Term Limits: Causes and Consequences

Author(s): Edward J. López

Source: Public Choice , Jan., 2003, Vol. 114, No. 1/2 (Jan., 2003), pp. 1-56
Published by: Springer

Stable URL: https://www.jstor.org/stable/30025937

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Springer is collaborating with JSTOR to digitize, preserve and extend access to Public Choice

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 Public Choice 114: 1-56, 2003.
@ 2003 Kluwer Academic Publishers. Printed in the Netherlands.

Term limits: Causes and consequences *

EDWARD J. LOPEZ
Department of Economics, University of North Texas, P.O. Box 311457, Denton, TX
76203-1457, U.S.A.; e-mail: [email protected]

Accepted 1 June 2001

Abstract. This paper consults multiple literatures to specify and evaluate the economic
tionales for term limitation, particularly on Congress. I first consider theories that arose
explain, among related issues, why individual states might unilaterally self-impose term lim
on their own delegations to Congress. Next I consider two main lines of argument for univ
limits, both of which begin with the empirical phenomenon of high and rising congressi
tenure. First, supporters of term limits argue that higher tenure biases legislatures towa
inefficiency big government (high spending). Second, higher tenure creates inefficient (a
competitive) conditions in the legislative election market. Term limitation would remedy t
inefficiencies by virtue of decreasing average tenure. These claims are then evaluated in li
of the evidence amassed in the literature. Based on the literature reviewed, this paper finds th
while term limits will reduce average tenure, there is no evidence to suggest that term lim
will affect the underlying causes of these inefficiencies. Further research on a more gen
reform, which would strike deeper at these underlying causes, is implied.

1. Introduction

Research on term limits grew significantly in the 1990s and left a variet
of unresolved questions in its wake. This article brings together the variou
contributions of the term limits literature, summarizes what the body of work
has to say, and highlights areas of ongoing debate and remaining problem
Understanding term limitation in this way requires understanding its context,
which takes us into the related literatures of congressional tenure, government
growth, campaign finance, welfare economics, and others. Thus, the purpo
of this paper is to integrate the work on term limits into the economics liter-
ature at large, and thereby answer otherwise unresolved questions. Doing s
makes clear the lessons that term limits research has to offer, which until now

* Thanks to the referee at Public Choice, Bruce Bender, Lawrence W. Kenny, W. Robe
Reed, Noel D. Campbell, Alex Tabarrok, Bernard Grofman, William A. Niskanen, Gord
Tullock, James M. Buchanan, Tyler Cowen, and Randall G. Holcombe for their helpful com
ments. Thanks also to Hsiu-Ju Yang for her valuable research assistance. Thanks to the At
Economic Research Foundation and the Earhart Foundation for financial support. Spec
thanks to Robert D. Tollison. I am exclusively responsible for oversights and/or errors.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
2

have been somewhat scattered and ambiguous. This is an extensive exercise.

To make it tractable I have organized the survey around an imaginary hub,
with causes for term limits occupying one spoke and consequences the other.
By "causes" I refer to reasons why term limits may be a desirable institu-
tion. There are many reasons a polity, or private groups within a polity, may
wish to limit terms of office. In the interest of brevity and parsimony of ana-
lysis, I ignore non-economic arguments for term limitation. For example, the
journalist George Will (1992) has argued term limits will return Congress to
the founding fathers' vision of representation. Similarly, Washington public
policy groups have argued that term limits will restore a citizen legislature, or
that term limits will increase racial and gender diversity in office.' One could
be persuaded that these are largely rhetorical instruments for the true under-
lying economic reasons for supporting term limitation. The present focus is
on expected social and private gains as causes for supporting term limits. I
therefore stick to the economic arguments.
By "consequences" I mean the positive effects of term limitation on polit-
ical, economic, and public policy variables. I will pay special attention to the
consequences vis-a-vis the arguments advanced in support of term limitation.
This enables me to make a positive economic evaluation of term limits later
in the paper.
I have two caveats before proceeding. First, I generally restrict the survey
to the economics and political science literatures. There is a legal literature on
term limits, discussing original intent (Price, 1996), the qualifications clause
of the Constitution (Zubler, 1995; Thomas, 1995), states' rights, and other
constitutionality issues. I am not specialized in this area so as to discuss
these important questions. Moreover, they fall slightly afield from this forum.
Second, I also emphasize term limits for legislatures, particularly Congress,
over other offices.2 Congress is the only form of elected office in American
government not regulated by term limitation. As a result, there are open ques-
tions on congressional term limits, particularly regarding the consequences.
And there is, therefore, considerably more literature on legislative and con-
gressional term limits than executive or judicial. There are important lessons,
however, to draw from the research conducted on these other offices, particu-
larly gubernatorial limits. So the emphasis on term limits for legislators will
not be to the complete exclusion of term limits for executives and judges.
The plan of the paper is as follows. In the remainder of Section 1, I briefly
discuss the interest group theory of government (because most term limits
studies are either implicitly or expressly in this context) and the chronology
of term limits. These are not intended to be thorough treatments but merely
to "set the stage" for the thorough discussion of the term limits literature that
follows. Next, in Section 2, I review the literature on why a polity or group

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 3

- a state, in particular - would desire term limits. This area of the literature
is closely tied to relative tenure among the states, whereas the support for
universal term limits draws more closely on high and rising average tenure.
In Section 3, I discuss the history of congressional tenure, specifically the
transition from low and stable tenure to high and increasing tenure. In Sec-
tion 4, I use the causes of this transition to present the two leading arguments
for term limits. In Section 5, term limits become an independent variable,
and I discuss their consequences on political and economic variables, paying
particular attention to the "causes" laid out thus far in the paper. Section 6
presents my evaluation of the arguments for term limits based on the literature
reviewed, and Section 7 briefly summarizes the results obtained in the paper.

1.1. The market for wealth transfers and economic vs. "political" efficiency

The literature on term limits has developed within the context of the interest
group theory of government.3 This is foremost a positive theory: the gov-
ernment's primary function is (as distinct from should be) to redistribute
wealth among private groups in the polity. Collective action costs increase
with group size and heterogeneity (Olson, 1965). Legislatures broker wealth
transfers from politically ineffective groups, those with high Olson-costs, to
politically effective groups, those with low Olson-costs. The particular mode
of legislation is not essential to achieving the transfer,4 but the one well-
known caveat is that the wealth transfer typically will not take the form of
a cash payment (Stigler, 1971: 4).
Legislators maximize their net political support rather than social wel-
fare. I use the term "politically efficient" throughout this paper to mean the
lowest-cost attainment of a given quantity of transfers. The election mar-
ket, to illustrate, ought to ensure that the lowest-cost candidate gets elected
to provide the transfers desired by constituents. But imperfections in the
election market may interfere, and higher cost candidates may acquire and
maintain office (which might manifest in, for example, "shirking"). Note
that "political" efficiency, as I am currently using the term, is distinct from
economic (e.g., Pareto) efficiency. The quantity of wealth transfers that max-
imizes net political support is, in a political sense, the efficient quantity. As in
economic markets where competition is equilibrating, political competition
among interest groups pushes the legislator to match marginal demanders
with marginal suppliers of wealth transfers, and the politically efficient quant-
ity of transfers is reached. In the economic (i.e., wealth-maximizing) sense
of efficiency, wealth transfers through legislation are inefficient due to both
deadweight and rent seeking costs (Tullock, 1967). The market for wealth
transfers tends toward political efficiency but away from economic efficiency.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
4

There is a premium on seniority in the market for wealth transfers, both

to the legislator and the individual district. The more productive a legislator
at making the marginal adjustments to political efficiency, the greater will be
the probability of reelection and the returns to holding the office. Productivity
in brokering wealth transfers is a function of parliamentary rights such as
committee seats, agenda power, and leadership positions, all of which are
positively correlated with relative tenure, or seniority. Each legislator has the
incentive, therefore, to accumulate tenure vis-a-vis other legislators. And their
electoral support groups, toward which wealth transfers are directed, share in
this incentive.
This is where term limitation becomes relevant: social welfare would be

improved were the incentives to transfer mitigated or outright eliminated.

That is, term limitation may be a vehicle for recovering some or most of th
economic costs associated with wealth transfers. Second, incumbents collect-
ively control many of the institutions determining the competition they wil
receive from challengers to their seats. If incumbents have erected barriers
to entry such that the congressional election market is anti-competitive, term
limitation may enhance political efficiency. Each argument presents a case
for term limits based on excessive tenure. It is along these two possibilities,
and within the context of the interest group model of government, that I will
discuss the causes and consequences of term limits.

1.2. Historical sketch of term limits

It is necessary, before proceeding, to establish basic facts about term limits

in history. The idea of term limitation, like so many democratic institutions,
derives from antiquity. Aristotle endorses rotation of public office in The
Politics:

Such being our foundation and such the principle from which we start,
the characteristics of democracy are as follows: the election of officers by
all out of all; and that all should rule over each, and each in his turn over
all; that the appointment to all offices, or to all but those which require
experience and skill, should be made by lot; that no property qualification
should be required for offices, or only a very low one; that a man should
not hold the same office twice, or not often, or in the case of few except
military offices; that the tenure of all offices, or of as many as possible,
should be brief...

Aristotle, The Politics, Book VI, Section II

In addition, the Greeks practices term limits in many forms. Oakley (1994:
14 ff.) shows that term limits were a feature in most of the Greek constitu-

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 5

tional regimes. Particularly in the democratic regimes, the term length and
maximum tenure in many public offices were specifically enumerated. Some
branches of the government, such as the 500-member Boule ("Council") were
subject to stringent limits of only one term.5
Term limits disappeared in practice until the American founding era, but
resurfaced in thought with the renaissance of republican theory, roughly in
16th century Italy and England. Continuing with the Levelers, Physiocrats,
and Enlightenment thinkers, this heritage was brought to the American found-
ing, where de jure term limits again came into practice under the Articles of
Confederation:6

No State shall be represented in Congress by less than two, nor more

than seven members; and no person shall be capable of being a delegate
for more than three years in any term of six years ...

Articles of Confederation, Article V, Paragraph 2

Term limits were nearly included in the U.S. Constitution as well, but their ex-
clusion is not well understood. James Madison strongly favored de jure term
limits, and included them in his Virginia Plan. Similarly, George Mason wrote
term limits into the Virginia Declaration of Rights, which later informed the
Bill of Rights. Despite strong intellectual support from such respected figures
as Madison and Mason, and an apparently firm agrarian coalition of support
that included Washington himself, mandatory rotation of office was dropped
from the Constitutional debate with little episode. This is not to say the
framers deemed term limits an unworthy idea; some of them did, and others
did not.7 But when ideological, legal, and economic constraints were brought
to bear on the array of ideas proposed before the convention, term limits
did not survive. A systematic, rational choice explanation of exactly how
this played out eludes the literature.8 For now, we rely on the conventional
interpretation that the farmers considered term limits an unnecessary detail
due to their expectations that tenure problems would take care of themselves
through naturally (voluntary) rapid turnover of representation.
Term limits almost entirely disappeared until the 1990s, when a "term
limits movement" swept change throughout the political institutions of Amer-
ican democracy. Table 1 provides chronological detail of these institutional
changes. Beginning with the 1990 general elections, 23 states voted to im-
pose term limits on their delegations to the U.S. Congress. These laws were
contested, and ultimately deemed unconstitutional under a 5-4 U.S. Supreme
Court ruling (U.S. Term Limits, Inc. v. Thorton, 115 S.Ct., 1995). In 1992,
term limits were a pillar in the platform of presidential candidate H. Ross
Perot. In 1996, term limits were an issue in the Republican presidential

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
6

primaries. And in that year's election cycle, initiatives in 10 out of 14 states

passed that enacted so-called "Informed Voter Laws" instructing their state
legislatures to vote for an Article V constitutional convention with the explicit
intention of imposing congressional term limits.
Term limits were an important issue within Congress as well. In 1994,
virtually all of the 73 freshmen elected to the House of Representatives made
promises to vote for term limits. During the 104th Congress that followed,
term limits were part of the Republican "Contract with America". Just weeks
before the Supreme Court's Thorton ruling, the House brought term limits to
floor votes. Later that year the Senate took a procedural vote on term limits,
marking the first time in forty years that term limits came to the floor in
both chambers. None of the measures in either chamber gained the majorities
required to pass. Republicans, now in the majority, again made term limits a
campaign issue in the 1996 election cycle, and the bill was again introduced
at the start of the 105th Congress.

2. Theories of term limits: Why a polity or group would want them

A set of inter-related puzzles or paradoxes arises out of the experience of

term limits in the 1990s, particularly at the state level. Why, for example,
did roughly half the states pass congressional and legislative term limits but
the others did not? Or why do voters consistently and overwhelmingly reelect
their incumbents and support term limits?9 More generally, what is the voter's
optimal tenure for his representative? A substantial body of work, mostly
theoretical, was developed to solve these puzzles and more. In this section, I
now place term limits on the left-hand side of the equation to consider how
this body of work might constitute the "causes" for a state desiring unilateral
term limitation.

2.1. Term limits as solution to free rider problem

Dick and Lott (1993) construct a model in order to address the apparent
paradox I alluded to just now: why do voters continually re-elect their own
members of Congress while strongly favoring term limit initiatives? The
answer, they claim, is to conceive of voting for legislators as troubled with
free-ridership, and term limits as providing the solution to the free-rider
problem.
The theory advanced by Dick and Lott (henceforth DL) is an application
of the interest group theory of government. Because legislators compete for
net transfers vis-t-vis one another, tenure is beneficial to a district both in ab-
solute terms and in relative terms. However, tenure also enables the legislator

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 7

Table 1. Chronology of action on congressional term limits.

Date Location Action

11/90 CO Voters approve initiative imposing term limits on con-

gressional delegation.
11/91 WA Voters reject initiative imposing term limits on congres-
sional delegation.
3/92 ID State legislature passes statute recommending Congress
propose Article V amendment to limit congressional
terms.

11/92 1 Voters in 14 states approve initiatives imposing term

limits on congressional delegations.
2/94 WA Federal District Court rules state's term limits law
unconstitutional.

3/94 AR State Supreme Court rules state's term limits law un-
constitutional.

5/94 UT State legislature passes and governor signs statute im-

posing term limits on congressional delegation, making
Utah first state to do so.

5/94 NE State Supreme Court rules state's term limits law un-
constitutional - signature requirements.
10/94 OK Voters again pass term limits on congressional delega-
tion.

11/94 2 Voters in seven states approve initiatives imposing term

limits on congressional delegations.
11/94 U.S. Republicans win control of Congress. Term limits part
of "Contract with America".

1/95 U.S. House and Senate hold hearings specifically on term

limits.

3/95 U.S. House conducts first-ever floor debate on term limits.

Rejects constitutional amendment.
4/95 NH State becomes 23 to imposed term limits on con-
gressional delegation, only second to do so via st
legislative statute.
5/95 U.S. United States Supreme Court rules Arkansas term limits
law unconstitutional, effectively nullifying legality of
all 23 state laws.

10/95 U.S. Senate conducts non-binding "Sense of Senate" vote,

rejects term limits 45-49.
4/96 U.S. Senate conducts floor debate on constitutional amend-
ment. Motion fails cloture vote 58-42, two sho
requisite 60 to close debate.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
8

Table 1. Continued.

Date Location Action

11/96 3 Voters in nine states pass "Informed Voter Laws".

11/96 U.S. Republicans retain control of Congress. Promise term
limits vote.

1/95 U.S. House conducts hearings on term limits.

2/97 U.S. House conducts floor debate, rejects constitutional
amendment to impose term limits.

1AZ, AR, CA, FL, MI, MO, MT, NE, ND, OH, OR, SD, WA, and WY.
2AK, CO, ID, ME, MA, NE, NV.
3AK, AR, CO, ID, MO, ME, NE, NV, and SD.

to accumulate brand name capital, which insulates him from voter

ences and marginally increases the scope for inefficient representati
shirking). So tenure is a double-edged sword that creates a free rider pro
In this manner, DL reflect the model of executive term limits in Ad
Kenny (1986). High turnover can be costly to the electorate by subst
inexperienced politicians for experienced ones, and by exacerbating in
problems associated with last term effects. However, high tenure can
weaken the electoral sorting mechanism by worsening Downsian info
deficiencies; an electorally comfortable politician creates transfers to
and to influential coalitions, reducing net transfers to the electorate
creasing welfare costs of redistribution. Adams and Kenny (1986: 304
to see term limits emerge when the latter effect dominates:

We propose that the existence of tenure limits reflects the imp

of deadweight costs (largely because of redistribution), which ri
tenure, relative to turnover costs, the latter falling with tenure.

