Content uploaded by Catherine Moury
Author content
All content in this area was uploaded by Catherine Moury on Oct 25, 2017
Content may be subject to copyright.
The Fulfillment of Parties’ Election Pledges:
AComparativeStudyontheImpactofPower
Sharing
Robert Thomson
Monash University
Terry Royed
University of Alabama
Elin Naurin
University of Gothenburg
Joaqu´
ın Art ´
es
Universidad Complutense de Madrid
Rory Costello
University of Limerick
Laurenz Ennser-Jedenastik
University of Vienna
Mark Ferguson
Bennett College
Petia Kostadinova
University of Illinois at Chicago
Catherine Moury
Universidade Nova de Lisboa
Franc¸ois P´
etry
Universit´
eLaval
Katrin Praprotnik
University of Hamburg
Abstract: Why are some parties more likely than others to keep the promises they made during previous election campaigns?
This study provides the first large-scale comparative analysis of pledge fulfillment with common definitions. We study the
fulfillment of over 20,000 pledges made in 57 election campaigns in 12 countries, and our findings challenge the common
view of parties as promise breakers. Many parties that enter government executives are highly likely to fulfill their pledges,
and significantly more so than parties that do not enter government executives. We explain variation in the fulfillment of
governing parties’ pledges by the extent to which parties share power in government. Parties in single-party executives, both
Robert Thomson is Professor, School of Social Sciences, Monash University, Melbourne, Victoria 3800, Australia (robert.
thomson@monash.edu). Terry Royed is Associate Professor, Department of Political Science, University of Alabama, Box 870213,
Tuscaloosa, AL 35487, USA (troyed@ua.edu). Elin Naurin is Associate Professor, Department of Political Science, University of
Gothenburg, Box 711, 405 30 Gothenburg, Sweden (elin.naurin@pol.gu.se). Joaqu´
ın Art´
es is Associate Professor, Departmento de
Econom´
ıa Aplicada, Facultad de Derecho,Universidad Complutense de Madrid, Av. Complutense S/N, Ciudad Universitaria, 28040 Madrid,
Spain (jartes@ucm.es). Rory Costello is Lecturer, Department of Politics and Public Administration, University of Limerick, Limerick,
Ireland (rory.costello@ul.ie). Laurenz Ennser-Jedenastik is Assistant Professor, Department of Government, University of Vienna, Roo-
seveltplatz 3, 1090 Vienna, Austria (laurenz.ennser@univie.ac.at). Mark Ferguson is Assistant Professor, 218 Pfeiffer Science Hall, Ben-
nett College, Greensboro, NC 27401, USA (mjferguson@crimson.ua.edu). Petia Kostadinova is Assistant Professor, Department of
Political Science, University of Illi nois at Chicago, 1102 Behavioral Science Building, 1007 W. Harrison Street, Chicago, IL 60607-
7139, USA (pkostad@uic.edu). Catherine Moury is Senior Lecturer, IPRI and Faculdade de Ci ˆ
encias Sociais e Humanas, Universi-
dade Nova de Lisboa (FCSH-NOVA), Av. De Berna, 26-C, 1069-061, Lisbon, Portugal (catherine.moury@fcsh.unl.pt). Franc¸ois P´
etry
is Adjunct Professor, Department of Political Science, Universit´
e Laval, 1030, av. Des Siences-Humaines, Qu´
ebec, G1V OA6, Canada
(francois.petry@pol.ulaval.ca). Katrin Praprotnik is Postdoctor al Researcher, Institute for PoliticalScience, University of Hamburg, Allende-
Platz 1, 20146 Hamburg, Germany (katrin.praprotnik@uni-hamburg.de).
We are grateful to all colleagues who offered constructive criticism and encouragement on this project and our individual country studies.
We received particularly influential advice on this comparative study from Shaun Bowler, Michael Huelshoff, Michael McDonald, Bing
Powell, Su san Stokes, and Guy Whitten. We thank the editor of AJPS and four anonymous referees for their insightful guidance throughout
the review process. We thank Lucy Mansergh for making available most of our data on Ireland, and Stephen A. Borrelli and J. Norman
Baldwin for assisting in data collection for the United States. We acknowledge financial support from the following sources: the Austrian
Science Fund (FWF), Grants S10903-G11 and S10903-G08; the Fonds de recherche du Qu ´
ebec-Soci´
et´
e et culture (FRQSC Grant nr.
SE171427); FCT-Fundac¸ ˜
ao para Ciˆ
encia e a Tecnologia (PTDC/CPJ-CPO/111915/2009; PI: Catherine Moury); the German Research
Foundation (DFG), project “Political Economy of Reforms” (SFB 884: C1); the Spanish Ministry of Innovation (Grant CSO2013-40870-
R); the Swedish Research Council (421-2007-7276), Riksbankens Jubileumsfond (RRD10-1418:1), the COFAS Marie Curie Fellowship
Programme (2011-0106), and the Multidisciplinary Opinion and Democracy Research Group at the University of Gothenburg; and the
U.S. National Scie nce Found ation (Grant No. SBR-97 30785).
American Journal of Political Science,Vol. 00, No. 0, xxxx 2017, Pp. 1–16
C⃝2017, Midwest Political Science Association DOI: 10.1111/ajps.12313
1
2ROBERT THOMSON ET AL.
with and without legislative majorities, have the highest fulfillment rates. Within coalition governments, the likelihood of
pledge fulfillment is highest when the party receives the chief executive post and when another governing party made a
similar pledge.
Replication Materials: The data, code, and any additional materials required to replicate all analyses in this arti-
cle are available on the American Journal of Political Science Dataverse within the Harvard Dataverse Network, at:
https://doi.org/10.7910/DVN/YJUIBI.
The fulfillment of election pledges is highly rel-
evant to the theory and practice of representa-
tive democracy. If parties channel societal de-
mands into government policies effectively, there should
be a substantial level of congruence between the pol-
icy content of their election programs or manifestos and
subsequent government policies. A strong program-to-
policy linkage is central to the mandate theory of democ-
racy and the responsible party model (Downs 1957;
Klingemann, Hofferbert, and Budge 1994; McDonald and
Budge 2005; Powell 2000).
For Mansbridge (2003, 515), “the idea that dur-
ing campaigns representatives made promises to con-
stituents, which they then kept or failed to keep” is the
focus of the traditional model of democratic representa-
tion, also known as “promissory representation.”
This topic also features prominently in political
debate around the world. Politicians often claim to
hold a mandate to carry out their election platforms
(Grossback, Peterson, and Stimson 2005). Parties’ elec-
tion programs and specific election pledges receive con-
siderable media attention (Krukones 1984). The popular
narrative is generally negative. In 31 of the 33 coun-
tries studied in an international survey, more respon-
dents disagreed than agreed with the following statement:
“People we elect as MPs try to keep the promises they have
made during the election” (ISSP Research Group 2008; see
also Naurin 2011; Thomson 2011). Such skeptical views
are by no means confined to uninformed mass opinion.
Manin (1997, 180) writes, “Even assuming that voters
choose to pay some attention to the candidates’ promises,
they know, or should know, that the credibility of those
promises is an open question. It is not reasonable on their
part to suppose that candidates will necessarily honor
their commitments.” Our research examines whether and
when such views reflect the reality of politics.
Parties’ election programs and the policy pledges
they contain also have important intra- and interparty
functions. Party leaders must respond to and represent
factions to maintain internal support (Thomassen 1994,
256–58). Factions view their parties as vehicles through
which they can attain political goals, and one way in which
this can be done is to secure the party’s commitment
to specific policies in its election pledges. Also, in the
absence of detailed election programs, party leaders may
be undisciplined. With reference to Latin American par-
ties, Mainwaring and Scully (1995, 25) note: “Unfettered
by party platforms, [political leaders] make policy choices
that tend to be short-term and erratic. They are more
prone to demagoguery and populism, both of which have
deleterious effects on democracy.”
The interparty function of election pledges is relevant
in systems where power sharing is the norm. Coalition
governments are typically formed after negotiations be-
tween prospective governing parties based on each party’s
election program (Strøm, M¨
uller, and Bergman 2010).
Parties’ campaign promises are relevant to government
formation since they are clear statements of the policy
positions with which parties enter these negotiations. In
coalition governments, the extent to which each govern-
ing party’s policy proposals are turned into policies is
a mark of its relative power in the coalition (Laver and
Shepsle 1996, 42).
