be concerned that public goods production may be reduced.

The Executive and the Bureaucracy

Term limits will reduce the power of Congress compared with the bureaucracy and executive. [8] Although term limit proponents typically do not argue that this is an advantage of term limits, much of the support for term limits does comes from dissatisfaction with Congress. I discuss how term limits reduce the power of Congress and then evaluate the merits of this effect.

Congress can exert control over the bureaucracy through the committee and budget appropriation process, but such control requires information (Weingast and Moran 1983). The less information the committee has relative to the bureaucracy the closer one gets to the conditions assumed in William Niskanen's (1971) monopoly bureau model. Term limits will reduce the average tenure of committee members. Reduced tenure implies less experienced representatives who require more information and assistance from the civil service (including congressional staff). Also, the ability of the bureaucracy to safely ignore or waffle on implementation of legislation could increase if key committee members are known to be nearing the end of their terms. In the United States, senior officials in the civil service have an average tenure of 23 years (Aberbach, Putnam, and Rockman 1981: 68). Similar lengths of term are not uncommon among Senators or even Congressmen in the present unlimited system but would be impossible under a term limit system. [9]

Term limits reduce the ability of Congress to effectively monitor and control the bureaucracy but this does not necessarily create an independent bureaucracy. The bureaucracy has often managed to maintain some degree of independence by playing the executive and Congress against one another. Executive control over the bureaucracy may grow as congressional control declines. Presidential tenure is eight years at most and presidential appointees, especially those brought in from the private sector, often exit government as soon as they gather enough human capital to be effective private lobbyists (Chandler 1990). Representatives and bureaucrats tend to have much longer tenures and hence are able to found long-term tacit alliances with one another. Aberbach et al. (1981: 235) note that "cabinet officials often are obstacles to be circumvented in accord with long-standing informal treaties among more durable actors." As was noted above, term limits reduce the average tenure of representatives and make understandings between Congress and the bureaucracy more difficult to establish, thereby increasing the relative power of the executive branch. Term limits in other words could move the United States towards a more European system with an "active" executive and "reactive" parliament. [10]

One facile argument for term limits is that they are necessary to create a "balance of power" because presidents are already term limited. A "balance of power," however, need not imply equal treatment. Many factors have worked to change the balance of constitutional powers since the ratification of the constitution. [11] Among these are the continually growing size and diversity of society and the increasing demands on Congress's time and resources. These have tended to weaken Congress as it has become less unified and has had to assign more responsibilities to the executive branch and the bureaucracy (Aberbach 1990). The increasing average tenure of representatives has been a force working in the opposite direction. Security of tenure has given congressional leaders a strong foothold from which presidential and bureaucratic power can be resisted.

Supporters of presidential power argue that a weak Congress would be beneficial. Without a single decisionmaker able to exercise leadership, chaos reigns. If only Congress would not interfere, the nation's problems could be handled as swiftly and efficiently as the Gulf war. Term limits in this view are an explicit and rational vote against Congress and for presidential power.

In the face of political gridlock the presidential power thesis can look compelling. Congress, however, is the preeminent representative institution of the U.S. government. It is factionalized and divided because the nation is composed of many parts each with its own interests. Congress acts slowly and sometimes indecisively because it requires a measure of consensus to make a decision. Transaction costs in Congress are an unlikely explanation for its failure to act (Weingast and Marshall 1988, Wittman 1989). Thus, if there were political actions which could benefit everyone or nearly everyone, as the presidential power thesis suggests, Congress would soon find them. [12] Gridlock in Congress is less a sign of institutional inefficiency than it is a sign of diverse preferences. Removing gridlock via strong presidential action cannot create Pareto-optimal benefits.

Types of Output and the Last-Period Solution

Hayek (1979: 28-29 and passim) and, more recently, Amihai Glazer and Martin Wattenberg (forthcoming 1995) argue that long-term, general-interest legislation is underprovided in the current political system. They argue that term limits would shift the composition of output away from pork towards general-interest public policy.