But the resemblance of DL to this argument is not complete: in

with vying electoral constituencies - a legislative polity - it is relat
ure that matters to the district's acquisition of net transfers. As a re
tenure/turnover tradeoff creates a free rider problem among the pertin
tricts. An electorate that removes an inefficient legislator, while other d
continue to re-elect their own, forfeits relative tenure and reduce
transfers. Meanwhile, districts that retain their incumbents gain te
net transfers relative to districts that remove their incumbents. Knowin
all districts rationally free-ride and the equilibrium is the high-tenu
quo. Legislative term limits enter the analysis as a solution to this fr
problem:

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 9

If all voters could agree simultaneously to limit the tenure of incumbents,

... [this would] ... lower the cost incurred by any district that removes
its incumbent, because term limits place an upper bound on the decline
in relative political experience that a district suffers when replacing its
incumbent with a challenger (Dick and Lott, 1993: 4).

Hence, while each district is a free-rider, the universal imposition of term

limits can be seen as a solution to this the free-rider problem because when
all districts vote for term limits at once, none has the incentive to free-ride. In
general equilibrium, this explains why voters like both incumbents and term
limits.

However, this does not mean that each district is in a partial equilibrium at
the optimal balance between tenure and turnover (Adams and Kenny, 1986).
The DL free rider theory predicts that term limits will be adopted universally
- in all states/districts at once. But as discussed in Section 1.2 above, that
did not happen. On the contrary, 23 states unilaterally self-imposed term
limits, and 27 did not. It seems the literature has had difficulty explaining this
anomaly partly due to the dominance of the DL theory, which understates
the motivations behind voters' desires for term limits. Yes, voters want term
limits because their members become insulated over time as accumulated ten-

ure erects barriers to entry that become sufficient to allow representatives to

shirk constituent interests. But this is only half the story. A state/district's net
transfers are affected also by members from other states, and a state's voters
might want term limits because of the wealth transfers going elsewhere. The
DL paradox could perhaps be understood as the fact that voters appreciate
their own members because of their ability to bring home wealth transfers, but
they also dislike members from other states because they take wealth transfers
elsewhere. In experience, the passage of term limits was not a solution to the
free-rider problem as much as an indication that free-riding existed and 27
states were doing it. The free rider argument, therefore, explains why voters
in 27 states do not want term limits on their representatives, but not why
voters in 23 states do.
In another paper (L6pez, 2002) I attempt to sort out these issues using
regression analysis of the congressional votes on term limits. The House
brought a term limits constitutional amendment to the floor in the 104th Con-
gress (1995-96) and again in the 105th (1997-98). The amendment gained
simple majorities in both votes, but not the two-thirds supermajorities re-
quired to pass. After the 104th vote but before the 105th vote, the Supreme
Court struck down the 23 state term limits law. I examine voting behavior
before and after the Court's ruling, and demonstrate how it changed for
different types of legislators. In the 104th vote, being from a term limited
state added between 12 and 25 percentage points to the probability of voting

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
10

for universal term limits, ceteris paribus. But in the 105th vote, this ceteris
paribus difference in voting disappears, adding at most 0.7 probability points.
Moreover, using multivariate maximum likelihood, I demonstrate that the
probability (both joint and conditional) of voting 'yea' and then switching
to 'nay' was higher for members from unilaterally term-limited states. What
do these results suggest? Prior to the Court's ruling, members from states
that had unilaterally imposed term limits were at a relative disadvantage,
and they wanted to impose term limits on all other members as well. In
contrast, representatives from states without term limits were at a relative
advantage, and strongly opposed the amendment. It is not a free-rider result
per se, but certainly a similar calculus of strategic advantage produced it.
Moreover, the results are maintained whether the returning member or the
electing constituency is treated as the unit of choice. This would seem to
indicate a clear awareness, on the part of voters and representatives, of other
districts in deciding whether to support term limits.

2.2. Why unilateral term limits?

The results from L6pez (2002) are akin to a states' voters placing their del-
egation into the low payoff of an off-diagonal cell in the simple prisoners
dilemma game. So it is still a mystery why a state/district would unilater-
ally impose term limits (or why 23 states did so while 27 did not). More
broadly, how can the DL free-rider theory argue that states will coordinate
their behavior in a way that belies the underlying model it employs?
Tabarrok (1994, 1994) takes up this question and is also concerned with
the DL paradox, focusing on the effect that term limits will have on the dis-
tribution of power among "conflictual coalitions" (such as political parties
or interest groups). Each coalition in a heterogeneous political jurisdiction
is risk averse to being exploited while another coalition is in power. Term
limits are desirable not because they will reduce tenure, but because they will
increase the number of open seat elections, and therefore accelerate the rota-
tion of power between these coalitions. Tabarrok cites statistics from House
elections between 1960 and 1990 in which open seat elections result in a
change of party between three and five times more frequently than incumbent-
challenger races (Tabarrok, 1996: 237). Therefore, term limits should result
in more frequent changes of coalition control, from which risk averse co-
alitions benefit. They prefer imposing term limits on themselves as a means
of ensuring that other coalitions would not remain in power long enough to
exploit them excessively. This model can be applied to predict unilateral term
limits by individual states. The more extreme the differences between groups
in the same political jurisdiction, the greater the benefits they derive from
rotating their opponents out of office routinely, and thus the more willing

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 11

they are to rotate themselves. Hence, the "conflict theory" predicts that states
with greater conflict will be more likely to pass term limits. Empirical work
on this prediction is limited to a logit model across states, but since there is
near perfect correlation between states that have term limits and states that
have initiatives and/or referenda, meaningful parameter estimates were not
produced.
By similar reasoning, Glaeser (1997) models voters as heterogeneous in
ideological views as well as risk averse. In districts where coalitions differ
greater (ideologically), each prefers even odds of being in power for con-
secutive periods to even odds of being in power permanently. Though rather
stylized, the model generates a welfare claim that term limits are more likely
to emerge from more risk averse districts so long as the costs of losing relative
tenure are not too great. The degree of heterogeneity also matters (though it is
not clear from the model whether more heterogeneous districts are necessarily
more risk averse). The model also predicts ideologues will support term limits
and Downsians will not. So states that take higher net transfers would, ceteris
paribus, not self-impose.
Konrad and Torsvik (1997) dynamically model bureaucratic incentives,
and the effect that political institutions have on efficient oversight by legis-
lators. The authors assume a model with asymmetric information between
bureaucrats and legislators (cf. Weingast, 1984; Weingast and Moran, 1983).
Bureaucrats generally have longer tenure and, relatedly, more specialized ex-
perience and knowledge of the cost/production functions underlying the tasks
assigned to the bureau. In contrast to conventional agency solutions such
as an efficiency wage, however, it is possible to overcome the information
asymmetry by letting legislators have longer tenure. Legislators would then
accumulate knowledge of the bureau's functions and can work to alleviate
agency inefficiencies. On the contrary, Konrad and Torsvik apply the "ratchet
effect" reasoning of the agency literature, which shows that agents bound by
long-term contracts will conceal information about their true productivity in
early periods to avoid the principal using (i.e., "ratcheting") that information
later on. What the principal needs is a time-consistent way to commit to a
short-term contract (or at least the strategies thereof). Term limits provide
such a commitment. A complication arises because the principal in their
model is himself an agent in the broader interest group model of the legislator.
But the authors show that the commitment problem reverts back to voters
under certain plausible assumptions.

2.3. Dynamic prisoners dilemma

It is possible that none of the above theories is necessary to explaining unilat-

eral action to impose term limits. It is a well-known property of the PD model

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
12

that when played in supergames the dynamic equilibrium may reach the vol-
untary cooperative, globally optimal combination. Ferejohn (1986) develops
this results for political markets. The question then becomes whether voters
in the states that passed term limits perceived an effectively infinite horizon
to the game. One would also have to ask why voters in other states did not
perceive, or had not yet perceived, the same thing. These are questions that
the literature on term limits has not yet taken up. But clues to their answers
can be gleaned from the facts as they unfolded. First, direct democracy played
a key facilitating role in the passage of term limits: of the 23 states that passed
term limits, 22 did so via initiative; every initiative state passed term limits;
and none of the 27 states without term limits has an initiative. Second, the Su-
preme Court in the 1995 U.S. Term Limits v. Thorton rules that state-imposed
term limits on Congress were unconstitutional, halting any existing dynamic
process toward more states adopting them. Certainly the explanation that only
those states with direct democracy pass term limits is an unsatisfying explan-
ation. One possible alternative explanation is that direct democracy facilitates
institutional change more quickly, and states that would have imposed term
limits and had to do so through representative democracy (i.e., legislative state
constitutional amendment) did not have the opportunity to do so because of
Thorton (Tabarrok, 1996). This is a theoretical hunch, but it has some support
in that New Hampshire is both the only state to pass term limits via legislative
statute and the last to do so - just one month before Thorton. The argument
implies that more states would have passed term limits after New Hampshire,
but were disallowed by the Court.
Even so, this rationale only diverts attention away from the correlation
between direct democracy and self-imposed term limits. The underlying
question of the political-economic factors why a state - direct democracy
or not - would self-impose term limits still remains. One possibility is the
presence of Stackelberg-type leadership among the first movers in the game.
States unilaterally acted knowing the asymmetric losses they would incur,
but they did so with the expectation that other states would follow. Why the
others would follow is disputable. Certainly a state cannot penalize another
for not adopting term limits, so there would appear to have been no "stick"
with which to prod the other states to adopt. On the other hand, there is the
"carrot" of global efficiency (general equilibrium) gains to mutual coopera-
tion. But then there is the still bigger carrot of local (partial equilibrium) gains
to free riding. Even in the supergame construction of the prisoners dilemma,
the cooperative equilibrium is sensitive to large numbers of players, costs of
volunteering, and the magnitude of efficiency benefits of mutual cooperation.
One would have to argue that states looked away from all these hindrances
and still made the first move to self-impose term limits. It seems to stretch the

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 13

limits of plausibility. Still, it would benefit the argument greatly if one could
empirically show that the states that passed term limits conditioned their laws
on whether voters in other states also passed. Indeed, five of the 23 states that
passed term limits included "trigger clauses", by which their laws would only
take effect after one-half of the other states also self-imposed term limits. On
the other hand, 18 of those 23 states did not include such trigger clauses.
It would seem, therefore, that the dynamic prisoners dilemma applied to
term limits suggests numerous testable propositions. But we cannot be sure.
To my knowledge, no study either theoretically or experimentally shows the
supergame notion advanced here to hold. While we have some decent clues
and plausible theoretical hunches why states might self-impose, we are not
yet sure.

2.4. Term limits as redistribution of political power

Friedman and Witman (1996) view term limits as a way to redistribute

political power among branches of government and across districts. They
write, "term limits enable voters in one legislative district to influence the
outcome in other legislative districts" (p. 229). The formal theory models
the production of policy outcomes as a function of the balance of powers
and "alignment" of constituent interests with representatives' performance.
They first assume political power increases monotonically with absolute and
relative tenure. They next assume constituents compare executive to legislat-
ive performance endogenously (a formal property that lets them aggregate
evaluations across districts.) Third, they assume constituents vote rationally
(to maximize alignment between interests and performance). Finally, they
assume that term limits will diminish the relative power of the branch on
which they are imposed. Three important results are proposed. First, voters
favor incumbents over challengers, ceteris paribus. Second, voters wish to
impose term limits on the branch of government that is least aligned with
their interests. Third, voters in a district can rationally wish to reelect their
own representative, but still want legislative term limits if: 1) the executive
branch is more aligned with their interests; and 2) the seniority of their own
representative is relatively low. (The weak form of the proposition still holds
even if one of the conditions does not, so long as the other condition is suffi-
ciently pronounced.) Hence, in the Friedman and Wittman theory, term limits
are a means to transfer political power: 1) from districts with more senior
representation to districts with less; 2) to districts whose interests are more
aligned with the executive (congress) in the case of congressional (executive)
term limits; and 3) from districts whose representation is in control of the
congressional (executive) branch.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
14

The empirical test of their theory was conducted using the vote on Califor-
nia Proposition 140 in 1990 that imposed term limits on the state legislature
(Friedman and Wittman, 1995). They find that Republican districts voted
more in favor of term limits, as did districts whose representatives were less
tenured. Also, districts whose legislators had been deemed more effective in
constituent surveys were less prone to vote in favor. Examining the Congres-
sional vote on the 22nd Amendment limiting presidential terms, the authors
also find those representatives/senators who were weak in executive relative
to congressional political power - Republicans and Southern Democrats -
voted consistently in favor of the amendment. Those who were strong in ex-
ecutive relative to congressional political power - highly populous states and
non-Southern Democrats - voted consistently against. Friedman and Wittman
(1995: 80) conclude:

When Democrats were in control of the Presidency and Republicans were

in control of Congress, Republicans were in favor of term limits on the
Presidency; and when Republicans were in control of the Presidency and
Democrats in control of Congress, Republicans become more interested
in imposing term limits on Congress and less so on the Presidency.

On casual observation republicans were more supportive of term limits prior

to the 1994 elections, but lost considerable interest in the idea once in the
majority. (I will clarify this observation in the section on party balance be-
low.) Moreover, in congressional voting on term limits L6pez (2002) finds
statistically significant opposition among representatives who are females,
democrats, more tenured, and from districts with greater net transfers, ceteris
paribus.
To summarize, unilateral term limits appears to be a strategic decision.
This meets squarely with the intuition of the matter, and there is ample evid-
ence to support this claim. Concluding anything further, however, is difficult
because the literature has not decidedly shown why a state would unilaterally
self-impose. Relative tenure seems to be an important variable, particularly if
one perceives term limits as a way to redistribute political power. But party,
voter heterogeneity, risk preference, and the balancing of tenure costs against
turnover costs also matter. In contrast, the causes for universal term limits
are comparatively easy to identify. They involve high and increasing average
tenure, to which I now turn.

3. Congressional tenure

In Section 1.2 I argued that the only explanation that exists for excluding
term limits from the Constitution is that the framers expected congressional

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 15

turnover to remain naturally high. The increase in professionalism in the 20th

century, particularly as evidenced by high and increasing tenure, renewed
interest in term limits. In this section, I discuss the history of congressional
turnover and tenure because it is necessary to evaluate the effectiveness of
term limits vis-a-vis the stated causes for desiring term limitation.