Our research on the fulfillment of election pledges
complements the saliency approach to the mandate
model, in which scholars focus on the relative em-
phases parties place on different policy themes (Budge,
Robertson, and Hearl 1987; Klingeman, Hofferbert, and
Budge 1994; Robertson 1976). The saliency approach in-
vestigates the program-to-policy linkage by examining
the association between parties’ emphases of various pol-
icy themes in their programs and subsequent government
spending in related policy areas. Similarly, Sulkin (2009)
examined issue emphases in U.S. congressional candi-
dates’ campaign appeals and found that these correlate
with relevant legislative activity once in office. The pledge
approach is distinct from the saliency approach in that it
features the specific policy contents of parties’ electoral
appeals, which allows a more fine-grained analysis of the
program-to-policy linkage.
The research approach we use is quite straightfor-
ward: We identify pledges in party programs and evalu-
ate the extent to which these pledges were fulfilled (for
reviews of this approach and comparisons to other ap-
proaches, see Mansergh and Thomson 2007; Naurin 2014;
Royed 1996). Election pledges are commitments in par-
ties’ programs to carry out certain policies or achieve
certain outcomes. These commitments are sufficiently
THE FULFILLMENT OF PARTIES’ ELECTION PLEDGES 3
detailed for researchers to test whether they were ful-
filled after the election. In one of the earliest pledge
studies, Pomper (1968; Pomper and Lederman 1980) ex-
amined the fulfillment of election pledges made by U.S.
parties. Our study takes the pledge approach forward with
the first truly comparative analysis of pledge fulfillment
across 12 countries: Austria, Bulgaria, Canada, Germany,
Ireland, Italy, the Netherlands, Portugal, Spain, Sweden,
the United Kingdom, and the United States. Our data
come from a comparative project in which scholars have
worked to develop common definitions, test the reliability
of different approaches, and produce data that are compa-
rable. We merge some previously analyzed data (recoded
so that they use a common definition of pledge)along
with thousands of new pledges not previously analyzed.
Our integrated analysis allows us to make comparable as-
sessments of pledge fulfillment in different institutional
and economic contexts and to control for pledge char-
acteristics when examining variation in fulfillment. We
find that some governing parties fulfill high percentages
of their previous election pledges, but also that there is
substantial variation in pledge fulfillment. Our main fo-
cus is on the impact of power-sharing arrangements on
pledge fulfillment. One of our main findings is that parties
in single-party executives, both with and without legisla-
tive majorities, are more likely to fulfill their pledges than
parties in coalitions.
Hypotheses
We begi n t h e anal y sis by a d dres s ing th e m ain de s c rip-
tive question raised by mandate theory: To what extent
do governments fulfill parties’ election pledges? We also
examine whether there is a stronger program-to-policy
linkage for governing parties than for opposition parties.
We refer to a party as a “governing party” if it held execu-
tive office after the election and an “opposition party” if
it did not, regardless of whether the party was in power
during the election campaign in which it made the pledge.
The fulfillment of opposition parties’ pledges can be ex-
plained at least in part by the fact that governing parties
made the same or similar pledges on some issues, or that
the pledges concerned uncontentious policies that any
government would enact. Governing parties may also en-
act and take credit for popular policies proposed by their
opponents.
Our main focus is the impact of power-sharing
arrangements on the fulfillment of pledges made by
governing parties. Single-party executives with legisla-
tive majorities are a defining characteristic of political
systems. According to Lijphart’s (1999) influential ty-
pology, winner-takes-all majoritarian systems usually
have single-party majority executives, whereas consensus
democracies, in which power is dispersed more widely,
generally have minority governments and coalitions. For
Powell and Whitten (1993), single-party majority govern-
ments make for greater clarity of responsibility, because
this gives a single party the most control over policymak-
ing. Their argument concerning clarity of responsibility
refers specifically to the contexts in which voters hold
governments to account for variation in economic per-
formance. However, the idea that single-party majority
governments have the most control over economic poli-
cies and performance is of obvious relevance to partisan
control in other policy areas. Our main hypothesis with
respect to power sharing is the following:
H1: Parties that form single-party executives with
legislative majorities are more likely to fulfill their
election pledges than governing parties that are
compelled to share power with other parties.
Governments in which parties are compelled to share
power are single-party minority governments, major-
ity coalitions, and minority coalitions. Our main anal-
yses control for government duration, which tends to be
shorter for coalitions and minority governments (Saalfeld
2013).
On balance, we expect single-party status to matter
more than majority status. Previous research suggests that
the difference between governments with and without
legislative majorities may not be large. Mayhew’s (2005)
research on the United States indicates that divided gov-
ernments, in which the executive party does not control
a legislative majority in one or both houses, can be as
productive as those in which a single party controls both
branches. Strøm (1990) also finds that minority govern-
ments, which are common in many small European coun-
tries, work effectively.
We exa m i ne sev e ral fa c tors th a t may a f f ect pl e dge
fulfillment in minority governments, the first of which
is the presence of support agreements with opposition
parties. Strøm (1990, 108–9) describes minority govern-
ments with support agreements as majority governments
in disguise, and defines genuine minority governments
as those that seek support for each new initiative on
an ad hoc basis. Second, single-party minority govern-
ments that control the median legislator are likely to be
stronger than those that do not. Crombez’s (1996) model
of government formation predicts that as the largest party
becomes more centrally located and larger, although still
holding less than 50% of legislative seats, the likelihood of
asingle-partyminoritygovernmentincreases.Moreover,
4ROBERT THOMSON ET AL.
when these conditions are met, Crombez’s model sug-
gests that minority status is a sign of strength rather than
weakness. Third, legislative procedures such as investi-
ture rules define the way in which minority governments
form. While there are many differences among investiture
rules, it is common to distinguish between positive and
negative rules (Rasch, Martin, and Cheibub 2015). Under
positive investiture rules, prospective governments must
win the support of simple or absolute majorities in order
to take office. Under negative rules, by contrast, support
is assumed to exist unless a legislative vote, usually by
absolute majority, proves otherwise. Negative investiture
rules reduce the need for prospective minority govern-
ments to make bargains with opposition parties that may
dilute their parties’ campaign commitments. Nonethe-
less, even in the presence of negative investiture rules,
minority governments still need to win majorities for the
specific bills they propose, which puts them in a weaker
position compared to majority governments.
The theory of veto players is relevant to the ex-
pectation that governing parties in coalitions are less
likely to fulfill their election pledges. Tsebelis (2002)
argues that policies tend to be more stable in systems
with more veto players who have greater ideological
diversity among them. The ideological range of parties in
governing coalitions is commonly used as a measure of
the heterogeneity of the set of relevant veto players (e.g.,
Tsebelis and Chang 2004). Since most election pledges
concern proposals to bring about change, veto players
theory could be taken to imply that pledge fulfillment
is lower in coalitions with more parties and with parties
that span a wider ideological range.
It is, however, questionable whether veto players the-
ory implies hypotheses regarding the effect of the ide-
ological range of coalitions on pledge fulfillment, and
we therefore simply include this variable as a control.
McGann and Latner (2013, 827) point out that governing
parties are not veto players according to the accepted def-
inition of this term as members of every possible winning
coalition; instead, governing parties are simply members
of the winning coalition that happens to form. One coali-
tion member’s attempt to block change in the form of the
fulfillment of its partner’s election pledges may not result
in nonfulfillment, but rather in the formation of a new
government. Furthermore, the link between the spatial
model of politics and election pledges may not be a simple
one. Large ideological distances need not imply low levels
of pledge fulfillment. The saliency theory of party com-
petition (Budge, Robertson, and Hearl 1987; Robertson
1976) states that ideologically distant parties tend to fo-
cus on—and, by implication, make pledges on—different
issues or themes. This makes it theoretically possible for
all parties in ideologically diverse coalitions to have high
levels of pledge fulfillment.
Within governing coalitions, we expect the party of
the chief executive or prime minister to have an advan-
tage in terms of pledge fulfillment. According to models
of coalition formation and policymaking in coalitions,
the party of the chief executive has greater influence over
policy than does its junior coalition partner(s). Several
models focus on the proposal power of the party se-
lected to initiate government formation, which puts them
in a strong position relative to other coalition members
(Austen-Smith and Banks 1988; Baron 1991; Diermeier
and Feddersen 1998). The party that leads the process
of coalition formation is generally the largest party that
goes on to control the chief executive. For Huber (1996),
the vote of confidence procedure enables the prime min-
isterial party to raise the stakes in any legislative vote by
making it a vote of confidence in the government, thereby
limiting the extent to which other coalition parties influ-
ence policies. Some models show how the prime minister’s
party can shape policies by reconfiguring the jurisdictions
of ministerial portfolios (Dewan and Hortala-Vallve 2011;
Thies 2001).