The argument for underprovision of public policy is as follows. Assume there are two types of legislators, public policy producers, and pork producers. Public policy producers spend most of their time on broad issues related to the national interest, pork producers spend their time producing local public and private goods for their constituents. Voters prefer that Congress spend most of its time on public policy. Electing a single public policy producer, however, will not appreciably change Congress's output of public policy. The rational strategy for voters therefore is to elect a pork producer because the output of public policy will not change but the pork producer will increase the provision of local goods. Rational politicians respond to these incentives by producing pork rather than public policy. The result of all districts acting rationally is socially irrational. This type of argument may explain why voters can simultaneously express disgust at Congress yet reelect their representatives. [13] Given that other districts will also reelect their representatives there is nothing to gain by electing a public policy producer and much to lose.

Term limits, according to Hayek (1979), and Glazer and Wattenberg (1995), will create more policy-orientated politicians and fewer pork producers. Consider the last period of a term-limited representative. In the last period the net benefit to a politician of producing pork is severely reduced because the politician faces no election constraint. With no election constraint the politician is free to engage in general-interest policy work. Recognizing this, Hayek suggests (1979: 113) that a one-term limit of 15 years would be ideal. Moreover, to prevent the representative from engaging in pork production to aid a post-Congress career Hayek would provide every representative with a lifetime sinecure to ensure freedom from all political pull. [14]

According to the Hayek and Glazer and Wattenberg (HGW) theory the problem is that politicians work too hard to get reelected. Term limits reduce the incentive to get reelected and so leave politicians free to pursue higher goals like public policy production. Note that this argument is in direct opposition to the shirking-deviation theory. The shirking-deviation theory argues that the problem is that senior politicians get reelected too easily and therefore they shirk their duties and deviate from the public's preferences.

There is nothing in the HGW theory to suggest that politicians freed of election constraints will not increase shirking or deviation. Hayek (1979) argues that prestige, monitoring, and selection effects will keep shirking within tolerable limits. It is probably for these reasons that the Supreme Court does not have a reputation for shirking even though judges have life tenure and comfortable salaries. There are only 9 justices on the Supreme Court, however, and 535 members of Congress. It is questionable whether the forces Hayek mentions can eliminate serious shirking in an institution of this size. The HGW theory also assumes that if politicians are freed of electoral constraints they will simply choose the right sort of public policy. It is not explained, however, why the judgement of politicians as to the right public policy should cohere with the judgments of voters. The HGW theory sacrifices election constraints, which give politicians an incentive to discover and represent the wishes of their constituents, for a tenative argument that term limits will increase the output of (good) public policy.

Incumbency Advantage Reexamined

We now return to the issue of incumbency advantage. Two questions strike to the heart of the incumbency advantage-term limits debate: Why do voters reelect their representatives if this causes politicians to shirk and deviate from voter preferences? And the related question: Why do voters vote for term limits and simultaneously reelect their representatives? [15]

One answer to the first question is that voters reelect incumbents because they are risk averse and prefer the representative they know to an unknown gamble. A variant of this theory holds that the incumbent targets information to those voters who benefit most from the incumbent's policies. The targeted voters vote for the incumbent and the non-informed voters vote randomly. Since the non-informed voters are a "wash" the incumbent tends to win with high probability. [16]

This theory and its variant suggest that term limits would make voters worse off. In the first case, term limits introduce more unknowns into the voter's calculus, thus increasing risk and making voters worse off. In the second case, the targeted group of voters is made unambiguously worse off and there is no reason to believe that the uninformed voters are made systematically better off. [17]

The most popular answer for the two questions posed above is the free rider-seniority argument (Buchanan and Congleton 1994, Dick and Lott 1993). The argument asserts that there are two opposing forces at work in the voter's decision. Voters reason that they should reelect their representative because senior representatives have more power in Congress and will be able to bring home more pork. But all voters reason in this way, so each region reelects its representative but none gain the hoped-for extra pork because seniority is relative. Even though reelecting incumbents does not increase seniority (and so does not make voters better off) not reelecting dramatically reduces seniority and so makes voters worse off. Voters are trapped into continually reelecting incumbents even though it makes the voters no better off. In fact, continual reelection makes voters worse off because politicians understand that the more senior they are the less likely the public is to vote them out of office. Since there is little threat to their jobs, senior politicians feel free to shirk and deviate from the voter's preferences.