3.1. 19th century congressional tenure

For more than a century, the framers' expectation of high congressional

turnover proved true. Figure la illustrates this with the simple time series
of turnover (percent of incumbents replaced) for the history of Congress.
Until 1890, average turnover was close to 46%. Since then, it is barely 20%.
Until the 1880s, mean tenure was close to two terms, but by the 1900s it
had increased to 3.5 terms (Gilmour and Rothstein, 1996). This long-term
adherence to the framers' vision was followed by a relatively abrupt reversal
near the end of the 19th century. As can be seen with the six-year moving
average in Figure lA, turnover experiences a period of steep decline in the 25
or so years beginning about 1870. As Kernell (1977: 671) observed, "From
the Civil War through the 1920s ... there appears to have been near linear
growth of congressional careerism ..." This can be seen in Figure IB. From
1865 to 1921, turnover declined by an average of 1.01% every year. Between
1883 and 1910, turnover declines by 2.1% annually, as shown in Figure IC.
These visual inspections merely illustrate what historians of Congress have
recognized for generations: that the last few decades of the 19th century
ushered in the professionalism of Congress.
But why? To specify and evaluate the arguments for term limits, it is ne-
cessary to first understand why low and stable tenure patterns became high
and rising tenure patterns. Fortunately, we can rely on a rather cogent literat-
ure investigating this change, despite its multiple layers of causation among
numerous variables. Two classic studies in particular present a fuller picture.
Polsby (1968) describes the transition of the House of Representatives into
a more efficient and internally professional body with its own rules, division
of labor, and both de jure and de facto norms. As the House took on dis-
tinct organizational boundaries, it became more complex and specialized - as
evidenced by increased autonomy from the executive branch, the importance
of committees to producing legislation, and increased salaries, staff, facilities
and perquisites of office. The House also took on specific rules and meta-
rules, such as the seniority system for committee assignments.1o In short, the
House had become a distinct and identifiable institution, both requiring and
affording the professionalization of its members. This "Institutionalization"
argument suggests several incentives toward increased tenure. Polsby reports

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
16

Turnover in Congress (1789-1997)

80

A
70

60

Six-period moving average

50

40

30

Percent Turnover in
20
House of Representatives

10

0
1789 1799 1809 1819 1829 1839 1849 1859 1869 1879 1889 1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999

Turnover in Congress (1869-1911)

70

B
60

50
Linear Trendline:
y = 54.93 - 1.01x
40
Percent Turnover in R2 = 0.62
House of Representatives
30

20

10

1865 1875 1885 1895 1905 1915

Turnover in Congress (1880-1911)

60

50

Linear Trendline:
40
y = 53.69 - 2.07x
R2 = 0.68
Percent Turnover in
House of Representatives
30

20

10

C
0

1883 1887 1891 1895 1899 1903 1907

Figure 1.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 17

a decreasing percent of first-term members, higher mean tenure, and other

similar measures starting sometime in the second half of the 19th century."1
The other classic study on which we rely is Kemell (1977), which intro-
duces developments outside the Congress that added to the internal incentives
toward increased tenure. Three factors, in some combination, prevail in Ker-
nell's analysis: 1) less party competition for seats; 2) increased political
ambition among members; and 3) diminished voluntary rotation. We come
to a clearer understanding of the transition to professionalism by examining
Kernell's factors in the context of the broader and more recent literature.
First, diminished competition for congressional seats. Due to the use of party
ballots, straight party voting characterized the late 19th century. An incum-
bent - even one with seniority - was vulnerable to losing office if his party
lost. With the adoption of the Australian ballot through the 1890s one of the
earliest incumbent advantages was introduced, and incumbents became less
constrained in controlling their individual seats (Rusk, 1970). Most scholars
pinpoint the change at the 1896 election, after which increasing numbers of
incumbents sought reelection (Price, 1975). Aside from ballot rules, gerry-
mandering was frequently in use to protect the incumbent party's seat, and
racial disenfranchisement in the south perpetuated the tenures of safe white
seats in the south (Erickson, 1995). We see, therefore, strategic barriers to
entry long before the 20th century, the era that has been the focus of all
modem economic research on competition in the congressional labor market.
Nonetheless, this diminished competition for seats explains why the decline
in turnover accelerated about 1896, but not the origination of the trend itself,
which began thirty years prior.
Second, ambition among representatives. Driven mostly by centralizing
political power and interest group demand for transfers, representatives were
given greater incentive to accumulate terms in office. Certain symptoms of
increasingly centralized federalism, while difficult to measure and perhaps
endogenous to congressional ambition, most plausibly added to the careerist's
attraction to Congress. Increasing authority of the national government took
many forms after the Civil War. The Interstate Commerce Act (1887) created
the nation's first regulatory bureaucracy and an important oversight role for
Congress. An emboldened Congress followed with the Sherman Antitrust Act
(1890) and a comprehensive array of import tariffs later that year.12 These
were conspicuous signals to the politically ambitious that their functions as
would-be members of Congress were becoming of increasingly central im-
portance to economic welfare across the nation. Congress' greater power was
not lost on interest groups either, which served to fan the fires of ambition
yet more. Union troops after the Civil War were the first major, identifi-
able interest group to successfully acquire private transfer legislation from

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
18

the national government. They succeeded because they effectively argued a

legitimate claim on the government's resources (at that time in significant
surplus), and their political support provided an expedient item of exchange
with the controlling Republican party (Holcombe, 1999). This experience
had the twin effects of setting a precedent for transfers and attracting both
additional interest groups and visionary legislators who saw the potential
of the new wealth transferring polity. As Holcombe (1999: 312) notes, "the
increased activity of interest groups, led by Civil War veterans, beginning
around 1870, increased the demand for transfers and laid the foundation for
the transition to government growth." Particularly in the form of economic
regulation, Congress began to broker increasingly significant transfers of
wealth. A large body of scholarship, for example, points to the private interest
nature of antirust (McChesney and Shughart, 1995), including the origins of
the 1890 Sherman Act (Stigler, 1969) and similarly the 1914 Clayton Act
(Ekelund, McDonald, and Tollison, 1994; Ramirez and Eigen-Zucchi, 2001).
Railroad regulation transferred wealth to farmers and alternative shippers.
Agriculture secured another transfer in the regulation of grain elevator rates
(Anderson and Hill, 1980). In sum, the post-Civil War powers of Congress,
combined with the increasing demand of interest groups, increased the value
of the marginal term in office. This was to the advantage of the politically
ambitious - and simultaneously to the disadvantage of the amateur - and led
to increasing careerism in Congress.
Third, the decline in voluntary rotation. This effect is less endogenous
yet more difficult to accurately measure. Congressional districts in the 19th
century routinely contained multiple counties and municipalities, each vying
to represent the district's seat. Parties saw very early on the need to nominate
a single candidate to the general election, and organized caucus conventions
to do so. To resolve these simultaneous intra-district and inter-party disputes,
gentlemen agreements were entered such that a member elected to two terms
would step aside and support the subsequent nomination of one prior conven-
tion rival.13 The institution was never formalized, however, and soon began to
wane under the increasing attraction - among individual careerists and their
districts/regions - to multiple terms. Eliminating the party caucus in favor of
direct primaries solidified the defeat of rotation. This additional incumbent
advantage - control of the nomination process - reinforced the trend toward
increased tenure.

Which of Kernell's three factors mattered most? The answer is important

because, if it can be shown that term limits can alter one or two factors
but not another, this would identify the potential for term limits to recover
inefficiencies. Kernell takes effort to distinguish the secular impact of the
three factors, and concludes that ambition accounts for the largest effect.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 19

But his observations are casual. Gilmour and Rothstein (1996) offer more
rigorous insight. They develop a dynamic model of congressional tenure
that contrasts electoral defeat with retirement as reasons for turnover. They
examine cohorts of entering House members and their continuation (or suc-
cessful re-election) rates, which are strictly determined in their model by
voluntary quit (q) and electoral loss (1) rates. For any ith cohort, the con-
tinuation rate is r' = (1 - q - iP). Looking at these rates over time for
multiple cohorts, the model can derive the number and/or proportion of any
ith cohort in any tth House. For example, if N freshmen enter the House in
period 1, then their number in period t is simply the geometric progression:
Ni = N (rj) (r2) ... (r"_ ).14 This methodology is used to construct profiles
of the tenure and turnover structure of congresses over time. And since the
entire model is driven by q and 1, a given tenure profile can be explained
in terms of prior loss and voluntary retirement rates. Their findings are gen-
erally consistent with Kernell's story: declining retirement (i.e., increasing
ambition) accounts for three-fourths of the decline in turnover between 1870
and 1930. And fewer electoral losses explain the remaining 25%. The shares
vary within decades, however. Gilmour and Rothstein find, for example, that
losses began to have an impact in the 1880s (whereas Kernell said it was the
1890s). Similarly, while many scholars point to the 1896 election as a turning
point, Gilmour and Rothstein (1996: 63) conclude it was "a culmination of
a long process" due largely to declining retirement. Among Kernell's (1977:
671) three factors, ambition was the key to the transition to high and increas-
ing tenure; in the broader context, the changes within and around Congress
created a situation in which:

investing one's life work in the House became safer as automatic rules
replaced the Speaker's discretion in determining committee advancement.
If the environment suggested a complex division of labor and the party
system made change possible, the increasingly careerist orientation of
its members can be viewed as providing the House with the necessary
impetus for structural change.

These structural changes provided adequate incentive for legislators to

abandon the principles that the founders relied on in forming their expecta-
tions of the congresses of the future. Gradually an amateur Congress became
professional, and the framers' nobler vision was slowly etched into the back-
ground. The market for wealth transfers became the primary descriptor of
Congress, and, as the data show, tenure began to increase at the turn of the
century. This trend has continued throughout much of the 20th century, to
which we now turn.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
20

3.2. 20th century congressional tenure

As apparent in Figure lA, congressional turnover remained relatively stable

between 1900 and the early 1950s but by the New Deal era a professional
and insulated Congress had emerged. Voluntary rotation was history, Con-
gress had become the premier political body in American federalism, and
Washington had become a reasonably enjoyable place to live. "By the 1930s"
write Gilmour and Rothstein (1996: 56), "the House was a professional le-
gislature in the modern sense. Incumbents also became safer in their seats.
Racial gerrymandering, Jim Crow, literacy tests, and female suffrage served
to homogenize congressional districts across the nation, making reelection
easier (Erickson, 1995).
Tenure did not experience a secular change again until just after World
War II. Both increasing incumbent advantages and higher wealth transfers to
attract more ambitious politicians played roles. Gilmour and Rothstein (1996)
report that turnover declined 15% between 1926 and 1988. To explain, they
contrast 1926-46 (the era of "pre-incumbency") with 1948-80 ("incumbency
advantage"). Electoral losses in pre-incumbency averaged 17% compared to
just eight percent during incumbency advantage. Because retirement rates
stayed constant over the two periods, Gilmour and Rothstein attribute all
of the increase in tenure to incumbency advantage. Similarly, Scully (1995)
shows that the probability of being in office after any number of years is
significantly higher in the 86th Congress (1959-60) than in the 57th (1901-
02). However, Scully attributes the change to increasing marginal value of
tenure in producing net wealth transfers to interest groups. Indeed, federal
spending (both civilian and military) rose steadily after an initial spike during
WWI.

Before campaign finance reform in 1974, incumbents in their first few

terms were more vulnerable to electoral defeat. This is developed in a series
of articles by Bob Reed and Eric Schansberg. It begins with the difficult task
of evaluating tenure data in recent congresses, due to the fact that stays of
sitting members are not yet complete. This may seem an innocuous challenge
but consider, for example, the 96th House that commenced in 1989. Included
in this Congress were members with first terms going back to 1953. Of all
members who entered office between 1953 and 1989, fully 32% were still in
office at the start of the 96th Congress. Measuring the stays of each entering
cohort with a mean term statistic would count only the members who had
already left office, underestimating the cohort's actual tenure insofar as its
members remained. Its importance arises in presenting accurate tenure and
turnover statistics for modemrn congresses. Use of mean tenure methodology
had limited the samples selected in previous studies including Glazer and
Grofman (1987) and Scully (1995).

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 21

Table 2. Increase in expected tenure post-1975 as calculated by Reed and Schansberg (1992).

Year entered

Continuation rate Party 1953-75 1977-89 Pct. change

used as substitute (1) (2) (3)

1977-91 Democrats 12.67 14.66 +15.7

Republicans 10.56 12.26 +16.1

1985-91 Democrats 13.31 19.16 +43.9

Republicans 10.67 13.11 +22.9

Reed and Schansberg (1990, 1992) avoid this p

rate rkn measures the probability that a member o
tinue on to the (nth + 1 term. Using this probabili
members' remaining time left in office for each j
current time in office as t, the total eventual tenu
expected remaining tenure as simply: E(SIt
By first calibrating the expression with pre-1975 c
selectively substituting post-1975 rates, they are able
effect of a given cohort's continuation rate on the
Their findings over the 1953-1991 sample are tw
an upward trend over the entire time period, but
celerates after 1975. Substituting the later continu
representatives entering the House from 1977-91
increase in expected tenure over those entering f
ings are summarized in Table 2.16 Second, Reed
to further pinpoint the source of these changes by su
four post-1975 continuation rates. They find that
tenures that result are very close to those produce
1975 continuation rates (reported in column (3) of
the continuation rates associated with the four coh
explain nearly all of the post-1975 increase in e
particularly useful finding because it coincides with
in 1974, which, by these data, appears to have inc
incumbents in their first few terms.

3.3. Tenure overview

Overall, we have a good understanding of historical patterns in Congressiona

tenure, and also reasonably cogent explanations for these patterns. Tenure
patterns in Congress were stable and low until the middle 19th century, which

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
22

was followed by an upward trend with distinct "jumps" in the late 1890s, mid
1950s, and early 1970s. The causes of increasing tenure are a combination of
internal and external factors. Most of the early increases in tenure are attribut-
able to Congress attracting increasingly career-minded individuals to office,
possibly due to the increasing role of Congress. Once there, legislators began
a gradual, incremental process of insulating the value of incumbency through
policies and institutions that dissuade challengers. High and increasing tenure
is a combined result of ambition and barriers to entry, and also serves as the
chief purpose for desiring term limitation, the arguments for which I now
discuss.

4. Two arguments for term limits

What are the adverse consequences of high and increased tenure that would
support arguments for term limits? Proponents of term limits uphold two
related efficiency arguments. First, since federal spending increases with
tenure and term limits would unambiguously decrease average tenure, then
term limits should reduce federal spending. Thus, term limits would en-
hance economic efficiency. Second, since competition for congressional seats
diminishes with tenure and term limits would regularly nullify incumbent ad-
vantage, then term limits would then make the congressional election market
more competitive. Thus term limits would enhance political efficiency. My
purpose in this section is to carefully lay out these two parallel arguments in
order to evaluate them.

4.1. Tenure and spending

It is intuitive to some that a direct relationship between tenure and spending is

relatively obvious, but the research in this area has reached highly disparate
results. In the first place, one must ask whether the supposed relationship
should hold at the individual or the aggregate level. That is, do individual
members begin to support higher spending as they accumulate tenure? Or
does Congress spend more as average tenure increases? Considering turn-of-
the-century data, aggregate federal spending did not begin its upward trend
until 1917 while congressional tenure began to increase perhaps four decades
earlier (Borcherding, 1977). Looking at post-WWII data, the four panels in
Figure 2 show that neither tenure nor turnover has a readily apparent relation-
ship to either real federal spending or federal spending as a percent of GDP
(cf. Aka, Reed, Schansberg, and Zhu, 1996: 193). In particular, real spending
increases monotonically while turnover fluctuates about it and tenure remains
relatively constant - in fact, in the early 1990s tenure falls appreciably while

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 23

spending continues to rise. In the aggregate, therefore, it would seem the

relationship between tenure and spending is far from obvious. But this is
a visual inspection only, and we should look at some of the more rigorous
econometric evidence on the tenure-spending relationship.
The supposed positive relationship is articulated by the "culture of spend-
ing" hypothesis, advanced primarily by Payne (1991). The argument is that
as a legislator spends more time in office, he becomes used to voting for
spending increases. Payne reasons that this is a function of the environment
in which legislators are immersed: from all directions they are laden with
pressure to increase spending. With more time in this environment, voting
to increase spending becomes routine; hence, tenure should be positively
related to spending. Reed, Schansberg, Wilbanks, and Zhu (1998) entertain
two additional hypotheses. First, logrolling opportunities could increase with
higher tenure, and thereby increase spending. The theoretical literature is
moot on the relationship between logrolling and spending (see discussion
in Mueller, 1989: 82-87). Stratmann (1992, 1995) offers evidence that highly
concentrated interests form coalitions around narrow issues such as individual
agricultural subsidy programs and defense programs. This is generally con-
sistent with the higher spending thesis. No study, however, either theoretical
or empirical, supports a systematic relationship between tenure and the ex-
tent of vote trading. Still, the hypothesis is intuitive, given that successful
logrolling coalitions require stability, which depends on relationships, ta-
cit agreements, and reputations that build over time. Second, tenure could
increase shirking possibilities, which could have In effect on spending de-
cisions. More likely, however, is a relationship between the last period and
shirking: the higher an individual member's tenure the more likely he is to
be in his last term. Even so, the expected sign of shirking on spending is
ambiguous.
At the individual level, some cross-sectional work has been done, with
mixed results. Sobel and Wagner (1998), for example, find that tenure is
not a statistically significant predictor of tax or spending bill sponsorship for
democrats, but it is for democrats voting in favor of spending bills. For repub-
licans, they find that tenure reduces tax bill sponsorship, increases spending
sponsorship, and does not affect voting on spending packages. This kind of
mixed result is typical, leaving the relationship between tenure and spending
muddled. Other contributions to the literature advance a slightly more certain
relationship between tenure and spending. Scully (1995), for example, uses
hazard rate analysis to compare tenure and a state's marginal value of wealth
transfers between the 57th Congress (1901-02, roughly the starting point
for increasing federal expenditures) and the 86th Congress (1959, roughly
the point of tapering off). He then compares geographic constituencies by