Within coalitions, we also expect that pledges are
more likely to be fulfilled if the party that made them went
on to hold the relevant ministerial portfolio. In Laver and
Shepsle’s (1996) model of coalition policymaking, parties
have little say in policy areas over which they do not re-
ceive ministerial control. According to this model, parties
will be persuaded to participate in a coalition only if they
believe it is credible in terms of policy, and the distri-
bution of ministerial portfolios provides such credibility.
In addition, models of ministerial drift posit that minis-
ters may pursue initiatives that differ from their govern-
ment’s common platform (Huber and Shipan 2002, 185;
Martin and Vanberg 2004, 15–16). There are, however,
other models of coalition policymaking in which pol-
icymaking is a process of compromise between coali-
tion partners and ministerial autonomy is constrained
(Dunleavy and Bastow 2001; Warwick 1999).
Our hypotheses regarding the fulfillment of pledges
by parties within coalitions are the following:
H2a:Incoalitiongovernments,partiesthatreceivethe
prime ministership are more likely to fulfill their
election pledges than other coalition members.
H2b:Incoalitiongovernments,apledgeismorelikely
to be fulfilled if the party that made it receives
the relevant portfolio.
The power-sharing arrangements considered so far,
particularly the comparison of single-party majority gov-
ernments with other government types, are part of a
THE FULFILLMENT OF PARTIES’ ELECTION PLEDGES 5
broader set of institutional constraints on governing par-
ties. Different theoretical approaches highlight different
institutional features in addition to power-sharing ar-
rangements within executives. Lijphart’s (1999) typology
also includes federalism, bicameralism, judicial review,
and central bank independence. Powell and Whitten’s
(1993) index of clarity of responsibility also includes
the power of opposition parties in legislative commit-
tees, party cohesion, and bicameralism. McGann and
Latner’s (2013) theory incorporates the proportionality
of the electoral system and an index of characteristics that
constrain governments, including bicameralism, federal-
ism, presidentialism, and referenda. A common thread
running through these theories is that it is harder for
governments to get things done in the face of more in-
stitutional constraints. Competing centers of power, or
competitive veto points in the language of veto players
theory, make it more difficult to realize policy change.
They in effect raise the threshold of support required for
adecisiontobetaken.Intermsofpledgefulfillment,this
leads to the following expectation:
H3: Governing parties that are subject to fewer insti-
tutional constraints are more likely to fulfill their
election pledges.
Given the 12 countries we include in the present
study, we focus on (semi-)presidentialism, bicameralism,
federalism, and European Union (EU) membership as
possible constraints on governing parties that reduce the
likelihood of pledge fulfillment.
In addition to the degree of government and institu-
tional control, which are our main concerns, we expect
other factors to matter and control for these other vari-
ables. Pledges are more likely to be redeemed when gov-
erning parties have more resources in terms of finances
and time. Economic conditions are among the most im-
portant factors in explaining variation in public policy
outputs (Huber and Stephens 2001). Growth provides
government revenues, which are directly relevant to ful-
filling pledges that increase expenditure or cut taxes, and
indirectly relevant to fulfilling regulatory measures, which
often have implications for the allocation of government
personnel. Like revenue, time is a resource that enhances
the ability of governments to get things done; thus, when
governments are short in duration, we anticipate lower
rates of pledge fulfillment.
Characteristics of the pledges themselves and the pro-
grams in which they are made may also affect the like-
lihood of fulfillment. We control for the total number
of pledges made by the party, but we do not formulate
a specific expectation about the direction of the effect.
On the one hand, it might be more difficult to fulfill
more pledges than fewer. On the other hand, a larger
number of pledges may indicate that the party is more fo-
cused on pledges as part of its electoral and government
strategy, which may mean a higher likelihood of fulfill-
ment. We also control for incumbency and government
experience by categorizing parties into three groups ac-
cording to their status during the election campaign in
which they made the pledges we examine: incumbents,
opposition parties with prior governing experience, and
oppositions parties without prior governing experience.
We expe c t that pl e dges b y i ncum b ents ar e most li kely t o
be fulfilled, whereas pledges by opposition parties with
no prior governing experience are least likely to be ful-
filled if they enter office. By identifying parties without
governing experience, we consider parties’ expectations
regarding the likelihood they will enter government of-
fice. Parties without prior governing experience may not
expect to enter government and therefore have to imple-
ment their pledges, which may make them more inclined
to formulate pledges that are difficult to fulfill.
Characteristics of pledges themselves may matter too.
Our hypotheses refer to pledges that involve changes to
the status quo. In a minority of pledges, parties promise
to maintain the status quo on particular issues, and we
analyze the fulfillment of these pledges separately, antici-
pating a high rate of fulfillment given the incrementalism
of policymaking in large governments. We also explore
whether pledges of different types, such as pledges to cut
taxes or expand programs, differ significantly in terms of
their likelihood of being fulfilled.
The relations among pledges made by different par-
ties should also be considered, particularly when power-
sharing arrangements compel parties to cooperate with
others. We expect that governing parties’ pledges are more
likely to be fulfilled if other governing parties made the
same or similar pledges. Our models also control for the
fact that in six of the 57 election campaigns we study,
at least some of the pledges appeared in programs writ-
ten by pre-election coalitions of parties. We expect that
pledges made by pre-election coalitions are more likely
to be fulfilled than pledges made by separate parties that
enter coalitions.
It may be argued that parties consider the future in-
stitutional environment they expect to encounter when
making pledges, with parties that expect more obstacles
making more modest pledges. This could nullify the ob-
servable effects of institutions on fulfillment; to the extent
that we do find institutional effects, these may be under-
estimates of their true magnitude. There are two other
responses to this concern regarding endogeneity. First,
rational parties have good reason to make pledges they
know they may not be able to keep. Pledges that have a
6ROBERT THOMSON ET AL.
low likelihood of being fulfilled create negotiating space
in coalition or legislative negotiations. Such pledges also
serve to signal commitment to key supporters. Second, the
qualitative evidence casts doubt on the idea that parties
only make pledges that they expect to be able to fulfill.
Small parties do not issue more modest election plat-
forms than large parties just because they are less likely
to govern alone; instead, they set out what they would
ideally do if it were up to them. The U.S. case gives many
examples of the fact that parties do not necessarily tai-
lor their pledges to foreseeable constraints. Republican
Party platforms regularly pledge to enact a constitutional
amendment banning abortion, while this is unlikely to be
fulfilled without winning both the presidency and unre-
alistically large majorities in both houses.
Research Design
Case Selection
The study includes information on 20,023 pledges made
by parties prior to the formation of 57 governments in
12 countries. The governments included are representa-
tive in that they offer variation in the four types of execu-
tive government: (1) single-party executives with legisla-
tive majorities, (2) single-party executives with legislative
minorities; (3) coalition-based executives with legislative
majorities, and (4) coalition-based executives with leg-
islative minorities (see Table 1). Each of the governments
lasted at least 12 months and was the first government to
take office after the previous election.1The selection of
these countries was a consequence of researchers work-
ing independently on the countries on which they have
specialist knowledge and interest, and subsequently co-
ordinating their efforts to make comparisons possible.
For the pu rposes of com parison, we treat U.S. gover n-
ments as cases of single-party executive government with
or without legislative majorities. Divided governments in
the United States have more in common with single-party
minority governments than with coalition governments;
U.S. presidents must negotiate with the other main party
in the legislature on each piece of legislation, as opposed
to coalition governments, which typically have ongoing
agreements on a range of issues. We recognize, however,
that these U.S. governments may differ from the other
1The countries and time periods covered included several short-
lived governments of less than 12 months’ duration and caretaker
governments (in Bulgaria and Ireland), which we do not examine
here, but are considered in some detail in country-specific studies.
These cases provide insights into the tribulations of governing in
times of transition or crisis, rather than the effects of power-sharing
arrangements in which we are interested in this study.
single-party executives included, and we explore whether
our key findings differ when excluding the United States
from the analysis, which they do not (see the online sup-
porting information).
There is some variation among the country studies in
the policy areas included as well as the time periods cov-
ered, and we include controls and tests to check that this
does not drive our main findings. Most country studies
include all policy areas for all main parties, while a few
focus on a broad subset of socioeconomic policy for all
or some parties. The data for the Netherlands and Spain
and for one of the Irish governments include only socioe-
conomic policy. We control for the time period in which
each pledge was made with a categorical variable for the
decade since we do not expect time to have a linear effect.