Andrew Dick and John Lott (1993) argue that term limits allow voters to break out of the prisoner's dilemma. By fixing a maximum level of (low) tenure term limits stop voters from entering into a fruitless competition to outbid one another. Each representative in a term-limited system produces as much pork as in a non-limited system but they shirk and deviate less, thus voters are better off.

There are a number of problems with this theory. Dick and Lott (1993) apply the theory primarily to state legislatures where a vote for term limits is a vote to limit all legislators. A vote for term limits at the Federal level, however, is a vote to limit your congressional representatives but not those of other states. The seniority argument in this case implies that voters should vote against term limits. The theory also does not explain why voters voted for presidential term limits in 1951. Since everyone has the same president there is no prisoner's dilemma problem and no reason--if one follows the Dick and Lott theory--to vote for presidential term limits (Tabarrok 1995). (These problems also affect the HGW theory.)

A second problem with the theory is that the shirking-deviation assumption implies that there should be serious last period problems. [18] In the final period the congressperson has no reelection constraint and so is free to shirk and/or deviate. Dick and Lott (1993) assert that this shirking and deviation does not argue against term limits because it would occur anyway since there is always a final period. Every congressperson has only one final period but the relevant fact is that under a term limit system there are more final periods in total. Term limits will increase the number of times a congressional seat turns over in a given period of time, thus there will be more final periods and more shirking and deviation. Furthermore, since by backwards induction term limits reduce the election constraint in all periods, there will be more shirking and deviation in all periods, not just the last period. [19] These arguments suggest that term limits are unlikely to reduce shirking and deviation, a premise the Dick and Lott theory is based upon.

Term Limits or the Rotation of Office?

Up to this point we have examined a broad spectrum of arguments for term limits and for the most part the arguments appear to be weak. Yet 21 states, including large states like California, have passed term limit laws. It is troubling that we have so little rational understanding of this phenomenon.

The main problem with most term limit theories is that they assume that support for term limits represents dissatisfaction with current politicians. Voting against the current congressperson (or president, governor, or mayor) is assumed to be a close substitute for term limits. When this assumption is made it becomes very difficult to answer the two questions posed above, "Why do voters reelect their representatives if this causes politicians to shirk and deviate from voter preferences?", and "Why do voters vote for term limits and simultaneously reelect their politicians?" I now propose a new argument for term limits which does not make this assumption and can easily answer the two questions. I refer to the argument as the conflict theory of term limits.

Imagine that there are two rival coalitions in a region and that each fears the other will gain and hold on to power for an extended period of time. This fear can be motivated by two factors. First, the longer a coalition expects to be in power the more likely it is to exploit another coalition. Or, to put the issue the other way, the greater the expected rotation of power the less likely the presently ruling coalition is to exploit other coalitions. Until May of 1994 South African blacks had little legal political power. Nevertheless, as the prospect of future black political power increased the South African government became constrained in its actions towards South African blacks. The prospect of black power in the future moderated white power today. Second, if coalitions are risk averse they will prefer to rotate power with rival coalitions, and in this way share the spoils of governing rather than gamble upon winning or losing all political power for a lengthy period of time.

Term limits are one method of increasing the rotation of political power. When a politician's term is over the election for the open seat is more competitive than it would be if an incumbent were running. By increasing the number of open-seat elections term limits increase the rotation of power. Over the 1960-90 period, for example, House seats switched party 5.0 percent of the time when an incumbent ran and 25.7 percent of the time in an open election. In the Senate, parties rotated 15.5 percent of the time when an incumbent ran but 42.7 percent of the time when the election was open. [20] The probability of a rotation of power is five times more likely in the House and nearly three times more likely in the Senate in an open election than in an election with an incumbent. Thus, incumbency advantage has an enormous impact on party rotation. [21] Term limits, in fact, were historically referred to as the "rotary system" or the principle of "rotation in office" (Benton 1854, Petracca 1992). From this perspective the benefit of term limits is not the termination or limitation of current politicians but rather the expectation that new politicians will rotate into power.