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
24

$Trilons
$Trilons

120 10 80 40 20 0
120 10 40 20 60
80 60 0

RealFdrSpnig

RealFdrSpnig

PercntTuoviHs ofReprsntaiv

TenuradRlFSpig1947-

TurnoveadRlFSpig1947-
AveragTnuiHosf Reprsntaiv

1947568

1953678
5 0

25- 20 15 10 5 3 2 0
35- 30- 7 6 4

Turnove TermsinOfc

Spendig
Spendig

18.0 16.0 14.0

26.0 24.0 2.0 20.
20. 15.0 10. 5.0 0.
25.0

PercntTuoviHs ofReprsntaiv

FedralSpnigs PercntofGD

Figure2.Thtn-spdaovl:USHWI

FedralSpnigs PercntofGD

AveragTnuiHosf Reprsntaiv

TenuradPctGDSpig1953-7

TurnoveadPctGDSpig1947-

1947568

1953678
5 0

15 10 7 6- 5- 4 3- 2- 0
35 30 25 20

TermsinOfc
Turnove

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 25

income, and finds that lower income locales are significantly more likely to
feature longer tenures among their delegations. Scully interprets this to mean
the marginal value of a wealth transfer increases the marginal value of tenure
in office. In other words, Scully argues, poorer districts are willing to sacrifice
the benefits of greater rotation in office to obtain the greater income redistri-
bution that higher tenured representatives could acquire. Moore and Hibbing
(1996) present a more straightforward finding. Using data on federal outlays
by district and state from 1983 to 1990, the authors find that higher tenure in
the House does not necessarily increase spending within individual districts.
However, higher House and Senate tenure increases spending within the state.
Moreover, it is the House portion of the state's delegation that produces the
results - Senate tenure alone is insignificant.
The above studies are either cross-sectional or point-to-point intertem-
poral comparisons, Stronger evidence based on panel data is provided by
Aka et al. (1996) and Reed et al. (1998). Both studies examine NTU scores
(a spending measure based on roll call voting) of representatives and senators
entering office between 1975 and 1992. Reed et al. find only a weak rela-
tionship between attained tenure and spending. In certain sub-samples, the
relationship is actually negative. The culture of spending hypothesis gains
support in the data only with House Republicans entering between 1975 and
1982. The logrolling hypothesis holds only within the senate model. And the
shirking hypothesis receives no support in the their data. The authors con-
clude that their results do not support a clear relationship between tenure and
spending, nor can they explain why tenure might be related to spending. Aka
et al. (1996) pay particular attention to the variables that measure member-
specific variation over time (most importantly tenure) versus variables that are
member-constant but vary by observation in each cross-section (e.g., party,
race, gender, age at first election). They find that isolated, longitudinal tenure
effects are generally not statistically significant and are of tiny magnitude:
20 years in office increases a senator's NTU score by one-third of a standard
deviation, and decreases a representative's score by less than half a stand-
ard deviation. Instead, the member-constant variables explain far more of
the differences in spending. An advantage of their approach is the ability
to discriminate between members' own tenure effects and the effects that are
due to sample attrition. Assuming members who exit office earlier than their
cohort have lower spending preferences, then those who remain will increase
the average spending measure. The explanatory power of their cross-sectional
variables relative to their longitudinal variables suggests the sample-attrition
effect dominates. This also explains why simple cross-sectional studies might
find a positive and significant tenure effect.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
26

At best, therefore, the tenure-spending picture is a murky one. Even so,

atop the supposed direct relationship lies a primary cause for (i.e., reason for
desiring) term limits: to reduce federal spending by reducing congressional
tenure. This reigning in of Leviathan would recover welfare and rent seeking
costs associated with excessive wealth transfers. While this argument already
seems tenuous because the relationship between tenure and spending is un-
clear, I will evaluate it further in light of the consequences of term limits that
I discuss below.

4.2. Tenure and political barriers to entry

High tenure creates incumbent advantages like permanent staffs, franking

privileges, easier access to media, and particularly brand name recognition.
Such advantages have been modeled as barriers to entry, which allow incum-
bents to be less productive and dissuade more productive challengers from
running. Social losses occur in one of two ways. In partial equilibrium, one
set of constituents either receive a non-optimal quantity or composition of net
transfers, or the costs of acquiring optimal transfers are unnecessarily high.17
In general equilibrium, all constituents either receive non-optimal transfers
or pay super-optimal costs for transfers. Both conditions are potentially
remediable by term limitation.
Lott (1986, 1987b) approaches this problem with a straightforward and
useful thesis: because incumbent investment in brand name capital is both
sunk and non-transferable (except in rare cases of familial ties), incumbents'
past campaign expenditures constitute barriers to entry that limit the effect-
iveness of potential challengers. Thus, incumbents can inefficiently represent
their constituents (note the closeness to the idea of political shirking). This
is a simple point, but Lott's analysis and empirics draw an additional import-
ant implication: entry barriers also keep lower-cost legislators out of office.
Let TC' represent the total costs of the incumbent to produce an amount
of government services, G. The challenger's total costs of G are TCc. The
basic problem is that TC1 = C1 but TCc = Cc + B. In these equations,
C's with subscripts represent the fixed plus variable costs of producing G,
and B (for "brand name") represents sunk costs not yet encountered by the
challenger. But the incumbent acquires B through campaign spending and
on-the-job advertising. In short, B is an increasing function of prior campaign
expenditures, E : &B/aE > 0 and aB2/aE2 < 0. As mentioned, B is non-
transferable. In economic markets a firm can sell its brand name at a price up
to competitors' costs to duplicate it. But in the market for wealth transfers the
firm is a politician. An incumbent can endorse another candidate, and he can
pass on his name to family members, but a politician cannot sell his name to
another candidate.18 If the incumbent could sell his brand name, there would

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 27

be no problem because the challenger would purchase it, so long as Cc < C1,
for a price greater than the incumbent's present discounted value of holding
office but less than or equal to his own (Lott, 1986: 89). But this is not the
case. As a result, for a challenger to effectively compete for the seat, either
the incumbent must exhibit high CI (be a poor producer of G) or Cc must be
sufficiently lower than CI to offset the barrier B. Lott demonstrates that this
is unlikely because C, declines with incumbent tenure while B increases with
tenure. Therefore, Lott's main point is that least cost producers (in terms of
C's) are not elected to office, and this is (politically) inefficient.19
Lott (1987b, 1989) shows this empirically by comparing incumbents' past
investment in brand name with challengers' subsequent investment. If the
incumbent's brand name does cause TCc to be greater than TCI by B, then
higher levels of B decrease the challenger's return to expenditures and should
reduce them. He finds significant evidence, for House elections during the
1970s, that incumbents' current expenditures tend to increase challengers
spending, which is consistent with common results in the campaign finance
literature. However, current challenger spending is negatively related to past
incumbent spending. In addition, controlling for incumbents' time in office
indicates that more tenure decreases current challenger spending (at an in-
creasing rate). Lott is hesitant to interpret the tenure coefficient as a barrier
because time in office may simply reflect a low TCI. But time in office can
cause an incumbent's TCI to fall over time with committee seats, logrolling
relationships, etc., which are opportunities denied to challengers.
The broader literature is generally consistent with this entry barriers
theme. Bernhardt and Ingberman (1985), for example, show that voters may
be biased toward incumbency even though it is costly. They model voters
as uncertain whether candidates will actually fulfill the promises made in
their election platforms. They show that it is costly for an incumbent to:
a) campaign for re-election on a platform significantly different from his
reputation; or b) take actions after re-election that diverge from his reputation
established in prior terms. As a result, voters perceive less risk in consistent
incumbents than in challengers. Even if the challenger's platform is closer
than the incumbent's to the median voter, the incumbent may be re-elected
if his platform is consistent with his reputation and not so distant from the
median that it overwhelms the value of the economized risk. Thus, constitu-
ents may stick with an incumbent, even an inefficient one, because he is a
safe bet. Another cost of incumbency is the potential for rent-extraction (see
McChesney, 1991). As Buchanan and Congleton (1994: 49) write:

This potential differential value between incumbency and non-

incumbency provides the opportunity for the incumbent legislators to
secure rents that increase the costs [to voters] of all special benefit pro-

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
28

jects. Incumbents have a "cushion" that may be exploited to their own

advantage.

Buchanan and Congleton abstract in their model from many actual institu-
tions through which legislators would extract rents. For instance, Weingast
and Marshall (1988) stress the importance of long-term durability to wealth-
transferring contracts. Logrolling and the committee system are especially
noteworthy as examples of such durability-creating institutions. But each of
these, in turn, depends on individual members being in office for a long
time.20 Hence, we come back to the centrality of tenure to these many
inefficiencies. Since term limits would reduce tenure, they would reduce
political entry barriers, and reduce or eliminate the many associated political
inefficiencies.

5. Consequences of term limits

I now consider effects on economic and political data when term limitation is
moved to the right hand side of the equation. As Bernard Grofman (1996a)
noted in the important collection of term limits papers that he edited, this is
not a simple question. One is confronted less by answers than by "hypotheses
in search of data" (Grofman, 1996b). Legislative term limits were enacted
only recently, and in some of the most interesting cases (e.g., California), the
laws were significantly scaled back or changed. So data on their effects are
not especially prevalent. As a result, some of the evidence considered here
draws from calibrated simulations. In other cases, though - usually regarding
executive term limits - hypotheses have found their data and the literature
stands on actual empirical observations. I will emphasize the consequences
of term limits vis-h-vis the causes outlined earlier in the paper, but additional
effects that are separate from (or at least not strictly bounded by) the causes
will also be explained.
In tracing the consequences of term limits, I will take the view that term
limits represent an institutional change (Landes and Posner, 1975). Unlike
policy changes, that alter the incentive structures of economic agents in the
polity, an institutional change such as term limits alters the incentive struc-
ture faced by policy makers themselves. In other words, we can expect term
limits to impact pre-policy data such as the demographic, partisan, and ten-
ure/turnover characteristics of the legislature, the balance of powers among
the branches and levels of government, and the competitiveness of the elec-
tion market. In turn, these changes shift the policymaking landscape, so that
we can expect to see subsequent changes in the policy data such as tax laws
and spending programs. Following this cascade of effects, the consequences

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 29

of term limits can be compared to the causes developed earlier in the pa-
per, providing a basis for a positive evaluation of the net benefits of term
limitation.
Before proceeding further, I must make note of two caveats. First, the ef-
fects of term limits depend on the institutional setting and on the design of the
particular restriction (Cain, 1996). In reading the legislative history of term
limits in Congress,21 I found that the most important features of design are:
a) length of term; b) number of terms limited; c) grandfather specification;
and d) whether the limits were lifetime or allowed "reentry".22 To illustrate
this caveat, suppose reelection is a function of reputational capital, which is
in turn a function of tenure. Then the effects of term limits will depend on
the extant tenure profile and the functional relationship between reputational
capital and tenure. Paraphrasing Cain (1996: 27), if average tenure is high
and brand name capital depends strongly on tenure, then term limits should
increase electoral competitiveness. But if the proposed term limit is high
relative to average tenure, then the chain is broken. The House simulations
by Reed and Schansberg (1994, 1995), which I discuss in detail presently,
reinforce this point. Their results produce greatly different effects on tenure
profile when the number of terms limited is changed or when grandfathering
is removed. The net benefits of term limitation vis-a-vis the status quo and
other reforms may depend mostly on the form that the restriction takes.
Simulation models to predict the effects of term limits introduces the
second caveat, the "Lucas critique" (Lucas, 1976). Simulation models cal-
ibrate past continuation rates, then forecast future tenure profiles given these
same rates and a term limit. But term limits may incite changes to the under-
lying structural parameters that determine continuation rates. To the extent
of such an effect, these types of simulations may generate erroneous fore-
casts. Without exception, authors of these simulated studies acknowledge
this problem, as does Grofman (1996b: 7) in the introduction to his edited
volume. It is, of course, trivial to model the effects of structural changes once
they are known: simply update the forecasts with the altered continuation
rates. But the essence of Lucas' critique is that structural changes are partly
determined by the manner in which expectations are formed, in addition to
preferences and certain political institutions. Applying Lucas' argument in
this case, expectations depend partly on the absence of term limits, and are
therefore likely to change systematically following their enactment. In brief,
simulated forecasts of the effects of term limits using extant continuation rates
should be interpreted carefully because there is good reason to expect the
continuation rates to change systematically after term limits.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
30

5.1. Impact on political variables

5.1.1. Tenure turnover profile of legislature

It is widely believed that term limits will unambiguously reduce average ten-
ure, so long as the constraint is "binding". In order to "bind", the term limit
must be lower than what some critical proportion of the legislature's tenure
would have been without the restriction. For example, Moncrief, Thompson,
Haddon, and Hoyer (1992) gather recent tenure and turnover data for the state
legislatures, and find that a 12-year, no-grandfather, re-entry term limit law
would end the tenure of only 27% of legislators in lower chambers and 36%
in upper chambers. For state term limits laws to have a larger impact, the term
limit would have to be shorter in all but the most professionalized state legis-
latures. Opheim (1994) then examines an eight-year, no-grandfather, re-entry
law in states where such laws had already passed, and finds that, on average,
over 50% of state legislators would be affected. According to this conven-
tional view, the extent to which a term limits law "binds" will determine how
much it reduces average tenure.
The degree of this reduction is a more complicated question, and it com-
plicates the above answer. Legislators might anticipate the effects of term
limits and leave office before theirs "binds". Wayne Francis and Lawrence
Kenny address this question with "early" turnover evidence at the state level.
They assert two separate effects prior to reaching the term limit. First, the
"indirect" effect (Francis and Kenny, 1997) occurs because house members
facing a term limit see a move to the senate as a natural step, which is accom-
modated by term limits in the senate creating more open seats there. A house
member would pursue higher office before reaching the term limit if a senate
seat were to open. Second, the "present value" effect (Francis, Kenny, and
Anderson, 2000) considers how legislators perceive the marginal changes in
their private- and public-sector career opportunities. If the discounted value
of lifetime earnings through politics is reduced enough because of term lim-
its, legislators would substitute into private sector employment even before
reaching their maximum terms. Francis and Kenny compare extant turnover
rates to turnover in 1994 and 1996, when term limits laws enacted in 1990
began to take effect and the early effects of laws passed in 1992 could be
observed. They control for low extant turnover (more room to increase) and
the ratio of house to senate seats. They find that a term limits dummy vari-
able increases turnover by 4.3% in 1994 and 2.8% in 1996. And for each
year closer to the limit, turnover increases by .89 and .45%, respectively.
Higher turnover in the senate also increases house turnover in their results.
Keep in mind that they examine only states that had not yet hit the term
limit in 1994 or 1996. Therefore, these results are interesting because they
suggest, contrary to conventional wisdom, that term limits do not have to

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 31

"bind" in order to reduce average tenure. More fundamentally, they suggest

that term limits do not make legislators any less ambitious. The main cause of
increased congressional tenure, discussed in Section 3 above, would remain
even though term limits would reduce average tenure. The reason could be
that term limits alter the structural parameters underlying tenure/turnover
rates (e.g., legislators' expectations), but they do not alter the parameters
underlying political ambition (e.g., legislators' preferences). Moreover, this
interpretation is consistent with the Lucas critique so that these results temper
the results obtained in the balance of the literature on term limits and tenure,
which consists entirely of simulation studies.
I discussed the simulation methodologies of both Reed and Schansberg
(1990, 1992) and Gilmour and Rothstein (1996) in Section 3.1 above. These
methodologies are applied to simulate the effects of term limits in several
papers (Reed and Schansberg, 1994, 1995; Gilmour and Rothstein, 1994;
Franklin and Westin, 1998; Francis and Kenny, 1997). These papers all em-
ploy a similar kind of geometric progression based on continuation rates.
There are four reasons for a legislator to leave office: death, retirement,
pursuit of higher office, and electoral defeat (Schansberg, 1994).23 Death
is typically ignored because it is uncommon, and retirement and pursuit of
higher office are typically combined as "voluntary quits". Where 1 is loss
rates and q is quit rates, the continuation rate will be r = 1 - 1 - q. Now
suppose a term limit law is imposed at time period t. These studies gather
data on r for many years prior to t, and assume these rates will remain
somewhat constant. As such, they are able to simulate future tenure profiles
by iteratively calculating the number of members in their ith term at time
(t + k): Nt+k = N1 (I?)I () ... (t+k-1). Term limits can easily be incorporated
by adding the constraint that (t + k) cannot exceed the maximum number of
terms allowed by the particular law imposed. That is, if terms are limited to m
terms, then all continuation rates after period (t+m) equal zero. (Grandfather-
ing creates special cases. The above equation implies limited-grand fathering
because it treats extant incumbents the same as members entering in period
t. A no-grandfather law would mean ri == 0 for i > m and for periods t and
hence. A full-grandfather law would mean ri = r' for all time periods so long
as the cohort was extant at period t.)
The simulated effects of term limits on turnover depend on which period
of extant continuation rates is used. I will not here discuss the merits of one set
of rates versus others, though arguments can be made. Reed and Schansberg
(1994) predict the steady-state turnover to range between 11 and 15% in the
absence of term limits. With a 12-year, full-grandfather term limit, steady
state turnover increases to just above 20%. Limited- and no-grandfathering
eventually reach the same steady state. In their later paper (1995) Reed and

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
32

Schansberg expect a six-year limit to increase turnover to 37% in steady

state. Gilmour and Rothstein predict very similar results. From Figure 1, it
is evident that this would return Congress to a tenure profile not seen since
the 19th century. Moreover, in light of Francis and Kenny's "anticipatory"
effects, these steady state predictions must be considered conservative.
Political scientists have traditionally been concerned with high turnover
at the state level because it causes inexperience and could reflect a lack of
seriousness among legislators (Francis and Baker, 1986). Turnover at the
rates predicted above can potentially cause two problems, both related to the
"superclass" phenomenon. With term limits, Congress is inundated with large
waves of incoming freshmen as an equally large percentage of members reach
their maximum terms. Before reaching steady state, turnover spikes at inter-
vals equivalent to the maximum years allowed in office. Twelve years after
enacting a 12-year limit, 44% of Congress would be in their first term, and 24
years after 33% would be freshmen. With a six-year limit, the superclasses
would be 71 and 61% six and 12 years out (Reed and Schansberg, 1995).
The first problem with such superclasses is the low level of experience with
such large numbers of freshmen legislators. Without term limits the median
experience in Congress would be 10 years. With a six-year limit in Reed and
Schansberg's model, the median experience level would be one year during
and soon after a superclass. A 12-year limit is less severe, producing a me-
dian experience of seven years after a superclass. It is difficult to say how
this might matter. The legislature could fail to function, be incapable of re-
sponding to emergencies, and incur unnecessary organizational and learning
costs. Nothing in the literature investigates this. Superclasses might also cause
a related problem: the large numbers of outgoing lame ducks, which may
cause a widespread shirking problem. However, legislators do not appear to
vote systematically differently in their last term, at least in the current system
without term limits (Bender and Lott, 1996; though see the important caveat
to this result in the section on shirking below). The evidence from Francis and
Kenny (1997) and Francis et al. (2000) suggests legislators will anticipate the
term limit and rotate without such spikes at six- or twelve-year intervals. It
also seems intuitively implausible that competent people will not be available
for office:

The notion that one can understand policy tradeoffs only after years
in political office seems especially misplaced ... For starters, a pool of
academics study how policy is made and works; they could step into
legislative work with much background knowledge. Many people outside
academia have advanced training in public policy and public adminis-
tration programs. Similarly, many lawyers are trained in the writing and
implementation of legislation, and deal with policy questions day in and

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 33

day out in their work ... [M]any persons are qualified candidates for
legislative positions (Glazer and Wattenberg, 1996: 44-45).