Measurement
There are two steps to the pledge-testing approach: first,
identifying pledges in election manifestos, and second,
testing fulfillment. For a statement to qualify as a pledge,
it must contain language indicating commitment to some
future action or outcome. Pledges include both firm com-
mitment language, such as “we will” or “we promise to,”
as well as more softly described intention, such as “we
support” or “we favor,” as long as parties indicate that
they support the action or outcome referred to unequiv-
ocally. What determines whether a statement qualifies as
apledgeisthetestabilityoftheactionoroutcometo
which the party is committing itself. A pledge is a state-
ment committing a party to an action or outcome that
is testable:Thatis,wecangatherevidenceandmakean
argument that the action or outcome was either accom-
plished or not. Many statements that begin with hard
commitment language would be considered rhetoric, not
pledges, because they do not meet the testability criteria—
for example, “we will ensure that our government shows
respect for families” or “we support fair treatment for
all.” We define a pledge as astatementcommittingaparty
to one specific action or outcome that can be clearly deter-
mined to have occurred or not.Thesupportinginforma-
tion provides an extensive discussion of the conceptual
issues associated with the definition of pledges, as well as
the different approaches that scholars within our group
used before settling on this common denominator for the
purposes of comparative research.
Reliability tests were conducted on the identification
of election pledges. For the specific definition used here,
nine researchers independently coded part of the 2008
Canadian Conservative Party manifesto. The reliability
for each pair of coders was computed as x/n, where x is
THE FULFILLMENT OF PARTIES’ ELECTION PLEDGES 7
TABLE 1The12Countriesand57GovernmentsIncluded
Single-Pparty Executives with Legislative Majorities (15 Governments)
Bulgaria: 1997–2001, ODS
Canada: 1993–97, Liberals; 1997–2000, Liberals; 2000–4, Liberals; 2011–15, Conservatives
Ireland: 1977–81, Fianna F´
ail
Portugal: 2005–9, PS
Spain: 1989–93, PSOE; 2000–04, PP
UK: 1974–79, Labour; 1979–83, Conservative; 1983–87, Conservative; 1987–92, Conservative; 1992–97, Conservative
United States: 1977–81, Democrats
Single-Party Executives with Legislative Minorities (16 Governments)
Bulgaria: 2009–13, GERB
Canada: 2004–6, Liberals; 2006–8, Conservatives; 2008–11, Conservatives
Ireland: 1987–89, Fianna F´
ail
Portugal: 1995–99, PS
Spain: 1993–96, PSOE†;1996–2000,PP
†
Sweden: 1994–98, Social Democrats†;1998–2002,SocialDemocrats
†;2002–6,SocialDemocrats
†
United States: 1981–85, Republicans; 1985–89, Republicans; 1989–93, Republicans; 1993–97, Democrats; 1997–2001,
Democrats
Coalition Executives with Legislative Majorities (22 Governments)
Austria: 2000–3, ¨
OVP/FP ¨
O; 2003–7, ¨
OVP/FP ¨
O; 2007–8, SP ¨
O/ ¨
OVP; 2008–13, SP ¨
O/ ¨
OVP
Bulgaria: 1995–96, BSP/NS; 2001–5, NDSV/DPS; 2005–9, BSP/NDSV/DPS‡
Germany: 2002–5, SPD/ Greens; 2005–9, CDU-CSU/ SPD; 2009–13, CDU-CSU/ FDP
Ireland: 1982–87, Fine Gael/Labour; 1989–92, Fianna F´
ail/Progressive Democrats; 1992–94, Fianna F´
ail/ Labour; 2002–7,
Fianna F´
ail/Progressive Democrats; 2007–11, Fianna F´
ail/ Progressive Democrats/ Greens; 2011–16, Fine Gael/Labour
Italy: 2001–6, Berlusconi II’s coalition (FI/AN/LN/UDC/NPSI/PRI)‡;2008–11,BerlusconiIV’scoalition(PdL/LN/
MpA/PID)‡
Netherlands: 1986–89, CDA/VVD; 1989–94, CDA/PvdA; 1994–98, PvdA/VVD/D66
Sweden: 2006–10, Moderate Party/Centre Party/People’s Party/Christian Democrats‡
Coalition Executives with Legislative Minorities (4 Governments)
Ireland: 1997–2002, Fianna F´
ail/Progressive Democrats†
Italy: 1996–98; Prodi I’s Ulivo coalition (PDS/PPI/RI/FdV/UD)†‡;2006–8,ProdiII’scoalitionUnione(DS/DL/PRC/RnP-
PdCI/IdV/FdV/UDEUR)‡
Sweden: 2010–14, Moderate Party/Centre Party/People’s Party/Christian Democrats‡
Note:†=minority governments that had an agreement with one or more opposition parties or parliamentarians to maintain support;
‡=includes pledges made by governing parties that were part of pre-election coalitions. Data set does not include opposition parties in
Italy.
the number of statements that both coders identified as
pledges and n is the total number of statements identi-
fied as pledges by at least one of the nine coders. The
nine coders identified a total of 99 pledges in the mani-
festo, with an average paired reliability of 74%. Separate
reliability tests were carried out within several country
studies with higher levels of agreement between coders:
Ireland (80%), the Netherlands (88%), Spain (87%),
Sweden (94%), and the United States and United
Kingdom (together 84%). The reliability tests in the stud-
ies of Ireland, the Netherlands, and Spain and Sweden
used the same narrow definition of pledges as that used
here, whereas the studies of the United States and the
United Kingdom used a broader definition.
We also c onduc t ed a re l i abil i t y test o n t he cat e goriz a -
tion of pledges as “fully,” “partially,” or “not” fulfilled.
Depending on the nature of the action or outcome re-
ferred to in the pledge, a variety of sources were consulted
to test fulfillment, including legislation, ministerial de-
crees, budgetary or other data, and secondary sources.
A total of 40 pledges were randomly selected (five from
eight of the countries examined here) and examined by
8ROBERT THOMSON ET AL.
seven researchers. The researcher primarily responsible
for work on the country concerned provided the other
researchers with the evidence he or she used to evaluate
the fulfillment of each pledge without revealing his or her
evaluation, and translated the relevant material into En-
glish if necessary. Seven researchers then independently
categorized each pledge as “fully,” “partially,” or “not”
fulfilled. Across the 21 pairs of researchers, we found an
average agreement rate of 93%.
For 10 of the 12 countries selected, we have data on
the three-category indicator of fulfillment, but work done
on Italy and Spain used the dichotomous categorization
of “not fulfilled” and “at least partially fulfilled.” To maxi-
mize the numbers of cases, our multivariate analyses focus
on this dichotomous indicator in all countries.
The analyses also include information from our cod-
ing of the characteristics of each pledge. We coded whether
each pledge directly agreed or disagreed with a pledge
made by one or more of the other parties. Our mul-
tivariate analyses count multiple mentions of the same
or a similar pledge by different parties at the same elec-
tion only once.2We also d i stin g u ishe d b etwe e n p ledg e s
to maintain the status quo, which are a minority of cases,
and pledges to introduce some kind of change. We ex-
clude status quo pledges from the multivariate analyses
because our hypotheses are relevant to change pledges.
For a subset of six countries, we developed a more de-
tailed categorization of pledge type, which distinguishes
tax cut and expansionary pledges among others, and re-
port on this in the supporting information.
To estimate the ideological range of each coalition
government, we use measures of the ideological positions
of each party on the left-right dimension formulated by
Lowe et al. (2011), which adjust data from the Compar-
ative Manifestos Project (Budge et al. 2001; Klingemann
et al. 2006) for measurement error. We take the abso-
lute distance between the two most extreme parties in
the coalition as the measure of ideological range. We also
control for the ideological distance between each party
and the position of the party of the median legislator.
2In the multivariate analyses, we measure the fulfillment of a pledge
that was made by several governing parties in a coalition by assessing
the pledge as it was formulated by the party that obtained the prime
ministership or the largest of the governing parties if none held the
prime ministership. A pledge is coded as “similar” to another if
the fulfillment of one of the pledges would mean that the other
pledge was also fulfilled at least partially. So pledges by Party A
to cut the basic rate of tax from 20% to 15% and Party B to cut
the tax rate from 20% to 17.5% would be coded as “similar,” even
though they are distinct promises. Very few pledges are in direct
disagreement with other pledges, and our results are the same if
these are excluded from the analysis.