The conflict theory does not connect support for term limits with dissatisfaction with current politicians and, in the conflict theory, voting against current politicians is not a substitute for term limits. Consider the fact that voters overwhelmingly approved of the presidential term limit amendment yet reelected Roosevelt three times and indicated in polls that they would reelect Eisenhower if they were given the opportunity to do so (Sigel and Butler 1964). This behavior is odd if we think that support for term limits is connected with disapproval of current politicians. It is not odd, however, to reelect Roosevelt and vote for term limits if you support the Democrats but are worried that a future Republican President could be an incumbent as long as Roosevelt was. Similarly, it was not contradictory for voters in the 1992 elections to vote for term limits and simultaneously reelect their representatives.

The conflict theory implies that the more political conflict there is in a region the greater the demand for term limits. An excellent example of this principle is the new Bosnian constitution. Recently the Bosnian Muslims and Croats have joined together to establish a new country under a new constitution. [22] The Bosnian Presidency will be rotated among the Croats and Muslims. The Bosnian civil war has given both groups good reason to fear that one will exploit the other. By agreeing to rotate power both groups are constraining themselves, thereby limiting the possibilities for exploitation.

In the United States gubernatorial and congressional term limits are also more likely to exist the greater the amount of political conflict. In Tabarrok (1995) I measure conflict by state using a "cleavage index" developed by the political scientist Daniel Elazar (1984). [23] I then use statistical regressions to show that the probability a state has gubernatorial term limits is positively related to the amount of political conflict. The probability that a state votes for congressional term limits also increases with the amount of political conflict in a state.

If the political conflict theory is correct it provides strong normative support for term limits. According to the conflict theory all coalitions are better off with term limits than without them. The conflict theory also indicates that term limits need not be made universal in order to be beneficial. If there is a great deal of political conflict in a state it can benefit from term limits even if there is no effect on congressional output. [24]

Conclusion

Most of the traditional arguments for term limits are weak. Term limits are unlikely to create a citizen legislature--historically turnover rates have changed greatly with little or no effect on the composition or character of the legislature. Term limits may increase the amount of public policy which Congress produces or they may increase shirking--we are not sure which. Term limits may reduce the power of special interests but the effect is likely to be small and must be balanced against a possible reduction in public good production. Probably unintended effects of term limits include a shift in the balance of power towards the executive and an increase in the power of the political parties. It is difficult to argue that either effect is unambiguously positive.

A difficulty with almost all of the standard arguments for term limits is that they either do not explain why the public continually reelects politicians or they do not explain why the public reelects politicians and simultaneously votes for term limits. The conflict theory of limits suggests that term limits benefit risk averse and fearful coalitions by increasing the rotation of power. The majority of voters vote for incumbents because the voters believe the incumbents are doing a good job. A voter in today's majority, however, may fear being in tomorrow's minority. It is natural for him to express his satisfaction with current politicians by voting for their reelection and at the same time vote for term limits in order to limit the incumbency of future politicians.

____________________________________
Cato Journal Vol. 14, No. 2 (Fall 1994). Copyright Cato Institute. All rights reserved. The author is Assistant Professor of Economics at Ball State University. He thanks Tyler Cowen, Kevin Grier, Hugh Heclo, and seminar participants at George Mason University for their helpful comments.

References

Aberbach, J.D.; Putnam, R.D.; and Rockman, B.A. (1981) Bureaucrats and Politicians in Western Democracies. Cambridge, Mass.: Harvard University Press.

Aberbach, J.D. (1990) Keeping a Watchful Eye: The Politics of Congressional Oversight. Washington, D.C.: Brookings Institution.

Abramowitz, A.I. (1991) "Incumbency, Campaign Spending, and the Decline of Competition in the U.S. House Elections." Journal of Politics 53(1): 34-56.