Finally, it is conceivable that term limits would not decrease, and perhaps
even increase, average tenure. The logic here is that term limits cause poten-
tial challengers to postpone running until the seat is opened by mandatory
rotation. Grofman and Sutherland (1996) present a stylized model that pro-
duces this outcome. In their model, the probability of a challenger defeating
an incumbent is lower than the probability of winning an open-seat election.
The key feature of their model is that once a candidate is defeated (whether
running against an incumbent or not), the candidate's party finds another
(stronger) candidate for all subsequent election bids. If you lose once, you
cannot run again. Since strong candidates will then defer until term limits
create an open seat, then incumbents are re-elected with certainty at their
discretion. If the conditions of their model hold, and voluntary quits are low,
and the term limit does not "bind", then a term limits law could conceivably
increase average tenure. Such conditions are unlikely, however. As discussed,
evidence in Lott (1989) suggests that challengers use current campaigns to
invest in reputation for later campaigns, and the work by Francis and Kenny
(1997) and Francis et al. (2000) suggests a term limit does not have to "bind"
to affect rotation patterns.

5.1.2. Party balance

By far the conventional wisdom during the 1990s maintained that term limits
would benefit republicans and harm democrats. Democrats held majorities
in Congress and most of the state legislatures; term limits would eliminate
this incumbent advantage at regular intervals. Democrats are also more likely
to make lifelong careers in politics, and have higher mean tenures, both of
which are punished by term limitation. For these reasons, Grofman (1996b: 8)
notes, many journalists accused republicans of harboring a "hidden agenda"
in their support of term limits. This partisan split is generally consistent with
actual voting on term limits. Congressional republicans supported the 21st
Amendment more than democrats (Friedman and Wittman, 1995).24 State-
level initiatives in California found more support among republican voters
(Donovan and Snipp, 1994). And in the 1995 and 1997 roll calls, republicans
in both chambers were far more likely to vote for congressional term limits
(L6pez, 2002).
Observing such partisan support for term limits does not necessarily sup-
port any speculation as to why republicans would have a "hidden agenda".
My purpose in this section is to survey the predicted effects of term limitation
on party balance, and from this infer an empirically plausible argument for
why republicans wanted term limitation.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
34

Geometric progression models have been used to simulate the effects of

term limits oil partisan balance in Congress. First, party balance is calculated
at time t such that R: and D: are the numbers of incumbent republicans and
democrats. Continuation rates are then calculated by party, rr for republicans
and rd for democrats. If loss (1) and voluntary quit (q) rates differ between the
parties, this will influence the effect of term limits on each party. So patterns
of exit and new entry into Congress must be accounted for, again by party.
The number of non-incumbent republicans running for an open seat at time t

is both
So Rt =parties
(1-qr)(R:)+(1-qd)(D ). Similarly,
are running new faces DNby
in seats vacated = their
(1-qd)(D:)+(1-qr)(R:).
own and the
other party - the basic nature of open-seat elections. If we know the rate of
open-seat victory by party, vr and vd, then we can calculate the number of
republicans and democrats during any tth Congress:

Rt = (1 - 1r)(R_ 1) (qr) + (vr)(R_ 1) and

Dt = (1 - ld)(D_ 1) (qd) + (Vd)(DN ).25

The effects of an m-term limit imposed at time t can be easily incorporated
by adding the constraint that qr _= qd = 1 for any cohort in their t + mth
term. (Again grandfathering creates special cases. The above would hold for
the limited-grandfathering case, but a no-grandfather law would imply qr =
qd = 1 at time t for extant cohorts above m terms. Full grandfathering simply
means qr = qr and qd - qd for all time periods and all extant cohorts at time
t.)
By the above equations, the simulated effects of term limits on party bal-
ance depend on extant loss rates, quit rates, and also open-seat winning rates
- not initial party balance and tenure as the "hidden agenda" argument would
hold. Gilmour and Rothstein (1996) compare two sets of extant rates, the
90th-100th Congresses and the 98th-100th Congresses. In the 11-congress
sample, democrats are better at winning both incumbent and open seat elec-
tions: (1 - Id) > (1 - 1r) > Vd > Vr. But in the three-congress sample,
open-seat republicans have a higher winning rate: (1 - Id) > ( - ) >
vr > vd. Of great importance is that republicans have a higher voluntary quit
rate over all samples (qr > qd). Table 3 summarizes their simulated results.
Without term limits, the size of the democratic majority is smaller using the
three-sample loss rates, in which republicans win more open seat elections.
With term limits, however, Gilmour and Rothstein prove theoretically that
republicans will always experience net gains under these more recent condi-
tions (see their Theorem 4, p. 782). They also simulated term limits laws of
varying lengths, and found that republicans always win under extant loss and
retirement conditions. For example, a 12-year, limited-grandfather, lifetime

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 35

Table 3. Predicted Democrat-to-Republican seats; steady state with and without term limits;
results from Gilmour and Rothstein (1996).

Extant loss rates used No term limits Six-year term limit

90th-100th 264-171 241-194

98th-100th 242-193 not reported

Note. Retirement rates used are from 90th-100th Cong

term limit will produce a steady-state with the

from 264-171 to 241-194. Reed and Schansbe
to simulate steady-state party balance. Without
gain a net between 22 and 32 seats (using, r
party-specific continuation rates). Under a 1
ans would gain an additional 10 to 40 seats. E
the republicans became the majority party,
additional six to 12 seats (Reed and Schansbe
els suggest that term limits will always favo
however, conflict with the conventional "hid
Republicans benefit from term limits not b
party or because they have lower average te
depend on initial tenure or party balance. Ins
that republicans have a higher quit rate, and
more often. It is possible that the reason is a
open seat races. But the above results would
equal percentages of open-seat elections (Ree
precisely, it is the spread between incumben
rates that matters most:

It is true that each party is more likely to lose its seat in each race where its
incumbent cannot run. However, the increase in the likelihood of losing
these seats is not in general the same for both parties. This indicates
the logical possibility that the party that retires fewer incumbents may
suffer a net loss of seats if it also suffers a substantially greater increase
in the likelihood of losing each race ... (Gilmour and Rothstein, 1994:
771-772).

Republicans wanted term limits not because they were the minority party with
lower tenure, but because they lose reelection more often and were beginning
to win open seats more often. It has been observed that Congressional Re-
publicans lost a great deal of interest in term limits after they had gained the

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
36

majority in Congress. This analysis suggests the reason is not that they gained
the majority, but that they have been doing better in their open seat elections
relative to the Democrats.

5.1.3. Value of office

5.1.3.1. Value of office to legislators. A legislators' value of holding office
is defined by the returns to that office in the market for wealth transfers. Profi-
ciency in brokering wealth transfers is a function of parliamentary rights such
as committee seats, agenda power, and leadership positions, all of which are
positively correlated with tenure. So the value of holding office increases with
expected tenure in office. But the value of holding office depends also on how
soon such parliamentary rights are acquired and how long a member retains
them. More precisely, the value of holding office can be said to increase with:
1) higher expected length of completed tenure; 2) shorter waiting period prior
to become eligible for a leadership position; 3) higher probability of then at-
taining a leadership position; and 4) higher expected tenure in that leadership
position once attained (Reed and Schansberg, 1994).
The value of holding office depends on how term limits influence these
four effects. On the one hand, term limits would decrease expected completed
tenure. Using their methodology for calculating expected stays with truncated
observations (discussed above in Section 3.2), Reed and Schansberg (1995)
are able to compare expected length of completed tenure with and without
term limits. Based on 1973-94 continuation rates, length of tenure would
average 13.2 years. That average would drop to 6.1 years with a six-term,
limited-grandfather, lifetime limit, and to 3.1 years with an otherwise similar
three-term limit. This effect decreases the value of holding office. But term
limits would also decrease the wait for leadership. Without term limits, the
required wait for appointment to leadership is 16.2 years. If seniority is still
used after term limits to allocate leadership positions, that wait falls to 10 and
4 years under six- and three-term limits respectively. In the more likely case
that seniority is not the main criterion for becoming a leader, the wait would
drop even further. This effect increases the value of holding office. Thus,
despite the unambiguous decline in average tenures that can be expected from
the enactment of term limits, these simulations suggest that the total effect on
the value of holding office is ambiguous.26
Alternatively, the value of holding office can be inferred by legislators'
revealed preferences. In states that imposed legislative term limits, house
turnover has increased as ambitious members run for more open seats created
in their respective senates (Francis and Kenny, 1997; Francis, Kenny, and
Anderson, 2000). Representatives are demonstrating a willingness to forfeit
more (and virtually guaranteed) time in the House in exchange for lower

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 37

risk in pursuing a Senate seat. This implies that the reduced expected tenure
outweighs the quicker path to leadership, such that the net value of holding
the House office has declined after term limitation - at least relative to a seat
in the relevant Senate.27
A lower return to holding office would imply that the legislators of highest
marginal productivity leave. Similarly, a marginally less productive set of in-
dividuals is attracted to run for office. Whether term limits will cause a more

productive set of individuals to enter the legislature depends primarily on

defining legislator productivity, and secondarily on the functional relationship
between term limits and productivity so defined. From the term limits debate
of the 1990s, definitions of legislator productivity are largely normative in
nature.28 These normative arguments are largely non-falsifiable; and I will
avoid any evaluation of them here.29 I return to the positive aspects of this
issue in the discussion of political shirking below.

5.1.3.2. Value to external interests: Campaign expenditures. Interest

groups and advocacy organizations could face higher costs in relating to le-
gislators under term limits. Term limits might cause interest groups to face
higher variance in winning policies because relationships with individual
legislators and parties will be less stable. In addition, interest groups could
face higher winning coalition costs due to decentralized power, less internal
structure and specialization, and more frequent changes in leadership (Cap-
pell, 1996).30 In this view, interest groups would have more power after term
limits, but also more responsibility for writing, monitoring, and whipping
their own legislation. The value of office to these and other external interests
is perhaps most easily proxied by how term limits would affect campaign
contributions. If higher variance and higher costs of legislation increase the
marginal value of bribery, both explicit and through campaign contributions,
then term limits would increase campaign spending. An alternative view is
that campaign spending is a form of investment, both by interest groups in
future policies (Snyder, 1990, 1992) and by legislators in future reputational
capital (Lott, 1989). Term limits effectively tax this investment by driving
the return to zero after the maximum terms in office (assuming the legislator
does not seek higher office). The available evidence supports this latter view.
Mixon (1994) provides indirect support using crossstate regressions to indic-
ate statistically fewer lobbyists per capita in states with gubernatorial term
limits. More directly, Daniel and Lott (1997) use data from California elec-
tions between 1976 and 1994 to find that term limits there reduced election
spending by 31% (from roughly $310,000 to roughly $215,000 per race),
even with a positive and significant time trend.31

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
38

5.1.4. Efficiency of representation: The shirking problem

Shirking is conventionally thought to increase with either a stronger legislator
preference for shirking or a relaxation of the reelection constraint.32 Thus, a
given legislator may choose to shirk more after announcing his retirement,
or the farther away is the date of the reelection bid, or the greater are the
barriers to entry by effective challengers. Shirking may also depend on the
structure of the agency relationship, increasing as slack between principal
and agent increases. Term limits could alter shirking through its effect on
tenure, turnover, the form of the agency problem, or the returns to holding
office.

The costs of term limits associated with tenure depend firstly on the
relationship between tenure and legislator productivity. It is highly implaus-
ible that this function is monotonic. Low and increasing tenure imparts
parliamentary rights (representative capital) that make the legislator more
productive and increase net transfers. On the other hand, high and increasing
tenure erects barriers to entry that worsen the effects of Downsian ignor-
ance and could lead to increased shirking. By historical standards, at least,
Section 3 has shown that tenure is high and increasing. Therefore, the costs
of high and increasing tenure are avoided with term limits, and one would
predict term limits to reduce legislator shirking. As I discussed in Section
2.1 above, Reed et al. use a panel-data regression approach to investigating
shirking behavior. They use regressions on panel data to try to relate shirking
to tenure, if shirking increases with tenure, and term limits reduce tenure,
then term limits should reduce shirking. They find no support for even the
first relationship, and conclude that term limits would not have an effect
on shirking. An important problem with this approach is finding adequate
proxies for the legislator's shirking preference and especially the amount
of shirking chosen. These problems cause biased estimates that mask true
shirking behavior (Bender and Lott, 1996). So this should not be considered
a conclusive result.

Shirking could be affected separately by the turnover effects of term limits.