For coalition governments, we include a control for
the Herfindahl index of concentration based on the seat
shares of each coalition member. This varies from .32 to 1
in our sample, and it becomes larger as the share of seats
held by any one of the parties becomes larger. The measure
of GDP growth uses data from the World Bank. It is the
average growth rate over the lifetime of the government.
Analysis of Pledge Fulfillment
Election pledges refer to many important policy changes.
To take exampl e s f r o m t h r e e countries in our data set,
in the United States, the Republicans under Reagan at
least partially fulfilled pledges to cut certain taxes, enact
deregulation, and tighten eligibility for food stamps. At
the same time, the Republicans failed to fulfill pledges to
reduce certain other taxes, enact a youth minimum wage,
and create tuition tax credits for private schools. Among
the U.K. Conservative Party’s fulfilled pledges were com-
mitments to reduce taxes, particularly for high earners;
sell off public housing; and privatize certain public-sector
companies. In the three-party majority coalition that took
office in the Netherlands in 1994, the Labour Par ty (PvdA)
partially fulfilled a promise to raise welfare payments in
line with wage increases in the private sector, while the
Liberal Party’s (VVD) pledge to freeze welfare benefits at
their 1994 levels went unfulfilled, as did the Democrat
66’s (D66) pledge to reduce the top rate of income tax.
Although the subsequent analyses focus on systematic
variation in pledge fulfillment, we should not lose sight
of the importance of pledges to the people they affect.
Figure 1 shows descriptive information on pledge ful-
fillment in the 12 countries included. The first main find-
ing is that governing parties fulfilled a clear majority of
pledges at least partially: 60% (5,439 of the 9,133 pledges
were fulfilled at least partially). There is substantial vari-
ation in levels of pledge fulfillment, and the aggregate
figures suggest that governing parties in single-party exec-
utives are more likely to fulfill their election pledges than
parties in coalitions. The highest rate of fulfillment for
governing parties is found in the United Kingdom, where
all five governments were single-party governments with
parliamentary majorities; the U.K. governing parties ful-
filled 86% (494 of 575 pledges) at least partially. The
single-party minority governments in Sweden also ful-
filled a remarkably high 87% (112 of 129) of their pledges
at least partially, whereas the Swedish majority coalition
of 2006–10 fulfilled a lower percentage of its pledges: 68%
(92 of 135). However, the Swedish minority coalition that
took office in 2010 fulfilled a higher percentage of pledges
THE FULFILLMENT OF PARTIES’ ELECTION PLEDGES 9
FIGURE 1TheFulfillmentofElectionPledgesbyCountryandGovernmentType
Note:UK=United Kingdom; SE =Sweden; PT =Portugal; ES =Spain; CA =Canada; DE =Germany;
US =United States; NL =The Netherlands; IE =Ireland; BU =Bulgaria; AT =Austria; IT =Italy; G =parties
that held executive office after the elections; O =parties that did not. The study of Italy does not include pledges
made by opposition parties. Sin maj =single-party majority governments; Sin min =single-party minority
governments; Co maj =majority coalitions; Co min =minority coalitions. Numbers above bars refer to the
total numbers of pledges tested for fulfillment.
in its pre-election coalition platform: 82% (182 of 223).
There are also relatively high rates of pledge fulfillment
in the two single-party Portuguese governments. The
lowest overall rates of pledge fulfillment are found in
Ireland, Bulgaria, Austria, and Italy. These governments
include several relatively short-lived coalitions, and in
the Austrian and Italian cases broad coalitions. Two of
the Austrian coalition governments include the far-right
populist Freedom Party, which had little government ex-
perience at the national level and faced ongoing internal
disagreements, which also led to the early termination of
the first cabinet. U.S. presidential parties fulfilled a higher
percentage of their election pledges than governing par-
ties in most coalition systems, but less than governing
parties in parliamentary systems where single-party gov-
ernments are the norm.
Figure 1 also contains information on the fulfillment
of pledges made by nonexecutive parties. Opposition par-
ties’ pledges have a reasonable likelihood of being fulfilled,
particularly when opposition parties face minority gov-
ernments and/or coalitions. In Germany, for instance,
42% of opposition parties’ pledges were fulfilled at least
partially. Nonpresidential parties in the United States also
saw relatively high percentages of their pledges fulfilled.
In most periods, nonpresidential parties held congres-
sional majorities. In the United Kingdom, pledges made
by opposition parties are less likely to be fulfilled, which
accords with the general view of the United Kingdom as
awinner-takes-allmajoritariansystem.Theparticularly
high percentage of fulfillment for opposition parties fac-
ing minority coalitions should be treated with caution
since this figure is based on pledges made by one Irish
party in 1997. The country study of Italy does not include
information on opposition parties’ pledges.
Table 2 presents the first multivariate models. Each
of the 7,770 observations refers to a campaign pledge
made by one or more parties that held executive office
after the election.3The dependent variable is whether the
3The data have a hierarchical structure to which a multi-
level model could be applied; pledges are nested in 81 party
platforms of pre-election coalition platforms, which are nested in
campaigns/governments, which are nested in countries. There are,
however, few obser vations at some of the higher levels (12 countr ies
and a maximum of four different election programs within any of
10 ROBERT THOMSON ET AL.
TABLE 2FactorsAffectingExecutiveGoverningParties’PledgeFulfillment
Model 1 Model 2
ebb(s.e.) p ebb(s.e.) p
Government type (reference: single-party majority)
Single-party minority 1.53 .43 (.18) .02 1.45 .37 (.19) .05
Coalition majority .36 −1.01 (.18) .00 .41 −.89 (.26) .00
Coalition minority .37 −1.00 (.32) .00 .41 −.89 (.36) .02
Chief executive 1.55 .44 (.15) .00
Relevant portfolio 1.26 .23 (.14) .10
Ideological range 1.05 .05 (.17) .77
Herfindahl index .72 −.33 (.47) .49
Presidentialism .31 −1.16 (.39) .00 .32 −1.15 (.40) .00
Semi-presidentialism 1.28 .25 (.34) .48 1.22 .20 (.35) .57
Bicameralism 1.13 .12 (.22) .59 1.06 .06 (.22) .78
Federalism .99 −.01 (.21) .95 .92 −.09 (.21) .68
EU member .94 −.06 (.22) .77 .91 −.09 (.22) .67
GDP growth 1.11 .11 (.04) .01 1.10 .10 (.04) .01
Duration in years 1.44 .36 (.06) .00 1.39 .33 (.06) .00
Governing experience (reference: incumbents)
Opposition parties with experience .73 −.32 (.16) .04 .74 −.30 (.17) .08
Opposition parties without experience .48 −.74 (.17) .00 .46 −.78 (.18) .00
Number of pledges (/10) .99 −.01 (.01) .32 .99 −.01 (.01) .28
Pre-election coalition 2.58 .95 (.28) .00 1.72 .54 (.30) .07
Ideological distance to median legislator 1.08 .07 (.18) .68
Decade (reference: 1970s)
1980s .67 −.40 (.29) .17 .67 −.41 (.32) .20
1990s .78 −.25 (.31) .42 .81 −.21 (.31) .51
2000s .89 −.12 (.29) .68 .89 −.12 (.30) .70
Subset of pledges tested .79 −.24 (.18) .18 .79 −.23 (.20) .25
Constant .80 −.22 (.37) .55 .75 −.29 (.62) .64
Log pseudolikelihood −4860.90 −4836.23
chi2(p) 531.13 (.00) 677.81 (.00)
npledges(programs) 7,770(81) 7,770(81)
Note: Logit models with dependent variable partially/fully fulfilled =1 and not fulfilled =0. Standard errors clustered by program.
Countries weighted equally.
pledge in question was at least partially fulfilled (1) or not
(0). The models include only governing parties’ pledges,
exclude status quo pledges, and count pledges made by
more than one governing party once, as described above.
This accounts for the smaller number of cases in Table 2
compared to the total number of governing parties’
pledges reported in Figure 1. Status quo pledges are a
minority of pledges and are very likely to be fulfilled; of
our governments), which makes multilevel modeling problematic.
Two of the explanatory variables—Relevant por tfolio and Agreement
between coalition partners—vary within programs, which makes
us prefer individual pledges as the unit of analysis, rather than
platforms.
the 9,133 pledges made by governing parties, 871 were sta-
tus quo pledges. Of these 871 pledges, 768, or 88%, were
at least partially fulfilled, and there is little variation in
the fulfillment rate across different types of government.
While status quo pledges can be substantively important
promises, they are qualitatively different from pledges to
change the current state of affairs, which makes it appro-
priate to exclude them from the multivariate analyses.