Becker, G.S. (1985) "Public Policies, Pressure Groups, and Dead Weight Costs." Journal of Public Economics 28: 329-47.

Becker, G.S. (1990) "Reforming Congress: Why Limiting Terms Won't Work." Business Week, 6 August: 18.

Benton, T.H. (1854) Thirty Years' View. Vol. I. New York: D. Appleton and Company.

Bogue, A.G.; Clubb, J.C.; McKibbin, C.R.; and Traugott, S.A. (1976) "Members of the House of Representatives and the Process of Modernization." Journal of American History 63(2): 275-302.

Buchanan, J.M., and Tullock, G. (1962) The Calculus of Consent. Ann Arbor: University of Michigan Press.

Buchanan, J.M., and Congleton, R. (1994) "The Incumbency Dilemma and Rent Extraction by Legislators." Public Choice 79: 47-60.

Chandler, J.A. (1990) "The United States." In Kingdom J.E. (ed.) The Civil Service in Liberal Democracies, 163-81. London: Routledge.

Coyne, J.K., and Fund, J.H. (1992) Cleaning House: America's Campaign for Term Limits. Washington, D.C.: Regnery Gateway.

Cowen, T.; Glazer, A.; and McMillan, H. (1994) "Rent Seeking Promotes the Provision of Public Goods." Economics and Politics 6(2): 131-46.

Crane, E.H. (1990) "Term Limits for a Citizen Legislature." Vital Speeches, 15 December: 147-48.

Dick, A.R., and Lott Jr., J.R. (1993) "Reconciling Voters' Behaviour With Legislative Term Limits." Journal of Public Economics 50: 1-14.

Glazer, A., and Wattenberg, M.P. (forthcoming, 1995) "Promoting Legislative Work: A Case for Term Limits." In Grofman, N.N. (ed.) Legislative Term Limits: Public Choice Perspectives. Amsterdam: Kluwer-Nijhoff.

Gwartney, J.D., and Wagner, R.E. (1988) "Public Choice and the Conduct of Representative Government." In Gwartney, J.D., and Wagner, R.E. (eds.) Public Choice and Constitutional Economics, 3-28. Greenwich, Conn.: JAI Press.

Hayek, F. A. (1979) Law, Legislation, and Liberty, Vol 3. Chicago: University of Chicago Press.

King, G., and Gelman, A. (1991) "Systemic Consequences of Incumbency Advantage in U.S. House Elections." American Journal of Political Science 35(1): 110-38.

Kreps, D., and Wilson, R. (1982) "Reputation and Imperfect Information." Journal of Economic Theory 27: 253-79.

Kreps, D.; Milgrom, P.; Roberts, J.; and Wilson, R. (1982) "Rational Cooperation in the Finitely Repeated Prisoner's Dilemma." Journal of Economic Theory 27: 245-52.

Kreps, D. (1990) A Course in Microeconomic Theory. Princeton: Princeton University Press.

Lott J.R., Jr. (1987) "The Effect of Nontransferable Property Rights on the Efficiency of Political Markets." Journal of Public Economics 32: 231-46.

Lott J.R., Jr., and Reed, R.W. (1989) "Shirking and Sorting in a Political Market With Finite-Lived Politicians." Public Choice 61: 75-96.

Maisel, L.S. (1990) "The Incumbency Advantage." In Nugent, M.L., and Johannes, J.R. (eds.) Money, Elections and Democracy, 119-42. Boulder, Col.: Westview Press.

Mayhew, D.R. (1974) "Congressional Elections: The Case of the Vanishing Marginals." Polity 6: 295-317.

Mezey, M.L. (1971) Comparative Legislatures. Durham, N.C.: Duke University Press.

Niskanen, W. A. (1971) Bureaucracy and Representative Government. Chicago: Aldine-Atherton.

Ornstein, N.J.; Mann, T.E.; and Malbin, M.J. (1992) Vital Statistics On Congress 1991-92. Washington, D.C.: Congressional Quarterly.