By definition, term limits increase turnover to the extent that they decrease
tenure. With increased turnover will come a greater frequency of legislators
serving their final term. A counter-prediction to the tenure effect emerges.
Because the reelection constraint does not bind in the last term, and term
limits increase the frequency of last terms, one would predict term limits to
worsen the shirking problem (Glazer and Wattenberg, 1996). While Francis
and Kenny (1997) and Francis et al. (2000) allay concerns about superclasses
of last-term members, their results actually worsen this problem because with
increased voluntary quits (pursuit of higher office) even fewer reelection con-
straints bind among sitting legislators. Cohen and Spitzer (1996) address this

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 39

using a modified prisoners dilemma supergame in which voters and repres-

entatives reach a cooperative equilibrium. Incumbents prefer staying in office
without shirking to losing an election bid, and constituents prefer keeping a
cooperative incumbent to electing a freshman, especially one who will shirk.
Because term limits increase the frequency of last terms, it is ensured that
"a substantial fraction of members will be lame ducks, and that another sub-
stantial fraction will have only one remaining election". (Cohen and Spitzer,
1996: 57) By this theory, term limits tend to increase shirking. These last term
arguments depend first on whether shirking increases in the last term. The
received empirical literature on this question demonstrates this is unlikely
to be the case: although absenteeism increases, legislators do not tend to vote
systematically differently in their last term (Bender and Lott, 1996). However,
this literature is somewhat narrowly reliant upon interest group ratings of final
roll call voting, such as ADA- and NTU-scores, to proxy shirking. Much of
a legislator's activities, including shirking, occur prior to floor voting. So this
literature uses relatively little information to derive the result that the last term
effect is not real. On this evidence, the Cohen and Spitzer effect may or may
not be a significant problem. Second, and more importantly, cooperation is
sensitive to both players' knowledge of being in the last period. Because of
anticipatory effects associated with legislators' progressive ambition, Francis
and Kenny (1997) show that members in their final term will generally repres-
ent a very small percentage of the legislature (6.7% for a four-term limit). If
this is the case, then voters and legislators are again uncertain of when the last
period will be, and the Cohen and Spitzer shirking problem disappears - irre-
spective of the evidence on last period shirking. Finally, the model relies on
an efficient electoral sorting mechanism; that is, on the unrealistic condition
that there is zero agency slack.
Term limits could separately affect the form of the agency relationship
and, thereby, shirking. Specifically, slack would likely increase. The cooper-
ative supergame outcome in Cohen and Spitzer (1996: 63) is characterized
as an "institutional equilibrium". Shirking is currently policed by internal
rules such as the Senate ban on government paid foreign travel after a failed
reelection bid, and a one-year ban on lobbying the Senate after leaving office.
But these institutions, according to Cohen and Spitzer, are endogenous to
electoral rules such as term limits. Internal regulation of shirking requires
a sufficient proportion of members willing to exchange last-term shirking
benefits for increased tenure benefits. Because they eliminate the benefits of
increased tenure, term limits would eliminate this gainful tradeoff and such
policing institutions can be expected to fall by the wayside. With greater slack
in monitoring, legislative shirking should increase by term limits.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
40

Yet another avenue is for term limits to affect shirking through the returns
to holding office. Term limits might reduce the reelection motive, but they
increase the election to higher office motive. Because success in seeking
higher office depends in part on owning a reputation as being productive, this
increases the cost of being known as a shirker (Glazer and Wattenberg, 1996).
Moreover, because the value of the lower office has declined, this could attract
legislators with a stronger or weaker preference for shirking.
Amid this muddled picture, Bender, Haas, and Kim (2000) offer some
potentially cogent answers (though their results are preliminary at the time
of this writing). They develop an estimable stochastic model of congress
based on the microeconomic behavior of individual members. Each member
chooses an optimal degree of shirking in a constrained dynamic maximiz-
ation of expected discounted lifetime utility. Voters, of course, provide the
constraint by punishing shirking with reduced reelection probability. Should
an incumbent be defeated or retire, he is replaced with a challenger whose
age and shirking preferences are randomly chosen from distributions for
each variable. Parameters for voter sorting, incumbent retirement, and the
distribution of both incumbents' and challengers' shirking preferences are
estimated from twelve terms of data using a minimum chi-square method-
ology and simulation. Specifically, an initial congress is constructed based
on a given set of parameter values, and the incumbents go through eleven
consecutive simulated elections. Parameter values are then perturbed and ini-
tial congress and simulated elections are performed until the chi-square value
used is minimized. These twelve simulated congresses closely fit the actual
data.33 The resulting parameter values are then used to simulate fifty addi-
tional congresses with and without term limits and observe optimal shirking
behavior. The term limits used are twelve- and six-year, fully grandfathered,
lifetime limits (where applicable, term limits are imposed starting with the
13th congress). Their results are threefold. First, they find effects on mean
tenure that are highly consistent with the findings of Reed and Schansberg
and Gilmour and Rothstein discussed above. Second, they find no change
in the preferences for shirking under term limits; the restriction does not
draw systematically better nor worse candidates - at least where shirking is
concerned.34 Third, they find that the optimal choice of shirking increases by
80% under a 12-year limit, and by 160% under a six-year limit.35 The major
advantages of this empirical approach are that there is no assumption that all
legislators have a positive demand for shirking and that there is no use of an
empirical proxy for (the not directly measurable) shirking.
The logic of this result is, by comparison to the empirical methodology,
quite simple. Term limits relax the reelection constraint built into the model.
As such, the expected costs of shirking are reduced, and the quantity of shirk-

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 41

ing increases. Shirking increases more for a six-year limit because there is
even less time for voters to punish shirking - that is, for sorting to occur -
and also because members have a shorter time horizon and therefore greater
incentive to shirk. This is only one result - and again, a preliminary one - but
it appears to be somewhat compelling. It would support the last term effect
not being the vehicle for increasing shirking, and suggest that the pursuit of
higher office is overwhelmed by the relaxation of the reelection constraint,
such that the overall effect of term limits is to increase shirking. This is so
even under the assumption that term limits do not attract candidates with
stronger shirking preferences. It is important to keep in mind, however, that
the Bender/Hass/Kim model is institutionless in the Cohen/Spitzer sense, and
so it does not address the impact of term limits on agency slack.

5.2. Impact on policy variables

Much of the economics literature dealing with term limits treats them as a
regressor in various types of models aimed at explaining their effects on
economic variables such as state spending, taxes, public capital, etc. This
empirical literature focuses primarily on executive rather than congressional
or legislative term limits. Crain and Tollison (1993), for example, use a
time-inconsistency model to show that gubernatorial term limits foster more
frequent and more certain changes in regime, which helps to induce more use
of strategic fiscal policy (i.e., more variability in fiscal discretion variables).
The result is diminished fiscal stability and, ultimately, worse economic per-
formance. Similarly, Crain and Oakley (1995) use a strategic fiscal policy
model to reveal a larger per-capita stock of public capital in states with
gubernatorial term limits. Besley and Case (1995) present a thorough analysis
in which term limits are the primary regressor on taxes, expenditures, state
minimum wages, and workers' compensation. Their findings suggest that
term limits reduce the incentive for politicians to maintain political reputa-
tion (i.e., to work hard). Gubernatorial term limits, therefore, help to increase
sales taxes, income taxes and per-capita state expenditures, while they tend to
decrease the state minimum wage. In a preliminary paper, Crain and Johnson
(2001) echo these results for chief executives of OECD countries. Unlike Be-
sley and Case, however, they find that the effects on fiscal performance differ
systematically under a single-term versus a two-term limit. In countries with
a one-term presidential limit, government growth is higher than in countries
without, though fiscal volatility is dampened. A two-term limit significantly
increases fiscal volatility, and retards economic growth as in Besley and Case.
In both studies, the reduced incentive to establish and maintain a sound polit-
ical reputation drives the results in the presence of term limits. Increased fiscal
volatility has been demonstrated more generally to retard economic growth

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
42

(Hopenhayn and Muniaguerria, 1996). A small area of the literature examines

fiscal effects of legislative term limits, but the results are comparatively weak
and preclude direct inference as to the fiscal policy effects of legislative term
limits per se. As discussed above, Reed et al. (1998) find weak and conflicting
evidence on whether higher congressional tenure increased federal spend-
ing. Similarly, Owings and Borck (2000) find that states with professional
legislatures exhibit higher per capita state government spending.36 Reducing
tenure, however, will not necessarily reduce federal spending, an argument to
which I return below.

6. Will term limits improve efficiency?

The first part of this paper has been directed at an apparent paradox: why con-
stituents would support incumbency and term limitation. The second part of
this paper is directed at answering two related questions. First, will congres-
sional term limits reduce federal spending and therefore enhance economic
efficiency? Second, will term limits reduce barriers to entry by effective chal-
lengers and therefore enhance political efficiency? In this section, I evaluate
these two questions based on the findings as laid out thus far in the paper. The
literature speaks reasonably clearly in answering the first question, but less so
on the second.

None of the research that I have studied suggests reasonable evidence

that term limits would reduce federal spending by virtue of reducing tenure
and thus create economic efficiency gains. In the first place, the relationship
between tenure and spending is at best weak. Different studies have produced
conflicting or only loosely related results. To the extent that there is a dir-
ect correlation, it is a puzzle why spending might increase with tenure. We
simply know very little about this relationship. On the other hand, we can
be fairly confident that term limits will reduce mean tenure. Moreover, with
the geometric progression simulation models discussed throughout this paper,
we even have precise, though conservative, estimates of the magnitude of this
change. However, we cannot, based on the literature, then predict either the
direction or magnitude of the subsequent effect on spending. Redistribution
(spending) is a function of relative tenure, not absolute tenure, and univer-
sal passage of term limits does not alter relative tenure. Unless the value of
government to narrow interests is fundamentally changed, an electoral reform
such as term limits will simply rearrange the distribution of benefits. In other
words, term limits may alter the return to politicians being in office, but the
value to special interests of influencing policies remains the same. If anything,
in the unlikely event that term limits on legislators mimic term limits on
executives, the literature shows that the fiscal effects are roundly negative

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 43

and efficiency reducing. In light of these factors, the tenure-spending argu-

ment for term limits appears rather simplistic. It begs, for example, several
important questions. What is the causation between tenure and spending? An
increased role for Congress was the attraction for career politicians in the first
place; now decreasing tenure is supposed to reduce spending? If we cannot
address the endogenous properties of the relationship, we cannot predict the
consequences of limiting terms on this relationship. Nor can we legitimately
assume that these consequences would be beneficial. The tenure-spending ar-
gument is too simplistic. Its main practical weakness is that term limits do not
combat the single most important underlying force that increased both tenure
and, perhaps, spending: term limits do not make legislators less ambitious,
and do not attract a less ambitious set of candidates to office. Overall, based
on this review of the literature, term limitation must be regarded as a very
uncertain instrument regarding spending.
The literature indicates that term limits and political efficiency (as under-
stood in this paper) is a more complicated issue. First, the effects of term
limits on electoral competition depend at least partly on how the restriction
is designed. If the term limit is too long, or it allows re-entry into office after
a brief absence, then it will not have much of an effect. Conversely, if the
term limit is too short, it interferes with the ability of the election market to
do its job - for electoral sorting to occur. What is needed in the design of
term limits is an intermediate balance. But even here, on this simple point, is
where the interesting discussion begins. The effect of term limits on electoral
competition depends, less obviously, on whether the relationship between
tenure and entry barriers is robust to imposing term limits. It is conceivable
that, after term limits, entry barriers would increase with some metric other
than seniority - namely those attributes that lead to success in open-seat
elections. The creation of barriers could also shift to external criteria such
as interest group influence. And, depending on the direction and manner of
change in campaign spending brought about by term limits, the effects could
favor incumbents, challengers, or neither. It is apparent from the California
experience that term limits reduce campaign spending. Whether this reduc-
tion enhances electoral competition is the subsequent, unanswered question.
Clearly, the reduction narrows the gap between accumulated reputational cap-
ital of incumbents and low reputational capital by challengers. This suggests
that term limits reduce incumbent campaign spending relative to challengers,
which lowers entry barriers and makes the election market more competitive.
If so, then term limitation is a redundant instrument. Why not simply impose
asymmetric caps on campaign expenditures, as Lott recommends (Lott, 1986,
1989)? Whether term limits and campaign finance reform are substitutes is an

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
44

unanswered question that requires more research.37 It also suggests looking

at the underlying sources of political entry barriers.
Term limitation and properly designed campaign finance restrictions could
be substitutes in recovering political efficiencies if the spending/contribution
limit is properly designed. Gordon Tullock presented a paper entitled "Entry
Barriers in Politics" at the 1965 American Economic Association meetings.
Tullock (1965: 459, 464) argues that barriers exist because government itself
is a natural monopoly.

The natural monopoly here comes from a technological consideration

which amounts to a very strong economy of scale: only one majority can
exist at a time. ...

The scale advantage which acts as a barrier to entry is the majority vot-
ing rule (if another voting rule is actually used, then analogous problems
arise) which provides that the 'entrepreneurial group' which obtains half
the customers can drive the other entrepreneurial group or groups out of
the market.

Without effective competitors, in this view, occupants of the monopoly would

become inefficient suppliers of government services. Competitors are dis-
suaded by prohibitive expected costs and risk in attempting to acquire a
majority of the market. This is the same sort of political inefficiency dis-
cussed throughout the present paper. The more general question becomes:
what policy, including term limits, can address this underlying source of
political entry barriers?
Conventional policy options for dealing with natural monopoly are insuf-
ficient when the government itself is the monopoly. Laissez-faire amounts to
despotism and is undesirable. Regulation may work and seems sensible, but
would then create monopoly regulators. Finally, public ownership is redund-
ant and public operation resists clear definition (are political parties public?).
Tullock, in his hypothetical construction, instead suggests that competing
parties should regularly submit proposals to consumers (voters) for the rights
to the monopoly. To prevent myopic rule after the auction, bids would have
to include proposals for the mix and extent of government services, how pro-
duction methods would be improved, and how consumers would be treated. In
short, he suggests an institutional framework much like the U.S. Constitution.
This would be an improvement on laissez-faire regulation of government as
natural monopoly, but two problems still remain. First, consumers in this
political auction market have information deficiencies in evaluating altern-
ative bids. This is also true of economic markets, but Downsian ignorance
exacerbates the problem when voters decide an election.38 Second, aggreg-
ating preferences compounds the information problem and enables parties to

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 45

discount marginal consumer-voters through obfuscation in their bids. These

two problems cause the political inefficiency - in which government does not
pursue voter preferences but instead the preferences of those who operate the
government monopoly - to persist. Either transfers are suboptimal in quantity
or composition, or the costs of creating the optimal transfers is unnecessarily
high.
Considering Tullock's framing of the problem, his proposed policies, and
the inefficiencies that still remain, political entry barriers go much deeper
than simple matters of tenure. But the evidence surveyed by the present paper
suggests that the main vehicle of change from term limits will be reduced av-
erage tenure. Term limits are then too shallow to effect the kind of substantive
changes to political entry barriers needed to improve political efficiency.
Comparing Lott's and Tullock's arguments more closely suggests that
properly designed campaign finance reform may be a suitable substitute for
term limits. Lott (1986) explains that political entry barriers result not from
brand name capital, per se, but informational deficiencies that exist in all
markets. In his (1986: 91) view, economic markets are just better at redu-
cing information costs. This clearly reflects Tullock's argument based on
Downsian ignorance. Moreover, Tullock's (1965: 464) policy for reducing
entry barriers is to discipline incumbents with the threat of potential com-
petition. Lott discusses challenger investment overtime in much the same
manner as Tullock. And Tullock suggests shoring up the effectiveness of chal-
lengers by paying them, which is analytically equivalent to Lott's asymmetric
spending caps. While term limits may be incapable of effecting fundamental
change, properly designed campaign finance reform might. Moreover, the
likelihood that term limitation would affect political efficiency through its
effects on campaign expenditures suggests that it is at least partly a redundant
reform. This would be a fertile set of questions for the literature on campaign
finance and expenditures to take up.
The joint question of political and economic inefficiency is really one of
the second-best (Lancaster and Lipsey, 1956). Political markets apparently
are anti-competitive. A globally efficient improvement would be a natur-
ally competitive market. A second-best would be some form of institutional
intervention that would push the market to behave as if it were more compet-
itive. Economic monopolies are regulated by this underlying principle. But
as with economic regulation, it is healthy to question the original causes of
the monopoly power. The leading arguments for congressional term limits
have not done so. And as with economic regulation, it is necessary to es-
timate the effects of the intervention, and whether it reduces inefficiency.
The consequences of term limits, as laid out in the literature, are mixed in
answering this. It appears unlikely that term limits can reduce federal spend-

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
46

ing and recover economic losses. If so, a more general restriction - perhaps
appropriately designed campaign finance reform - is required to do so.

7. Conclusion

This paper has surveyed the causes and consequences of term limits an
tegrated the analysis into the broader economics literature in order to sp
the lessons offered from term limits research. The modern interest in term
limits has been based on high and increasing tenure in Congress. Tenure
was stable and low until the middle 19th century, but accelerated in the late
1890s because of increased political ambition, and again in the mid 1950s
and early 1970s because of increasing incumbent advantages. Term limits
advocates have proposed that tenure profiles in the modem congressional era
introduce inefficiencies of two sorts. First, the size of government (spending)
increases with tenure, creating additional deadweight losses in the economy.
Second, higher tenure enables incumbents to further insulate themselves from
competition by able challengers. As in economic markets, these barriers to
entry create associated political inefficiencies. Term limits would, by these
twin arguments, recover both economic and political efficiencies. Partly based
on these arguments, there was widespread popular support for term limits
during the 1990s, subsequent enactment in roughly half the state legislatures,
and enactment by 23 states on their congressional delegations. This support
and enactment was sometimes seemingly irrational, as when individual states
unilaterally restricted their own members, or when voters supported term
limits and simultaneously reelected their own members. In response came
several theories of term limits, typically with game theoretic underpinnings,
assigning rationality to certain forms of support for term limits. If rational,
then the consequences of term limits might be expected to be efficiency en-
hancing. This paper has identified these consequences: Term limits always
favor republicans because they lose reelection and win opens seats more often
than democrats do. Term limits also can be expected to increase the volatility
of fiscal policy and increase legislative shirking - not because term limit-
ation worsens a lame duck effect, but because it weakens electoral sorting
mechanisms. Term limits will almost certainly reduce average congressional
tenure, even before the term limit "binds". In this survey, however, I found
that in no way does this suggest that spending will also decline. Thus, the first
efficiency argument is not upheld. I found also that term limits will reduce the
value of the politician holding office, but the value of that office to outsiders,
namely special interests, is ambiguous. Insofar as term limits can reduce entry
barriers by reducing incumbent relative to challenger campaign spending, it
is a redundancy over properly designed campaign finance reform. Thus, I

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 47

have found no clear evidence supporting the second efficiency argument for
term limits either. To the extent that these inefficiencies are real, it remains a
problem for public choice research as to what kind of policies might enhance
efficiency. Properly designed campaign finance policy may be a solution, and
this suggests further study of the effects of alternative restrictions on electoral
outcomes.

