Patronage and Elections in U.S. States
1
Olle Folke
School of International and Public Affairs
Columbia University
IFN
Shigeo Hirano
Department of Political Science
Columbia University
James M. Snyder, Jr.
Department of Government
Harvard University
NBER
September, 2010
1
We thank Don Green, Ken Shepsle, and seminar participants at MIT and Stockholm School of
Economics for their helpful comments.
Abstract
This paper examines whether control of patronage jobs significantly increased a political
party’s probability of winning elections in U.S. states. We employ a differences-in-differencs
design, exploiting the fact that there is considerable variation in the dates different states
adopted civil service reforms. We find evidence consistent with the hypothesis that political
parties in U.S. states were able to use patronage to increase the probability of maintaining
control of their state legislatures and statewide executive offices. We also find evidence that
an “entrenched” party in power for a longer time period can use patronage more effectively
than a non-entrenched party. We consider several alternative hypotheses that might plausibly
account for the patterns in the data, but find no evidence to support them.
1. Introduction
Does the control of patronage significantly increase a political party’s probability of
winning elections? Perhaps surprisingly, at least for the U.S. we do not know the answer to
this question.
Theoretically, it seems hard to believe that the answer could be anything but “Yes.”
Patronage jobs constitute a valuable resource for the party in power. As Pollock (1937, p. 32)
notes: “Employees who are politically appointed are naturally expected to attend political
meetings, make speeches, canvass voters, and do all the other things involved in political
activity.” Parties also use patronage for “raising campaign funds through the assessment
of public employees. ‘Two per cent clubs’ manage to squeeze large sums of money out of
politically appointed employees to help defray campaign costs. The public pay roll is thus
used in an indirect way to pay party expenses.” Maranto and Johnson (2007, p. 79) states
the claim succinctly: “Before the 1950s, U.S. political appointments were primarily to help
the incumbent party win elections as ‘spoils’ patronage.”
1
On the other hand, electoral considerations are not necessarily the primary factor affect-
ing party leaders’ decisions regarding how to distribute patronage (e.g. Sorauf, 1959; Wilson,
1961; Johnston, 1979). Many patronage jobs may be used to maintain party organizations
or be consumed as rents. Party bosses must often provide loyal lieutenants or key factional
leaders with patronage, even if this is not the most efficient allocation from a vote maxi-
mization perspective, in order to maintain their position as bosses. Surveys of patronage
recipients often find that they were not particularly active in politics either before or after
receiving their jobs (e.g. Sorauf, 1956; Johnston, 1979).
2
Furthermore, by not allocating
jobs to maximize votes, patronage could even have a negative effect on a party’s electoral
support by alienating those who do not receive jobs.
3
Pollock (1937, p. 30) writes that
1
In a similar but less critical vein, Key (1964) notes: “The patronage system may be considered, too,
as a method of financing party activity. The operation of a party organization requires the services of
many men and women... Though much of this work is performed by unpaid volunteers, their efforts are not
adequate. Indirectly, a considerable part of party expense is met by the public treasury, and the chief means
of channeling public funds to party support is through the appointment of party workers to public office.”
2
In Sorauf’s (1956) study of Centre County, Pennsylvania, only 43% claimed to have engaged in any
campaign work for the party.
3
In his study of New Haven, Johnston (1979) found that patronage was severely miss-allocated from a
vote-maximizing point of view.
2
“for every appointment which is made, perhaps a dozen disappointed persons are made to
grumble – the one appointed in many cases becoming an ingrate.”
4
This paper provides evidence that patronage does – or did – help U.S. parties in power
retain power. To our knowledge, this is the first paper to do so. We study the U.S. states over
a one-hundred-and-ten year period, 1885-1995. Within this period there was large variation
in the timing of civil service reform across states. Illinois and Wisconsin adopted general
civil service laws in 1905, while Mississippi and Montana did not do so until 1976. The
period studied therefore includes about 20 years before 1905 and 20 years after 1976.
5
We
exploit this variation to estimate the degree to which control over patronage jobs affected a
political party’s probability of winning future elections.
In the first set of empirical analyses in this paper, we ask two simple questions: If a party
wins control of the state legislature at time t, does that party have a higher probability of
maintaining control of the legislature over the next few elections, at times t+1, t+2, etc.?
Second, and more importantly, is the probability of maintaining control of the legislature
over the next few elections even higher under a patronage system then it is after the adoption
of civil service reforms? We also study elections to statewide executive offices at t+1, t+2,
etc., in addition to control of the state legislature.
Our estimates indicate that the answer to both questions is “Yes,” and that the effects
of patronage are large. Consider a party that barely wins control of the state legislature
at time t. The probability that this party continues to control the legislature over the next
four elections is about 25 percentage points higher under a patronage system than under a
civil service system. We find a similarly large difference in the probability of winning future
4
More recently, Trounstine (n.d.) found that, conditional on running for re-election, mayors and city
council members in her sample were less likely to win in cities with large municipal workforces. As she
notes, in most cities most employees were covered by civil service law or union contracts. Restricting the
attention to cities where municipal employees were not unionized (in 1986) reversed the sign of the coefficient,
but it is statistically insignificant.
5
New York and Massachusetts adopted civil service laws very early on – in 1883 and 1885 respectively
– around the time the federal government adopted the Pendleton Act. West Virginia was the last state to
adopt a general civil service law, in 1989. Since these were relatively large outliers, we do not try to include
long pre-reform or post-reform periods for these states. Attempting to include a long pre-reform period for
New York and Massachusetts would be especially difficult, since it would take us into Reconstruction, and
also well into the era when most states still used party ballots rather than a secret ballot. These likely had
significant effects of their own on incumbent parties, and incorporating them would significantly complicate
the analysis.
3
elections for statewide executive offices. Overall, the evidence is strongly consistent with
the hypothesis that political parties in U.S. states were able to use patronage to increase
the probability of maintaining control of their state legislatures and also to increase the
probability of winning other statewide offices.
In the second part of the paper, we examine whether parties that control the legislature
for several years are better able to take advantage of patronage. We refer to these parties
as entrenched parties. Scholars have speculated that entrenched parties have an advantage
through the patronage system. For example, Sorauf (1959, p. 118) writes:
[J]ust as it takes money to make money, it takes political power to achieve greater
power. The party long out of office and desperately in need of new reservoirs of
strength is precisely the party that, should it suddenly find itself in office, would
be least able to use patronage for rebuilding. Weak parties lack the discipline,
the trained leadership, and the surplus of potential jobholders to use the system
to their maximum advantage.
However, there is little systematic evidence that experience controlling the government con-
fers an additional electoral advantage.
Thus, we ask another simple question: Is the higher probability of maintaining control of
the legislature over the next few elections we observe for winning control of the legislature
at time t under patronage even higher if the winning party also controlled the legislature for
several years prior to t?
Our estimates indicate that entrenched parties have an electoral advantage under the
patronage system relative to non-entrenched parties. Under a patronage system, the proba-
bility an entrenched incumbent maintains control over the subsequent elections is more than
30 percentage points higher than for a non-entrenched party. After the introduction of civil
service, we find little evidence that entrenched parties maintained their electoral advantage.
We also provide some evidence that this entrenched party advantage under patronage
may be related to entrenched parties’ ability to win “very close” elections. Prior to the
general civil service reforms, entrenched parties won about three out of four close elections.
Under civil service the entrenched parties did not have any advantage in winning these close
4