The headline finding from Table 2 is that parties
in single-party governments, both with and without
legislative majorities, significantly outperform parties in
coalitions on pledge fulfillment. Governing parties in
single-party majority governments by definition control
THE FULFILLMENT OF PARTIES’ ELECTION PLEDGES 11
the chief executive and all ministries, and they have an
ideological range of 0, a Herfindahl index of 1, and a
distance of 0 to the median legislator. Model 1 therefore
excludes these variables. In Model 1, the negative coeffi-
cients for the variables Coalition majority and Coalition
minority indicate that parties in these ty pes of government
are significantly less likely to fulfill their pledges than par-
ties in single-party majority governments, which is the
reference category. The coefficients for Coalition majority
and Coalition minority governments are also significantly
different from the coefficient for Single-party minority
governments (p =.00), indicating that parties in coali-
tions are significantly less likely to fulfill their pledges
than parties in single-party minority governments. The
magnitude of the effects of the coalition variables is also
substantial. The odds ratios for Coalition majority and
Coalition minority governments (ebof .36 and .37) indi-
cate that the odds of pledge fulfillment in these types of
government are 64% and 63% lower, respectively, than
for single-party majority governments.
Surprisingly, Model 1 also reveals that parties in
single-party minority governments are not significantly
less likely to fulfill their pledges than parties in single-
party majority governments. On the contrary, governing
parties in single-party minority governments are signif-
icantly more likely to fulfill their pledges according to
Model 1, although this positive effect becomes borderline
significant in Model 2, which adds additional explanatory
variables. The positive and significant effect also becomes
clearly insignificant when we exclude the duration vari-
able from Model 1. Despite the lack of robustness of the
significant positive effect of Single-party minority gov-
ernment, the absence of a significant negative effect is
noteworthy.
The single-party governments deserve further analy-
sis, particularly with respect to the comparison between
the 15 majority and 16 minority single-party govern-
ments. This analysis highlights that majority governments
do not outperform minority governments, even those of
arelativelyshortduration.Model1ofTable3examines
pledges made by parties that entered single-party gov-
ernments and distinguishes between the 11 single-party
minority governments that lasted longer than 3 years and
the five single-party minority governments that lasted less
than 3 years. None of the 15 single-party majority gov-
ernments lasted less than 3 years. Neither of the minority
government coefficients is significant, indicating that the
likelihood of pledge fulfillment for single-party minority
governments is not significantly lower than for single-
party majority governments. The government with the
shortest duration in this model is the single-party minor-
ity Canadian Liberal Party government in 2004–6, which
lasted 17 months and nonetheless fulfilled 72% (63 of 87)
of its pledges at least partially.
With respect to these single-party minority govern-
ments, we further explored (1) agreements with oppo-
sition parties, (2) seat share and control of the median
legislator as highlighted by Crombez’s (1996) model,
and (3) the presence of negative investiture rules, which
strengthen the executive relative to the legislature (details
are in the supporting information). Agreements with op-
position parties do not have a consistent effect on the like-
lihood of pledge fulfillment, although there is some ev-
idence that the Spanish minority governments benefited
from this arrangement. Neither seat share nor control
of the median legislator accounts for variation in pledge
fulfillment among minority governments. Our data ex-
hibit relatively little variation in these variables; of the
11 single-party minority governments outside the United
States, all consisted of the largest party in the legislature
with an average of 45% of the legislative seats, and only
one had less than 40% of seats. Eight controlled the me-
dian legislator, and the three that did not hold the median
legislator (one Spanish government and two Canadian
governments) did not fulfill markedly fewer pledges than
those that did. There is some evidence that negative in-
vestiture rules are associated with higher rates of pledge
fulfillment. Both Sweden and Portugal have negative in-
vestiture rules, which strengthen minority governments
relative to the legislature, and the minority governments
in Sweden and Portugal have high rates of pledge ful-
fillment. In a multivariate analysis reported in the sup-
porting information, the presence of a negative investi-
ture procedure is associated with a significantly higher
likelihood of pledge fulfillment by parties in single-party
minority governments.
We now t u r n to th e v a riabl e s Chief executive and Rel-
evant portfolio,whichtakedifferentvaluesforpartiesin
governing coalitions. Model 2 in Table 2 includes these
and other controls. The coefficients for Coalition major-
ity and Coalition minority remain negative and significant
in Model 2, indicating that the likelihood of pledge ful-
fillment is significantly lower in these situations than in
single-party majority governments. Both of the coalition
coefficients differ significantly from the Single-party mi-
nority coefficient, but they are not significantly different
from each other (p =.99). As expected, Chief executive has
a positive and significant coefficient in Model 2. The odds
of pledge fulfillment are 55% higher if the party receives
the chief executive. We do not, however, find significant
effects associated with Relevant portfolio or Ideological
range.
Model 2 in Table 3 is restricted to the subset of
cases concerning coalitions in which there is variation
12 ROBERT THOMSON ET AL.
TABLE 3Single-PartyGovernmentsandCoalitionsExaminedSeparately
Model 1 Single-Party Governments Model 2 Coalitions
ebb(s.e.) p ebb(s.e.) p
Government type (reference: majority
governments of at least 3 years)
Majority governments of less than 3 years .74 −.30 (.25) .22
Minority governments of less than 3 years .93 −.07 (.32) .83
Minority governments of at least 3 years 1.01 .01 (.24) .96 1.27 .24 (.52) .64
Chief executive 1.39 .33 (.14) .02
Relevant portfolio 1.24 .21 (.11) .06
Ideological range 1.09 .09 (.30) .76
Herfindahl index .81 −.21 (.48) .66
Agreement between coalition partners 1.90 .64 (.24) .01
Presidentialism .43 −.85 (.79) .28
Semi-presidentialism 1.43 .36 (.67) .59
Bicameralism .98 −.02 (.65) .98 1.26 .23 (.25) .36
Federalism .80 −.22 (.62) .72 .89 −.11 (.19) .57
GDP growth .96 −.04 (.06) .55 1.14 .13 (.06) .03
Governing experience (reference: incumbents)
Opposition parties with experience .63 −.46 (.33) .16 .80 −.22 (.15) .15
Opposition parties without experience .32 −1.14 (.29) .00 .74 −.30 (21) .14
Number of pledges (/10) .99 −.01 (.02) .75 .98 −.02 (.01) .20
Pre-election coalition 2.34 .85 (.47) .07
Ideological distance to median legislator 1.30 .26 (.30) .38 .87 −.14 (.22) .53
Decade (reference: 1970s for Model 1; 1980s for
Model 2)
1980s .96 −.04 (.38) .92
1990s 1.75 .56 (.46) .22 1.41 .34 (.21) .11
2000s .97 −.03 (.49) .95 2.28 .82 (.24) .00
Subset of pledges tested .85 −.16 (.36) .66 1.15 .14 (.26) .60
Constant 3.91 1.36 (.47) .00 .34 −1.07 (.58) .06
Log pseudolikelihood −1750.47 −2612.69
chi2(p) 102.71 (.00) 612.51 (.00)
npledges(programs) 2,946(31) 4,021(45)
Note: Logit models. Dependent variable partially/fully fulfilled =1 and not fulfilled =0. Standard errors clustered by program. Countries
weighted equally. Model 1 contains pledges from single-party executives in Bulgaria, Canada, Ireland, Portugal, Spain, Sweden, the United
Kingdom, and the United States. Model 2 contains pledges from coalitions with variation in the variables Chief executive and Relevant
ministry in Austria, Bulgaria, Germany, Ireland, the Netherlands, and Sweden. The table excludes Italy and the Swedish coalition from 2010
to 2014, for which we only have pledges from pre-election coalitions.
in the occupancy of the chief executive and relevant min-
istries within each coalition.4This model confirms the
finding that holding the post of chief executive within
coalitions has a marked and significant positive effect
4The Italian coalitions and the 2010-14 Swedish coalition are ex-
cluded from this model because all election pledges are sourced
from common programs of pre-election coalitions, which means
there is no variation in the variables Chief executive or Relevant
ministry.
on the likelihood of pledge fulfillment. Again, the ef-
fect associated with holding the relevant ministerial
portfolio falls just short of statistical significance. This
non-finding is contrary to the expectation from the model
ministerial-portfolio allocation, but in line with mod-
els of coalition bargaining in which ministers are con-
strained by the compromises reached with their coalition
partners.
The variables Ideological range and Herfindahl in-
dex do not affect pledge fulfillment in coalitions,
THE FULFILLMENT OF PARTIES’ ELECTION PLEDGES 13
FIGURE 2 The Probability of Pledge Fulfillment by Governing Party Status
Note: Bars refer to the 95% confidence intervals that pledges are at least partially fulfilled.