Notes

1. For these claims see Jacob (1996), Bandow (1995, 1996), Ferry (1994a), Armor (1994)
and Coyne and Fund (1992). Garrett (1996) critiques the claims of creating a citizen
legislature.
2. Some 3,000 municipal offices are term limited, many since the original founding of their
jurisdictions. The 22nd Amendment to the Constitution, ratified in 1951, limits presid-
ential terms. Nearly three quarters of the states have had gubernatorial term limits laws
at various times, with one fourth of these enacting stricter limits in the 1992 and 1994
elections.

3. The seminal contributions are Olson (1965), Tullock (1967), and Stigler (1971). Sub-
sequent formal models were presented by Peltzman (1976), McCormick and Tollison
(1981), Becker (1983) and Denzau and Munger (1986). Implications and empirical re-
search within the model are discussed in Tollison (1981, 1988). Shughart (1990: Ch.
2) provides a succinct summary. Discussions of the interest group theory can be found
in prominent textbooks in public choice (Mueller, 1989), public finance (Rosen, 1998),
industrial organization (Carlton and Perloff, 1994), and principles (Ekelund and Tollison,
1998; Gwartney and Stroup, 1997).
4. It can take the form of social insurance (e.g., Feldstein, 1998), industry regulation (e.g.,
Stigler, 1971), antitrust (McChesney and Shughart, 1994), environmental protection (e.g.,
Buchanan and Tullock, 1975; Pashigan, 1985), the tax code (Yagi and Tachibinaki, 1998),
formulation of budget rules (Garrett, 1998), et cetera. Other forms of redistribution
through legislatures are discussed in Tollison (1988).
5. By contrast, office holders in other branches of their government such as the courts, the
assembly, and the treasury, were hardly ever held to a specific number of terms, and were
sometimes explicitly allowed unlimited terms.
6. The interested reader can trace the intellectual heritage of the American founding,
including the concept of rotation in office, in Bailyn (1967). See also Richard (1994).
7. The framers' original intent on term limits has been subject of a prolific debate among
legal scholars surrounding the 1995 Supreme Court decision in U.S. Term Limits v.
Thorton. The ruling struck down states' term limits laws as unconstitutional additional
requirements for holding office. As a result, this body of work focuses the question of ori-
ginal intent on the Qualifications Clauses, and offers little on what the founders expected
term limits to accomplish, much less why the framers excluded term limits from the final
document. (Price, 1996). It remains a very interesting and unanswered question.
8. McGuire and Ohsfeldt (1984, 1986, 1989) present econometric studies of delegate voting
and ratification of the constitution. Their thesis and methodology provide a model for
a rational choice approach to the issue of term limits. Londregan (1999) empirically
examines strategic voting at the federal convention of 1787.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
48

9. The incumbent reelection rate has exceeded 90% in all elections since 1990. Opinion polls
indicate super-majority support for term limits among voters (Moore, Saad, McAnemy
and Newport, 1994). And term limits initiatives/referenda passed 22 of 23 times in the
1990s, typically by margins of 2-to-i and as high as 3-to-i (Ferry, 1994b).
10. The seniority allocation of committee assignments has ebbed and flowed into and out
of prominence. In general, see Polsby, Gallaher, and Rundquist (1969). Early congresses
had only the Committee of the Whole. With the emergence first of the Ways and Means
committee and then others, the Speaker typically exercised complete discretion. As early
as the 1860s this discretion began to wane as seniority determined increasingly greater
numbers of posts. Kernell (1977) argues ideological differences arising within political
parties at the turn of the century left party leadership an ineffective means of allocation.
After the 1910 revolt against Speaker Cannon, seniority was the exclusive medium of al-
locating chairmanships (Polsby, 1968: 161), which extended to committee assignments in
general and persisted until 1974. Modem political science and economics has researched
the committee assignment process in detail. For contrasting views, see Grier and Mun-
ger (1991), Cox and McCubbins (1993), Krehbiel (1992), Coker and Crain (1994), and
Leighton and L6pez (2002).
11. The other statistics include longer stays prior to accession to Speaker of the House, less
horizontal movement into leadership positions, and fewer mean years between retirement
from the speakership and death, all at the end of the 19th century.
12. Three months after the signing of the Sherman Act, its chief proponent introduced tariff
legislation popularly known as "The Campaign Contributor's Tariff Bill". For a discus-
sion of antitrust and protectionism as substitute policies, see DiLorenzo (1985). See
Shughart, Silverman, and Tollison (1995) for regression analysis providing evidence of
a complementary relationship.
13. Kemell's discussion of rotation (1977: 675-679) is engaging. For example: "[R]otation
was not always based on geography. In some districts rotation appears simply as a con-
venient way of allowing aspiring politicians to gain experience and exposure for future
elections to state offices. In a few areas the rotation period was a strict "one term and
out. Oregon followed this prescription without exception until the twentieth century, and
Lincoln's career, as well as the nation's history, may have looked far different had local
rules not forced him out of the House after his first term". Space requirements disallow
recounting such interesting details, but Kemell's article is highly recommended to the
interested reader.

14. This is my own simplified interpretation of the model. I also use my own variable names
for consistency with later parts of this paper.
15. This is taken from Reed and Schansberg (1992), Equation 2.
16. This is a reproduction of Reed and Schansberg (1992) Table 2. For fuller explanation of
these values, see particularly the discussion surrounding their Equation 4 and their Table
3.

17. This constitutes a social loss only if those districts experiencing losses outweigh those
districts that experience gains.
18. Though Wittman (1989) makes a counter claim.
19. Note again the proximity of this idea to political shirking. Lott's model, which builds
on Peltzman's (1976) theory, is more involved than this simple abstract, particularly in
understanding the incumbent's return to holding office. But I cannot explain the details
of the model without going on for some length. The essential point made in the text is
adequate for current purposes.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 49

20. Though see Landes and Posner (1975) for an exception to this, saying the independent
judiciary establishes durability to wealth-transferring contracts even in the presence of a
Congress whose membership turns over.
21. For House legislative history see H.J.Res.2, H.J.Res.73, H.Rept. 104-67, H.Rept. 104-82,
104Cong.Rec. H3553 ff., H3885 ff., H.Rept. 105-2, H.Rept. 105-4, 105Cong.Rec.E69 ff.
For Senate legislative history, see S.J.Res.21, S.Rept. 104-158, 104Cong.Rec. S3769-86,
S3805-17, S3864-80.
22. "Full grandfathering" completely exempts extant members; they can serve as many terms
as they and their voters like. "Limited grandfathering" does not count the terms already
served by extant members toward the term limit. "No grandfathering" applies the terms
served toward the limit, so, for example, upon enactment of a six term limit, all members
with tenure greater than six terms would immediately be forced out of office. See Schans-
berg (1991) for a good discussion. On reentry, see Reed and Schansberg (1994: 81, fn.
6).
23. Compare Schansberg's (1994) work on the U.S. House to Francis and Baker (1986),
who offer more specific explanations for voluntary quits from state legislatures: ambition
for higher office, increasing career and familial opportunity costs, dissatisfaction, and
health/age.
24. Kenny and Rush (1990) find similar partisan bias on a similar vote when democrats
supported the 17th Amendment far more than republicans.
25. This is a simplified version of the model in Gilmour and Rothstein (1994). I have changed
their notation to be consistent with mine.

26. On these calculations, Reed and Schansberg conservatively assume leaders spend only
one term in a leadership position. In a comment on this paper, Jacobson (1995) notes that
this assumption means 70% of the congress will never reach a leadership position.
27. The tradeoff requires their be a sufficient ratio of higher to lower offices. The effect can be
reasonably expected in state houses because movement to state senates has many oppor-
tunities. Likewise, movement to the U.S. House, and further to the U.S. Senate has many
opportunities. But once in the U.S. Senate, for example, there are not so many higher
offices to pursue. These indirect effects probably would not obtain.
28. First among these normative views is the ideal of the "citizen legislator", in which Con-
gress is forced by term limitation to return to its pre-professional era and work toward
the public interest. Congress would be populated not with political careerists but with
ordinary citizens in the mold of Cincinnatus, answering the call to public duty, serving
a short period of time, and then returning to their private sector lives (Petracca, 1991;
Bandow, 1995; Coyne and Fund, 1992). A citizen legislature would engender policies
closer to ordinary citizens' interests and restore Congress to its founding role as the body
of deliberative governance (Will, 1992). Second, one can argue without eschewing polit-
ical ambition that term limits give politicians the incentive to work in the public interest.
Glazer and Wattenberg (1996) use a conventional Olson/Stigler model of representation
to show that without term limits the frequency of legislators proposing general interest
legislation is approximately one-third the social optimum. But with term limits, a member
must pursue higher office to remain in politics. He therefore must, while in current office,
work to establish a broader constituency, and will favor general rather than parochial
policies on the margin. Third - a variant of Payne's "culture of spending" argument -
high tenure in office creates "legislative contamination", which leaves politicians jaded
and apathetic to seeing bad policies passed (Cain, 1996: 25-26).
29. Although Garrett (1995) counters the citizen legislature point by arguing that initial in-
vestment in political capital so increases the value of remaining in public office, and,

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
50

indeed, that the only way to receive the return on such investment is to remain in public
life.

30. This argument is laid out in the context of the California legislature; I take full liberty
in extending the argument to Congress. Cain (1996: 29) argues that the effect depends
on whether the contributor is a moneyed interest group or a public sector/public interest
lobby group or policy institute.
31. Their measure is defined as: "total expenditures by all candidates in either the general or
primary elections" (Daniel and Lott, 1997: 174).
32. For a survey of the shirking literature see Bender and Lott (1996).
33. The authors compare these simulated congresses to actual data, and fail to reject that there
is no statistical difference "In order to give an idea of how the model in general performed,
the mean probability of continuing in office is 0.844 for the twelve simulated congresses,
which compares favorably to the mean probability of 0.839 for the twelve congresses that
comprise our data sample". (Bender, Haas, and Kim, 2000: 31).
34. The mean incumbent taste for shirking is .21, while the mean for challengers is .26. Over
time, because the limit is lifetime and/or because of sorting, the congress' mean taste for
shirking will be upper-bounded by the mean of the challenger distribution. Therefore, the
estimated value for shirking taste does increase over time. It just does not increase more
or less under term limits.

35. When the simulations are run again assuming uniform distribution of challengers' pref-
erences for shirking, the effects are: tastes for shirking increase by 115 and 100%, and
shirking increases by 440 and 640%, respectively under 3- and 6-term limits. Both of
these changes are results of an earlier assumption about the challengers' shirking pref-
erences. By construction of the model, the distribution of preferences for shirking in the
challenger population is not affected by term limits. This is a strong assumption because
term limits will reduce the returns to challengers with low taste for shirking relative to
challengers with high taste for shirking. It is this very assumption that biases downward
their calculated degree of shirking under term limits, and what makes the increase in
shirking under term limits such a strong result.
36. Their measure for "professionalism" is a continuous { 0,1 } index based on legislative
compensation, expenditures on staff, and length of legislative session, relative to the U.S.
Congress. Full explanation can be found in their paper.
37. While revising this paper I received a preliminary draft of a study that examines the effects
of PAC contribution limits on electoral competitiveness across the state legislatures and
over time. The early results indicate that the presence of PAC limit lowers incumbent vote
share and margin of victory in state legislature races (Stratmann and Aparicio-Castillo,
2001).
38. It is easier and more important to evaluate one's own basket of goods and alternative
suppliers than the array of political goods, of which one consumes only a fraction, as-
sembled in party platforms. Additional uncertainty arises because economic bundles are
typically evaluated retrospectively (the goods have already been produced), whereas party
platforms must be evaluated prospectively (promises about future policies).

References

Adams, J.D. and Kenny, L.W. (1986). Optimal tenire of elected public officials. Journal of
Law and Economics 109: 303-328.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 51

Aka, A., Reed, W.R., Schansberg, D.E. and Zhen Zhu (1996). Is there a 'culture of spending'
in Congress? Economics and Politics 8: 191-212.
Anderson, T.L. and Hill, P.J. (1980). The birth of a transfer society. Stanford: Hoover
Institution Press.

Armor, J.C. (1994). 'Forshadowing' effects of term limits: California's example for Congress.
Term Limits Outlook Series, U.S. Term Limits Foundation, 3:1 (June).
Bailyn, B. (1967). The ideological origins of the American Revolution. Cambridge, MA: The
Belknap Press of Harvard University Press.
Bandow, D. (1995). Real term limits: Now more than ever. Policy Analysis, Cato Institute, No.
221 (6 April).
Bandow, D. (1996). The political revolution that wasn't: Why term limits are needed now
more than ever. Policy Analysis, Cato Institute, No. 259 (5 September).
Becker, G.S. (1983). A theory of competition among pressure groups for political influence.
Quarterly Journal of Economics 68: 371-400.
Bender, B., Haas, T.C. and Sunwoong Kim (2000). Sorting, shirking, and term limits.
Unpublished manuscript. University of Wisconsin-Milwaukee.
Bender, B. and Lott, J.R. (1996). Legislator voting and shirking: A critical review of the
literature. Public Choice 87: 67-100.
Bernhardt, M.D. and Ingberman, D.E. (1985). Candidate reputations and the 'incumbency
effect'. Journal of Public Economics 27: 47-67.
Besley, T. and Case, A. (1995). Does electoral accountability affect economic policy choices?:
Evidence from gubernatorial term limits. Quarterly Journal of Economics 60: 769-798.
Borcherding, T. (Ed.). (1977). Budgets and bureaucrats: The sources of government growth.
Duke University Press.
Browning, E.K. (1974). On the welfare cost of transfers. Kyklos 27: 274-377.
Browning, E.K. (1987). On the marginal welfare cost of taxation. American Economic Review
77: 11-23.

Browning, E.K. (1993). The marginal cost of redistribution. Public Finance Quarterly 21:
3-32. Buchanan, J.M. and Congleton, R.D. (1994). The incumbency dilemma and rent
extraction by legislators. Public Choice 79: 47-60.
Buchanan, J.M. and Tullock, G. (1975). Polluters' profits and political response: Direc
controls versus taxes. American Economic Review 65: 139-147.

Cain, B. (1996). The varying impact of term limits on legislative behavior and electora
sponsiveness. In B. Grofman (Ed.), Legislative term limits: Public choice perspectiv
Boston: Kluwer Academic Publishers.
Capell, E.A. (1996). The impact of term limits on the California legislature: An interest group
perspective. In B. Grofman (Ed.), Legislative term limits: Public choice perspectives.
Boston: Kluwer Academic Publishers.
Carlton, D.W. and Perloff, J.M. (1994). Modern industrial organization. 2nd ed. New York
Harper Collins.
Cohen, L. and Spitzer, M. (1996). Term limits and representation. In B. Grofman (Ed.)
Legislative term limits: Public choice perspectives, 47-65. Boston: Kluwer Academic
Publishers.
Coker, D.C. and Crain, W.M. (1994). Legislative committees as loyalty generating institutions.
Public Choice 81: 195-221.
Cox, G.W. and McCubbins, M.D. (1993). Legislative Leviathan: Party government in t
House. Berkeley: University of California Press.
Coyne, J.K. and Fund, J. (1992). Cleaning House: America's campaign for term limi
Washington, DC: Regnery Gateway.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
52

Crain, W.M. and Johnson, J.M. (2001). Effects of term limits on fiscal performance: Evid-
ence from democratic nations. Working paper, presented at 2000 Public Choice Society
Meetings.
Crain, W.M., Leavens, D.R. and Tollison, R.D. (1988). Laissez-faire in campaign finance.
Public Choice 56: 201-212.