Page, E.C. (1985) Political Authority and Bureaucratic Power: A Comparative Analysis. Knoxville: University of Tennessee Press.

Patterson, K.D., and Magleby, D.B. (1992) "Public Support for Congress." Public Opinion Quarterly 56: 539-51.

Payne, J.L. (1991a) "Limiting Government By Limiting Congressional Terms." The Public Interest 103 (Spring): 106-17.

Payne, J. L. (1991b) "Why Government Spending Grows." Western Political Quarterly 44: 487-508.

Petracca, M.P. (1992) "Rotation in Office: The History of an Idea." In Benjamin, G., and Malbin, M.J. (eds.) Limiting Legislative Terms, 19-52. Washington, D.C.: Congressional Quarterly.

Reed, W.R., and Schansberg, D.E. (1992) "The Behavior of Congressional Tenure Over Time: 1953-1991." Public Choice 73: 182-203.

Reed, W.R., and Schansberg, D.E. (1993) "The Relationship Between Congressional Spending and Tenure." Working Paper. Division of Business and Economics: Indiana University South East.

Reed, W.R., and Schansberg, D.E. (1994) "An Analysis of the Impact of Congressional Term Limits." Economic Inquiry 32: 79-91.

Schlesinger, A.M., Jr. (1973) The Imperial Presidency. Boston, Mass.: Houghton, Mifflin Co.

Sigel, R.S., and Butler, D.J. (1964) "The Public and the No Third Term Tradition: Inquiry into Attitudes Toward Power." Midwest Journal of Political Science 8(1): 39-54.

Stigler, G. J. (1971) "The Theory of Economic Regulation." Bell Journal of Economics (Spring): 137-46.

Tabarrok, A. (1995) "Presidential, Congressional and Gubernatorial Term Limits: Theory and Evidence." Manuscript, Ball State University, Muncie, Ind.

Tullock, G. (1967) "The Welfare Costs of Tarriffs, Monopolies and Theft." Western Economic Journal 5: 224-32.

Weingast, B.R., and Marshall, W.J. (1981) "The Industrial Organization of Congress." Journal of Political Economy 96 (1): 132-63.

Weingast, B.R., and Moran, M.J. (1983) "Bureaucratic Discretion or Congressional Control? Regulatory Policymaking By the Federal Trade Commission." Journal of Political Economy 91(51): 765-800.

Wittman, D. (1989) "Why Democracies Produce Efficient Results." Journal of Political Economy 97(6): 1395-1424.

Zupan, M.A. (1990) "The Last Period Problem in Politics." Public Choice 65: 167-80.

Notes

[1] The states with term limit laws are, Alaska, Arizona, Arkansas, California, Colorado, Florida, Idaho, Massachusetts, Maine, Michigan, Missouri, Montana, Nebraska, Nevada, North Dakota, Ohio, Oregon, South Dakota, Washington, and Wyoming.

[2] Some politicians will want to use the House as a stepping stone to the Senate and the Senate as a stepping stone to the White House but these will necessarily be a small minority since all will know from the beginning that the chances of successfully doing this are small.

[3] Payne's argument is similar to the capture theory of regulation proposed by George Stigler (1971).

[4] According to Payne (1991a,b) senior representatives are "captured" by special interests and therefore vote to increase spending. Gary Becker (1990) argues that it is the junior representatives who are captured and that only the senior representatives have the experience necessary to ignore the special interests. Becker's argument cannot explain why spending increases with seniority nor does it seem plausible psychologically.

[5] It is not necessary that each representative and each interest group have a long-term relationship with one another in order to build up trust. A reputation for keeping promises of support with some groups will carry over to other groups. This still requires that the representative and interest each have a history which can be examined.

[6] See Kreps and Wilson (1982); Kreps, Milgrom, Roberts, and Wilson (1982).