Snr. coalition =coalition members that hold the chief executive post; Jnr. coalition =coalition members that
do not. The estimates were derived from Model 2 in Table 2. Other variables are held at their relevant mode (for
categorical variables) or mean (for scale variables) values.
according to Model 2 (Table 3). The variable Agreement
has a positive and significant coefficient, indicating that
pledges are more likely to be fulfilled if they are supported
by more than one of the coalition members.
Figure 2 depicts the key findings as predicted proba-
bilities. The highest probabilities of pledge fulfillment are
found in single-party governments, and there is little dif-
ference between governments with and without legislative
majorities after controlling for other relevant variables,
including government duration. Parties in coalitions are
generally less likely to fulfill their election pledges. Senior
coalition members, which hold the chief executive, are
somewhat more likely to fulfill their election pledges than
junior coalition members.
With the exception of Presidentialism,theinstitu-
tional variables included in the models in Table 2 are
insignificant. The United States is the only presidential
system in the analysis, and the significant negative co-
efficient indicates that pledges in the U.S. presidential
system are significantly less likely to be fulfilled than
pledges in other systems. Semi-presidentialism, bicam-
eralism, federalism, and EU membership do not have sig-
nificant effects on pledge fulfillment. Since Portugal is the
only semi-presidential system in our analysis, the variable
Semi-presidentialism is the same as a dummy variable for
Portugal.
Regarding the other control variables, the signifi-
cant effect of financial resources is noteworthy. Overall,
pledge fulfillment is significantly more likely in times
of economic growth, although the effect is not con-
sistent throughout the subsets of cases. Resources in
terms of time also have a positive effect on pledge ful-
fillment, and we noted that minority governments are
disproportionately short-lived. With respect to govern-
ing experience, Models 1 and 2 of Table 2 indicate that
parties with no prior experience are less likely to fulfill
their election pledges than parties that were incumbents
when they made their election pledges. However, this ef-
fect becomes insignificant in the subset of coalition cases
(Model 2 of Table 3). The coefficients of the other control
variables are insignificant or not robust.
Further robustness tests are reported in the support-
ing information. These include the following additional
tests, among others: (1) models that explore the effect
of minority governments’ agreements with nonexecutive
parties or legislators; (2) models applied to subsets of
cases, such as only the parliamentary cases (excluding the
United States); (3) models with fixed effects for countries;
(4) models with an additional categorization of types of
pledges in a subset of countries; (5) multinomial models
with the three-category dependent variable; and (6) mod-
els with programs as units of analyses, which exclude the
variables that vary within programs. The main findings
highlighted here are robust to these alternative specifica-
tions. Our main finding that single-party governments
do better than coalitions in terms of pledge fulfillment is
robust even in a model similar to Model 1 in Table 2 that
includes country fixed effects.
14 ROBERT THOMSON ET AL.
This fixed-effects model uses only within-country
variation and is therefore very demanding given our data.
When adding the additional controls for Chief execu-
tive and other variables to this fixed-effects model, the
coefficients for Coalition majority and Coalition minority,
although still negative, become insignificant. This could
be due to the lack of sufficient within-country variation
in these variables. It may also be that the difference be-
tween single-party and coalition governments is due to
the fact that single-party governments by definition hold
the chief executive.
Conclusions
We pre s e nte d t h e firs t c ompa r ativ e s tudy o n t he ful f ill-
ment of election pledges across a broad range of countries
and institutional settings. The findings concern one of the
central principles of democratic theory: that parties make
promises to voters during election campaigns and then
fulfill those promises if they enter government office after
elections (Mansbridge 2003, 515). The evidence shows
that parties act according to this principle to a consider-
able extent. Parties that hold executive office after elec-
tions generally fulfill substantial percentages, sometimes
very high percentages, of their election pledges, whereas
parties that do not hold executive office generally find
that lower percentages of their pledges are fulfilled.
The fulfillment of pledges by governing executive par-
ties varies across governments in ways that reflect power-
sharing arrangements. The main power-sharing ar-
rangement that impacts pledge fulfillment distinguishes
between single-party governments and coalitions, not
between governments with and without legislative
majorities. We found the highest percentages of
pledge fulfillment for governing parties in the United
Kingdom, Sweden, Portugal, Spain, and Canada, most
of which governed in single-party executives. We found
lower percentages for governing parties in Germany,
the Netherlands, Austria, Bulgaria, Ireland, and Italy,
most of which governed in coalitions. Pledge fulfill-
ment by U.S. presidential parties lies at the higher end
of coalition governments, which suggests that U.S. pres-
idents are more constrained than governing parties in
single-party parliamentary systems, but less constrained
than most governing parties in multiparty coalitions.
We hope t h at our s t u dy en c ourag e s furth e r empi r i cal
research using the same measures of pledge fulfillment,
which expands the countries and time periods covered
to include more institutional variation and non-Western
democracies.
Our study provides evidence of the effectiveness of
governing parties that do not hold legislative majorities.
Single-party minority governments generally consist of
large, centrally located parties that control the median leg-
islator, which indicates that their minority status is a sign
of strength rather than weakness (Crombez 1996). Our
findings add to previous research that demonstrates that
minority governments and divided government can work
effectively in terms of legislative productivity (Mayhew
2005; Strøm 1990). We also found that negative investi-
ture rules, which are associated with longer tenure, are
also associated with a higher likelihood of pledge fulfill-
ment by parties in minority governments. Future research
could devote more attention to specifying and testing the
impact of legislative procedures on the strength of the
program-to-policy linkage, particularly in research that
includes more minority governments with greater vari-
ation in legislative rules on government formation and
termination.
Which party controls the chief executive explains
some of the variation in pledge fulfillment within coali-
tions, but the evidence indicates that the difference be-
tween holding and not holding this post is not always
large. By contrast, according to some prominent mod-
els of policymaking, the party of the chief executive
has far greater control over policy than do its coalition
partners (Austen-Smith and Banks 1988; Baron 1991;
Dewan and Hortala-Vallve 2011; Diermeier and
Feddersen 1998; Huber 1996; Thies 2001). Moreover, we
do not find clear evidence that a party’s pledges are more
likely to be fulfilled if it obtained the relevant minis-
terial portfolio after the election (cf. Laver and Shepsle
1996). Instead, the findings are more in line with theo-
ries that highlight mechanisms of collective policymak-
ing and constraints on ministers (Dunleavy and Bastow
2001; Warwick 1999). These findings may encourage the-
orists to further specify models of policymaking to iden-
tify the conditions under which the effects of these offices
are most marked. The advantage of holding these posi-
tions may depend on the presence or absence of certain
mechanisms for interministerial coordination or parlia-
mentary control that make chief executives and minis-
ters accountable (e.g., Kassim 2013; Martin and Vanberg
2004).
The effects of government type and economic con-
ditions on pledge fulfillment indicate that parties do not
fully adjust what they promise to the political and eco-
nomic constraints they will encounter in government. If
parties did respond accurately to such expectations, they
would make more modest commitments when they ex-
pect to enter coalitions or when the economy is weaker,
so that the probability of pledge fulfillment would be the
THE FULFILLMENT OF PARTIES’ ELECTION PLEDGES 15
same, regardless of these conditions. While more modest
commitments in these conditions would result in higher
levels of pledge fulfillment, they would not necessarily
serve the democratic process well. Parties make pledges
not only with a view to fulfilling them, but also to serv-
ing another two requirements of representation: bringing
together internal party factions and appealing to voters
during election campaigns.
References
Austen-Smith, David, and Jeffrey S. Banks. 1988. “Elections,
Coalitions and Legislative Outcomes.” American Political
Science Review 82(2): 405–22.
Baron, David P. 1991. “A Spatial Bargaining Theory of Gov-
ernment Formation in a Parliamentary System.” American
Political Science Review 85(1): 137–64.
Budge, Ian, Hans-Dieter Klingemann, Andrea Volkens, Judith
Bara, and Eric Tanenbaum. 2001. Mapping Policy Preferences:
Estimates for Parties, Electors, and Governments, 1945–1998.
Oxford: Oxford University Press.
Budge, Ian, David Robertson, and Derek Hearl, eds. 1987. Ide-
ology, Strategy and Party Change: Spatial Analysis of Post-
War Election Programs in Nineteen Democracies.Cambridge:
Cambridge University Press.
Crombez, Christophe. 1996. “Minority Governments, Minimal
Winning Coalitions and Surplus Majorities in Parliamentary
Systems.” European Journal of Political Research 29(1): 1–29.