Crain, W.M. and Oakley, L.K. (1995). The politics of infrastructure. Journal of Law an
Economics 38: 1-17.
Crain, W.M. and Tollison, R.D. (1993). Time inconsistency and fiscal policy: Empirica
analysis of the U.S. States, 1969-89. Journal of Public Economics 51: 153-159.
Daniel, K. and Lott, J.R. Jr. (1997). Term limits and electoral competitiveness: Evidence from
California's State Legislative races. Public Choice 90: 165-184.
Danielson, A.L. and Rubin, P.H. (1977). An empirical investigation of voting on energy issues.
Public Choice 31: 121-128.

Denzau, A.R. and Munger, M.C. (1986). Legislators and interest groups: How unorganize
interest get represented. American Political Science Review 80: 89-106.
Dick, A.R. and Lott, J.R. Jr. (1993). Reconciling voters' behavior with legislative term limi
Journal of Public Economics 50: 1-14.
Donovan, T. and Snipp, J.R. (1994). Support for legislative term limitations in Californi
Group representation, partisanship, and campaign information. Journal of Politics 56
492-501.
Ekelund, R.E., McDonald, M.J. and Tollison, R.D. (1994). Business restraints and the Clayton
Act of 1914: Public- or private-interest legislation? In ES. McChesney and W.F. Shughart
(Eds.), The causes and consequences of antitrust: The public choice perspective. Chicago:
University of Chicago Press.
Ekelund, R.E. Jr. and Tollison, R.D. (1998). Economics, 4th ed. New York: Harper Collins.
Erickson, S.C. (1995). The entrenching of incumbency: Reelections in the U.S. House o
Representatives, 1790-1994. Cato Journal 14: 397-420.
Feldstein, M. (Ed.). (1998). Privatizing Social Security. Chicago and London: University of
Chicago Press.
Ferejohn, J. (1986). Incumbent performance and electoral control. Public Choice 50: 5-25.
Ferry, J. (1994a). Women, minorities and term limits: America's path to a representative
Congress. Term Limits Outlook Series, U.S. Term Limits Foundation, 3:2 (July).
Ferry, J. (1994b). Coming to terms with term limits: A summary of state term limit laws. Term
Limits Outlook Series, U.S. Term Limits Foundation, 3:4 (December).
Francis, W.L. and Baker, J.R. (1986). Why do U.S. State legislators vacate their seats
Legislative Studies Quarterly 11: 119-126.
Francis, W.L. and Kenny, L.W. (1997). Equilibrium projections of the consequences of term
limits upon expected tenure, institutional turnover, and membership experience. Journal
of Politics 59: 240-252.
Francis, W.L., Kenny, L.W. and Anderson, B. (2000). The churning of membership in stat
legislatures: Effects of term limits and anticipatory behavior. Unpublished manuscript.
University of Florida.
Franklin, D. and Westin, T. (1998). Predicting the institutional effects of term limits. Public
Choice 96: 381-393.

Friedman, D. and Wittman, D. (1995). Why voters vote for incumbents but against incum
bency: A rational choice explanation. Journal of Public Economics 57: 67-83.
Friedman, D. and Wittman, D. (1996). Term limits as political redistribution. In B. Grof
man (Ed.), Legislative term limits: Public choice perspectives. Boston: Kluwer Academi
Publishers.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 53

Garrett, E. (1996). Term limitation and the myth of the citizen-legislator. Cornell Law Review
81: 623-697.

Garrett, E. (1998). Harnessing politics: The dynamics of offset requirements in the tax
legislative process. University of Chicago Law Review 65: 501-569.
Gerber, E.R. and Lupia, A. (1996). Term limits, responsiveness and the failures of increased
competition. In B. Grofman (Ed.), Legislative term limits: Public choice perspectives.
Boston: Kluwer Academic Publishers.

Gilmour, J.B. and Rothstein, P. (1994). Term limitation in a dynamic model of parti
balance. American Journal of Political Science 38: 770-796.
Gilmour, J.B. and Rothstein, P. (1996). A dynamic model of loss, retirement, and tenure i
U.S. House of Representatives. Journal of Politics 58: 54-68.
Glaeser, E.L. (1997). Self-imposed term limits. Public Choice 93: 389-394.
Glazer, A. and Grofman, B. (1987). Two plus two equals six: Tenure in office of Senators
Representatives, 1953-1983. Legislative Studies Quarterly 12: 555-563.
Glazer, A. and Wattenberg, M. (1996). Promoting legislative work: A case for term lim
In B. Grofman (Ed.), Legislative term limits: Public choice perspectives. Boston: Klu
Academic Publishers.

Grier, K.B. and Munger, M.C. (1991). Committee assignments, constituent preferences, and
campaign contributions. Economic Inquiry 29: 24-43.
Grofman, B. (Ed.). (1996a). Legislative term limits: Public choice perspectives. Boston:
Kluwer Academic Publishers.
Grofman, B. (1996b). Introduction to the term limits debate: Hypotheses in search of data.
In B. Grofman (Ed.), Legislative term limits: Public choice perspectives. Boston: Kluwer
Academic Publishers.

Grofman, B. and Sutherland, N. (1996). The effect of term limits when competition is endo-
genized. In B. Grofman (Ed.), Legislative term limits: Public choice perspectives. Boston:
Kluwer Academic Publishers.

Gwartney, J.D. and Stroup, R.L. (1997). Microeconomics: Private and public choice. 8th ed.,
United States: Dryden Press.
Holcombe, R.G. (1999). Veterans interests and the transition to government growth: 1870-
1915. Public Choice 99: 311-326.
Hopenhayn, H.A. and Muniaguerria, M.E. (1996). Policy variability and economic growt
Review of Economic Studies 63: 611-625.
Jacob, P. (1996). Whose government is it, anyway? U.S. Term Limits Foundation. Mimeo.
Jacobson, G.C. (1995). The House under term limits: A comment. Social Science Quarter
76: 720-724.

Kenny, L. and Rush, M. (1990). Self-interest and the Senate vote on direct elec
Economics and Politics 2: 291-302.
Kernell, S. (1977). Toward understanding 19th century congressional careers: Ambition
competition, and rotation. American Journal of Political Science 21: 669-693.
Konrad, K.A. and Torsvik, G. (1997). Dynamic incentives and term limits in bureaucrac
regulation. European Journal of Political Economy 13: 267-279.
Krehbiel, K. (1992). Information and legislative organization. Ann Arbor: University
Michigan Press.
Lancaster, K. and Lipsey, R.G. (1956). The general theory of the second best. Review
Economic Studies 24: 11-32.
Landes, W. and Posner, R.A. (1975). The independent judiciary in an interest-group perspect
ive. Journal of Law and Economics 18: 875-901.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
54

Leighton, W.A. and L6pez, E.J. (1999). Committee assignments and legislative loyalty.
Political Research Quarterly 55: forthcoming.
Londregan, J. (1999). Deliberation and voting at the Federal Convention of 1787. Unpublished
manuscript. UCLA.
L6pez, E.J. (2002). Congressional voting on term limits. Public Choice, forthcoming.
Lott, J.R. Jr. (1986). Brand names and barriers to entry in political markets. Public Choice 51:
87-92.
Lott, J.R. Jr. (1987a). Political cheating. Public Choice 53: 169-186.
Lott, J.R. Jr. (1987b). The effect of nontransferable property rights on the efficiency of political
markets: Some evidence. Journal of Public Economics 32: 231-246.
Lott, J.R. Jr. (1989). Explaining challengers' campaign expenditures: The importance of sunk
nontransferable brand name. Public Finance Quarterly 17: 108-118.
Lucas, R. (1976). Econometric policy evaluation: A critique. Journal of Monetary Economics
1: 19-46.

McChesney, F.S. (1991). Rent extraction and interest-group organization in a Coasean model
of regulation. Journal of Legal Studies 20: 73-90.
McChesney, F.S. and Shughart, W.F. II. (Eds.) (1995). The causes and consequences of
antitrust: The public choice perspective. Chicago: University of Chicago Press.
McCormick, R.E. and Tollison, R.D. (1981). Politicians legislation and the economy: An
inquiry into the interest-group theory of government. Boston: Martinus-Nijhoff.
McGuire, R.A. and Ohsfeldt, R.L. (1984). Economic interests and the American Constitution:
A quantitative rehabilitation of Charles A. Beard. Journal of Economic History 44: 509-
519.

McGuire, R.A. and Ohsfeldt, R.L. (1986). An economic model of voting behavior over specific
issues at the Constitutional Convention of 1787. Journal of Economic History 46: 79-111.
McGuire, R.A. and Ohsfeldt, R.L. (1989). Self-interest, agency theory, and political voting
behavior: The ratification of the United States Constitution. American Economic Review
79: 219-234.

Mixon, F.G. (1994). Constitutional and economic determinants of lobbyist activity in

resentative democracy. Rivista Internazionale di Scienze Economiche e Commercia
1043-1051.

Moncrief, G., Thompson, I., Haddon, M. and Hoyer, E. (1992). For whom the bell tolls: T
limits and state legislatures. Legislative Studies Quarterly 17: 37-47.
Moore, D., Saad, L., McAneny, L. and Newport, F. (1994). Contract with America: A G
Poll special report. The Gallup Poll Monthly 19-34 (November).
Moore, M.K. and Hibbing, J.R. (1996). Length of congressional tenure and federal spendi
Were the voters of Washington State correct? American Politics Quarterly 24: 131-1
Mueller, D.C. (1989). Public choice II. Cambridge, MA: Cambridge University Press.
Oakley, L.M. (1994). An empirical examination of direct democracy. Doctoral Disserta
Department of Economics, George Mason University, Fall.
Olson, M. (1965). The logic of collective action. Cambridge, MA: Harvard University P
Owings, S. and Borck, R. (2000). Legislative professionalism and government spendin
citizen legislators really spend less? Public Finance Review forthcoming.
Pashigian, B.P. (1985). Environmental regulation: Whose self-interests are being prote
Economic Inquiry 23: 551-584.
Payne, J.L. (1991). The culture of spending. San Francisco: ICS Press.
Peltzman, S. (1976). Toward a more general theory of regulation. Journal of Law
Economics 19: 211-240.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
 55

Petracca, M. (1991). The poison of professional politics. Policy Analysis, Cato Institute, No.
151, 10 May.
Polsby, N.W. (1968). The institutionalization of the U.S. House of Representatives. American
Political Science Review 63: 144-168.
Polsby, N.W., Gallaher, M. and Rundquist, B.S. (1969). The growth of seniority in the U.S
House of Representatives. American Political Science Review 63: 798-802.
Posner, R.A. (1975). The social costs of monopoly and regulation. Journal of Politica
Economy 83: 807-827.
Price, H.D. (1975). Congress and the evolution of legislative 'professionalism'. In N. Ornstei
(Ed.), Congress in change. New York: Praeger.
Price, P.J. (1996). Term limits on original intent?: An essay on legal debate and historica
understanding. Virginia Law Review 82: 493-533.
Ramirez, C. and Eigen-Zucchi, C. (2001). Understanding the Clayton Act of 1914: An analysis
of the interest group hypothesis. Public Choice, forthcoming.
Reed, W.R. Calculating the effect of term limits on federal spending is harder than it looks
Unpublished manuscript. Presented at 1996 Public Choice Society Meetings, Houston,
Texas.

Reed, W.R. and Schansberg, D.E. (1990). How long do congressmen stay in office? Economics
and Politics 2: 173-191.
Reed, W.R. and Schansberg, D.E. (1992). The behavior of congressional tenure over time:
1953-1991. Public Choice 73: 183-203.
Reed, W.R. and Schansberg, D.E. (1994). An analysis of the impact of congressional te
limits. Economic Inquiry 32: 79-91.
Reed, W.R. and Schansberg, D.E. (1995). The House under term limits: What would it
like? Social Science Quarterly 76: 699-716.
Reed, W.R. and Schansberg, D.E. (1996). An analysis of the impact of congressional te
limits on turnover and party balance. In B. Grofman (Ed.), Legislative term limits: Pub
choice perspectives. Boston: Kluwer Academic Publishers.
Reed, R.D., Schansberg, E.D., Wilbanks, J. and Zhen Zhu (1998). The relationship betw
congressional spending and tenure with an application to term limits. Public Choice
85-104.
Regens, J.L., Elliott, E. and Gaddie, R.K. (1991). Regulatory costs, committee jurisdictions,
and corporate PAC contributions. Social Science Quarterly 72: 751-760.
Rosen, H. (1994). Public finance. 5th ed. United States: Irwin-McGraw-Hill.
Rusk, J.F. (1970). The effect of the Australian ballot reform in split ticket voting: 1876-1908.
American Political Science Review 64: 1220-1238.
Schansberg, D.E. (1991). An analysis of tenure and turnovers in the congressional labo
market. Doctoral Dissertation. Department of Economics, Texas A&M University.
Schansberg, D.E. (1994). Moving out of the House: An analysis of congressional quits
Economic Inquiry 32: 445-456.
Scully, G.W. (1995). Congressional tenure: Myth and reality. Public Choice 83: 203-219.
Shughart, W.F. II. 1990. Antitrust Policy and Interest-Group Politics. Westport, CT: Quorum
Books.

Shughart, W.E. II., Silverman, D. and Tollison, R.D. (1994). Antitrust enforcement and foreign
competition. In ES. McChesney and F Shughart II (Eds.), The causes and consequences
of antitrust: The public choice perspective. Chicago: University of Chicago Press.
Snyder, J.M. Jr. (1990). Campaign contributions as investments: The U.S. House of Repres-
entatives, 1980-1986. Journal of Political Economy 98: 1195-1227.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms
56

Snyder, J.M. Jr. (1992). Long-term investing in politicians; or, give early, give often. Journal
of Law and Economics 35: 15-44.
Sobel, R.S. and Wagner, G.A. (1998). On the determinants of congressional bill sponsorship
and voting behavior with an application to government growth theories. Journal of Public
Finance and Public Choice 42: 2-3.
Stigler, G.J. (1969). The origins of the Sherman Act. Journal of Legal Studies 14: 1-12.
Stigler, G.J. (1971). The theory of economic regulation. Bell Journal of Economics an
Management Science 2: 1-21.
Stratman, T. (1992). The effects of logrolling on congressional voting. American Economi
Review 82: 1162-1176.

Stratmann, T. (1995). Campaign contributions and congressional voting: Does the timing o
contributions matter? Review of Economics and Statistics 77: 127-136.
Stratman, T. and Aparicio-Castillo, F.J. (2001). Effects of campaign contribution limits on
elections: Evidence from the states. Unpublished manuscript. George Mason University.
Tabarrok, A. (1994). A survey, critique, and new defense of term limits. Cato Journal 14:
333-350.
Tabarrok, A. (1996). Term limits and political conflict. In B. Grofman (Ed.), Legislative term
limits: Public choice perspective, 237-244. Boston: Kluwer Academic Publishers.
Thomas, C. (1995). Dissent. U.S. Term Limits, Inc. v. Thorton. United States Reports, Nos.
93-1456 and 93-1828.
Tollison, R.D. (1981). Rent seeking: A survey. Kyklos 35: 575-602.
Tollison, R.D. (1988). Public choice and legislation. Virginia Law Review 74: 339-371.
Tullock, G. (1975). The transitional gains trap. Bell Journal of Economics 6: 671-618.
Tullock, G. (1965). Entry barriers in politics. American Economic Review Papers an
Proceedings 55: 458-466.
Tullock, G. (1967). The welfare costs of tariffs, monopolies and theft. Western Econom
Journal 5: 224-232.

Weingast, B.R. (1984). The congressional-bureaucratic system: A principal-agent perspective.

Public Choice 44: 147-191.

Weingast, B.R. and Marshall, W.J. (1988). The industrial organization of Congress; or, w
legislatures, like firms, are not organized as markets. Journal of Political Economy
132-163.

Weingast, B. and Moran, M. (1983). Bureaucratic discretion of congressional oversight?: Reg-

ulatory policy-making by the Federal Trade Commission. Journal of Political Economy
91: 765-800.

Will, G.F. (1992). Restoration: Congress, term limits, and the recovery of deliberati
democracy. New York: Free Press.
Wittman, D. (1989). Why democracies produce efficient results. Journal of Political Econom
97: 1395-1424.

Yagi, T. and Tachibanaki, T. (1998). Income redistribution through the tax system:
simulation analysis of tax reform. Review of Income and Wealth 44: 397-415.
Zubler, T.C. (1995). Federal preclusion of state-imposed congressional term limits. Harva
Journal of Law and Public Policy 19: 174-188.

This content downloaded from

176.237.222.111 on Sun, 09 Oct 2022 07:53:33 UTC
All use subject to https://about.jstor.org/terms