[7] Note that one's evaluation of these effects depends on larger assumptions which we have not made. If the Coase theorem can be applied to the political process as in the models of Gary Becker (1983) and Donald Wittman (1989) then higher bargaining costs can only reduce welfare. If the Coase theorem does not apply, and politics is characterized by inefficient rent seeking and "majoritarian pathologies" as Tullock (1967), Buchanan and Tullock (1962), and Gwartney and Wagner (1988) point out then higher bargaining costs may improve welfare.

[8] By executive I mean the president, his advisers and his appointees. By bureaucracy I mean the relatively permanent civil service.

[9] One interesting proposal to overcome this problem is to have term limits for everyone including bureaucrats. Murray Rothbard (1991) has pointed out that this was effectively the case prior to the civil service reforms beginning with the Pendleton Act in 1883.

[10] Edward Page (1985), and Michael Mezey (1979) compare European and American bureaucracies and legislatures.

[11] See for example Arthur Schlesinger (1973) for one view.

[12] Moreover, we can be sure that the free exercise of presidential power would have considerable costs as well as potential benefits (Buchanan and Tullock 1962).

[13] Voters, however, do not seem to begrudge their support of incumbents. See the polls discussed in Kelly Patterson and David Magleby (1992).

[14] Even if we do not create a single long term, term limits can increase public policy work. Non-limited politicians always benefit more from elections than term-limited politicians. This is clear in the last period but consider the second-to-last period. In the second-to-last period the benefit to a non-limited politician of election is the opportunity to serve one additional period plus the option to run again. For a term-limited politician the benefit of being elected is only one additional period. Since the benefit of election is always smaller for the term-limited politician he has less incentive to produce pork than the non-limited politician.

[15] Voters who voted for term limits often reelected their representatives in the 1992 elections. This type of behavior was also present in the debate over presidential term limits. See Tabarrok (1995) for more discussion.

[16] From the binomial distribution note that if there are n uninformed voters and they vote randomly (p=.5 for the incumbent, p=.5 for the challenger) the incumbent can expect 1/2 of their votes with variance, p (1-p)/n = .25/n. As n increases the variance of the proportion of voters voting for the challenger decreases so any committed group exerts a larger effect on the final outcome.

[17] Lott's (1987) argument for term limits seems to suffer from this problem. He shows that there are barriers to entry in the political marketplace because of brand name effects. He asserts that this can explain support for term limits but he does not show how less information provision can benefit voters.

[18] On the last period problem see Mark Zupan (1990), and John Lott and Robert Reed (1989).

[19] See the backwards induction argument in footnote 14.

[20] These figures were calculated from figures in Tables 2-5 to 2-8 in Ornstein et al. (1992).

[21] Note that my analysis focuses on party rotation not on which party (if any) will gain more seats from term limits. See Reed and Schansberg (1994) for such an analysis. Also, coalitions are not necessarily identical with parties. Conflictual coalitions exist within parties as well as across parties. Thus, coalitions can rotate when parties do not and parties can rotate when coalitions do not.

[22] See the front page of the Washington Post, 2 March 1994, "Muslim and Croats to Link Territories."

[23] Elazar's index assigns a ranking of 0 (not important) to 3 (highly important) to seven categories of political cleavage or conflict; these are political culture, general culture, sectionalism, localism, urban-rural, metropolitan/non-metropolitan, and inter-metropolitan. The total score across these categories is then added to the number of categories with non-zero scores (see Elazar 1984 for more detail).

[24] States that do not pass limits may benefit at the expense of those which do because of relative seniority in Washington. This is not an argument, however, for passing term limits on all states because in the conflict theory some states may be worse off with term limits. It may be best, therefore, to ignore the externality and let each state choose as it will with full knowledge of the costs and benefits.

The Cato Journal is published in the spring/summer, fall, and winter by the Cato Institute, 1000 Massachusetts Ave., NW, Washington, D.C. 20001-5403. The views expressed by the authors of the articles are their own and are not attributable to the editor, editorial board, or the Cato Institute. Printed copies of the Cato Journal may be ordered by calling 1-800-767-1241. Back issues are also available on the Cato Institute Web site: http://www.cato.org. Email comments or suggestions to subscriptions@cato.org.