Dewan, Torun, and Rafael Hortala-Vallve. 2011. “The Three As
of Government Formation: Appointment, Allocation, and
Assignment.” American Journal of Political Science 55(3):
610–27.
Diermeier, Daniel, and Timothy J. Feddersen. 1998 “Cohe-
sion in Legislatures and the Vote of Confidence Procedure.”
Amer ican Political Science Re view 92(3): 611–21.
Downs, Anthony. 1957. An Economic Theory of Democracy.New
Yo r k : H a r p e r .
Dunleavy, Patrick, and Simon Bastow. 2001. “Modelling Coali-
tions That Cannot Coalesce: A Critique of the Laver-Shepsle
Approach.” Wes t Europ ea n Po lit ic s 24(1): 1–16.
Grossback, Lawrence J., David A. M. Peterson, and James A.
Stimson. 2005. “Comparing Competing Theories on the
Causes of Mandate Perceptions.” American Journal of Po-
litical Science 49(2): 406–19.
Huber, John D. 1996. “The Impact of Confidence Votes on Leg-
islative Politicsin Parliamentary Systems.” Ame rican Polit ical
Science Review 90(2): 269–82.
Huber, John D., and Charles R. Shipan. 2002. Deliberate Dis-
cretion? The Institutional Foundations of Bureaucratic Auton-
omy.Cambridge:CambridgeUniversityPress.
Huber, Evelyne, and John D. Stephens. 2001. Development and
Crisis of the Welfare State: Parties and Politics in Global Mar-
kets.Chicago:ChicagoUniversityPress.
ISSP Research Group. 2008. International Social Survey Pro-
gramme: Role of Government IV - ISSP 2006. GESIS Data
Archive, Cologne. ZA4700 Data file Version 1.0.0.
Kassim, Hussein. 2013. “The Europeanization of Member State
Institutions.” In The Member States of the European Union,
2nd ed., ed. Simon Bulmer and Christian Lequesne. Oxford:
Oxford University Press, 279–312.
Klingemann, Hans-Dieter, Richard I. Hofferbert, and Ian
Budge. 1994. Parties, Policies and Democracy. Boulder, CO:
West v iew P res s .
Klingemann, Hans-Dieter, Andrea Volkens, Judith Bara, Ian
Budge, and Michael McDonald. 2006. Mapping Policy Pref-
erences II: Estimates for Parties, Electors, and Governments
in Eastern Europe, European Union and OECD, 1990–2003.
Oxford: Oxford University Press.
Krukones, Michael G. 1984. Promises and Performance: Presi-
dential Campaigns as Policy Predictors. Lanham, MD: Uni-
versity Press of America.
Laver, Michael, and Kenneth A. Shepsle. 1996. Making and
Breaking Governments: Cabinets and Legislatures in Par-
liamentary Democracies.NewYork:CambridgeUniversity
Press.
Lijphart, Arend. 1999. Patterns of Democracy: Government
Forms and Performance in Thirty-Six Countries.NewHaven,
CT: Yale University Press.
Lowe, Will, Kenneth Benoit, Slava Mikhaylov, and Michael
Laver. 2011. “Scaling Policy Preferences from Coded
Political Texts.” Legislative Studies Quarterly 36(1):
123–55.
Mainwaring, Scott P., and Timothy R. Scully. 1995. “Intro-
duction.” In Building Democratic Institutions: Party Sys-
tems in Latin America, ed. Scott P. Mainwaring and
Timothy R. Scully. Stanford, CA: Stanford University Press,
1–34.
Manin, Bernard. 1997. The Principles of Representative Govern-
ment.Cambridge:CambridgeUniversityPress.
Mansbridge, Jane. 2003. “Rethinking Representation.”
Amer ican Political Science Re view 97(4): 515–28.
Mansergh, Lucy E., and Robert Thomson. 2007. “Election
Pledges, Party Competition and Policymaking.” Compar-
ative Politics 39(3): 311–29.
Martin, Lanny W., and Georg Vanberg. 2004. “Policing the Bar-
gain: Coalition Government and Parliamentary Scrutiny.”
Amer ican Journal of Political Science 48(1): 13–27.
Mayhew, David. 2005. Divided We Govern: Party Control, Law-
making, and Investigations, 1946–2002.2nded.NewHaven,
CT: Yale University Press.
McDonald, Michael, and Ian Budge. 2005. Elections, Parties,
Democracy: Conferring the Median Mandate.Oxford:Oxford
University Press.
McGann, Anthony J., and Michael Latner. 2013. “The Calculus
of Consensus Democracy: Rethinking Patterns of Democ-
racy without Veto Players.” Comparative Political Studies
46(7): 823–50.
Naurin, Elin. 2011. Election Promises, Party Behaviour and Voter
Perceptions.NewYork:PalgraveMacmillan.
Naurin, Elin. 2014. “Is a Promise a Promise? Election Pledge
Fulfilment in Comparative Perspective Using Sweden as an
Example.” West Europ ean Po lit ics 37(5): 1046–64.
Pomper, Gerald M. 1968. Elections in America: Control and
Influence in Democratic Politics.NewYork:Dodd,Meadand
Company.
16 ROBERT THOMSON ET AL.
Pomper, Gerald M., and Susan S. Lederman. 1980. Elections in
America: Control and Influence in Democratic Politics.2nd
ed. New York: Longman.
Powell, Bingham G. 2000. Elections as Instruments of Democracy:
Majoritarian and Proportional Views. New Haven, CT: Yale
University Press.
Powell, Bingham G., and Guy Whitten. 1993. “A Cross-National
Analysis of Economic Voting: Taking Account of the Polit-
ical Context.” American Journal of Political Science 37(2):
391–414.
Rasch, Bjørn Erik, Shane Martin, and Jos´
eAntonio
Cheibub. 2015. Parliaments and Government Format ion :
Unpacking Investiture Rules.Oxford:OxfordUniversity
Press.
Robertson, David. 1976. ATheoryofPartyCompetition.
London: Wiley.
Royed, Terry J. 1996. “Testing the Mandate Model in Britain
and the United States: Evidence from the Reagan and
Thatcher Eras.” British Journal of Political Science 26(1):
45–80.
Saalfeld, Thomas. 2013. “Economic Performance, Political In-
stitutions and Cabinet Durability in 28 European Parlia-
mentary Democracies, 1945–2011.” In Party Governance and
Party De mocracy,ed.WolfgangC.M¨
uller and Hanne Marthe
Narud. New York: Springer, 51–79.
Strøm, Kaare. 1990. Minority Government and Majority Rule.
Cambridge: Cambridge University Press.
Strøm, Kaare, Wolfgang C. M¨
uller, and Torbj ¨
orn Bergman.
2010. Cabinets and Coalition Bargaining: The Democratic
Life Cycle in Western Europe.Oxford:OxfordUniversity
Press.
Sulkin, Tracy. 2009. “Campaign Appeals and Legislative Ac-
tion.” Journal of Politics 71(3): 1093–1108.
Thies, Michael F. 2001. “Keeping Tabs on Partners: The Logic
of Delegation in Coalition Governments.” American Journal
of Political Science 45(3): 580–98.
Thomassen, Jacques J.A. 1994. “Empirical Research into Politi-
cal Representation: Failing Democracy or Failing Models?.”
In Elections at Home and Abroad, Essays in Honor of Warren
Miller,eds.M.KentJenningsandThomasE.Mann.Michi-
gan University Press: Ann Arbor, 237–65.
Thomson, Robert. 2011. “Citizens’ Evaluations of the Fulfill-
ment of Election Pledges: Evidence from Ireland.” Journal of
Politics 73(1): 187–201.
Tsebelis, George. 2002. Ve t o Pla yers : How P oli t ica l I nst i tut i ons
Wor k. Princeton, NJ: Princeton University Press.
Tsebelis, George, and Chang Eric. 2004. “Veto Players and the
Structure of Budgets in Advanced Industrialized Countries.”
European Journal of Political Research 43(3): 449–76.
Warw i ck, Pau l V. 19 9 9. “ Min ist eria l Au to n om y or Mi nis ter ial
Accommodation? Contested Bases of Government Survival
in Parliamentary Democracies.” British Journal of Political
Science 29(2): 369–94.
Supporting Information
Additional Supporting Information may be found in the
online version of this article at the publisher’s website:
The Supporting Information contains details of our de-
liberations that led to our common definition of election
pledge, the reliability tests, and the robustness tests de-
scribed in the analysis section.