Why Do Political Parties Keep Their Promises
5. MEASURING HOW POLITICAL PARTIES KEEP THEIR PROMISES1
François Petry and Benoît Collette
November 2008
Abstract
This chapter addresses three questions about the relation between political discourse and action: Do political parties keep their promises once elected? What are the methodologies used by scholars to demonstrate that political parties keep (or do not keep) their campaign promises? Are these methodologies valid and reliable? We answer these questions based on a review of 18 journal articles and book chapters published in English and French over the past forty years that report quantitative measures of election promise fulfillment in North America and Europe. We find that parties fulfill 67 percent of their promises …show more content…
on average, with wide variation across time, countries, and regimes. Most studies have major methodological weaknesses (no operational definition, no mention of relevant documentation, flawed research design) although the more recent ones tend to show higher levels of methodological sophistication and a modicum of scientific transparency.
The extent to which government actions fulfill election promises as a theoretical issue has raised an important scholarly debate. It is also an empirical issue that raises methodological debates. Despite the relative pertinence of such a question in representative democracy from both normative and positive perspectives, there is surprisingly few studies addressing it. The objective of this paper is to contribute to these debates by examining how the relevant scholarly literature assesses the relationship between campaign promises and government actions. More specifically, we ask the following basic questions. Do political parties keep their campaign promises once elected? What are the methodologies used by different scholars to demonstrate that political parties keep (or do not keep) campaign promises? Are these methodologies valid and reliable? The literature review will focus on 18 journal articles and book chapters published in English and French over the past forty years that report quantitative measures of election promise fulfillment. From this review, we try to draw some general patterns from a variety of empirical sources. We determine which methodologies contribute to differences across studies, and we also identify areas that have been neglected and warrant further investigation.
1. Theory: Why should we expect political parties to keep their election promises?
Ask people around you if they think that political parties keep their electoral promises and you will probably obtain a high rate of negative answers. For example, an overwhelming majority of respondents in a survey recently carried out by one of the authors in a large undergraduate political science class thought that electoral promises have “little or no importance” when it comes to know what a party will do once elected in office. Only 10% thought that election promises are “very important”. Although not entirely scientific, these lay opinions seem to coincide well with the conventional academic interpretation that party platforms bear little relationship to what a government will do eventually, and, consequently that election promises are not a significant element in the democratic debate. American political scientist E.E. Schattschneider (1942) argued long ago that “party platforms are fatuities. They persuade no one, deceive no one, and enlighten no one.” More recently, British political scientist Anthony King asserted that party manifestos are “empty and meaningless” documents having “virtually random relationship” with what the party will do in office (cited in Rose 1984). Davis and Ferrantino (1996) have even developed a positive theory of political rhetoric which predicts that political candidates will lie because they are unable to transfer the value of their reputations as honest politicians.
These negative pronouncements about the ability of governments to keep their election promises should not obscure the fact that several important theories hold the exact opposite, that politicians and political parties keep their campaign promises. There is first a normative theory of campaign promises. This theory states that political parties should follow the moral requirement of ‘decency’ which includes the following rules (Schedler 1998). First, avoid making promises which one knowingly cannot keep (realism criterion). Second, avoid making promises which one does not intend to keep (sincerity criterion). Third, avoid making contradictory promises (consistency criterion). The only exceptions to these moral rules are the occurrence of unforeseen events (natural disaster, economic crisis, war, unanticipated shifts public opinion) which allow a government to renege its moral obligation to keep its election promises.
Another theory of why politicians keep their campaign promises is the mandate theory of election. This theory comes from the positivist school of thoughts, and it is the one most frequently encountered in the literature. According to the theory political parties make specific pledges in their election platforms and they try to fulfill as many pledges as possible once elected in power. Hofferbert and Budge (1992) use the analogy of an architectural blueprint to illustrate this central idea of the theory. The theory is based on two postulates. First, the competing parties offer platforms that are quite distinct from each other; the differences between the platforms of the parties are sufficiently salient to allow the voters to make a rational non-arbitrary choice with a minimum of information. Party differences, and the consequences of electing one party rather than the other, are easily discernable by voters and the information costs of discovering these are small. Second, the voters prospectively compare the utilities provided by competing parties and they give their vote to the party which offers them the greatest utility.2 Based on these postulates, the theory predicts that the winning party carries through the platform on which it has been elected. The logic underlying this expectation is straightforward: The issues advocated by the party in government are the winning issues that contributed in getting the party elected in the first place. It is therefore rational for a utility-maximizing party to carry out its election promises. Another reason for keeping election promises is to avoid retaliation by disappointed voters at the next election.3
Although we have not found it in published form, it is also possible to build an explanation of why politicians keep their campaign promises following the constructivist approach. As with mandate theory, the constructivist approach would argue that there is a strong congruence between campaign pledges and subsequent government actions. However, this would be not because politicians find it rational to keep their promises but because the political discourse of campaign promises and of government actions goes through a same process of social construction of meaning (Faure et al. 1995). Unlike the mandate theory (and unlike positivist theories in general) which assumes that a true external reality can be discovered through the scientific method, constructivism postulates that we can only grasp different subjective constructions of reality produced by different people. Therefore, constructivism does not lead to empirically testable propositions. However, it is still a useful approach in that it challenges the positivist approach by raising important interrogations on the ambiguous nature of campaign pledges, on the relation of causality between campaign pledges and government actions, and on the meaning that the governing elites give to election promises in order to justify post hoc their political decisions. We will take up this thread in conclusion.
We therefore have a several theories that support the idea that politicians keep their promises, and that the pledges that parties offer in their election platforms do make a difference in subsequent policy making. This is against the background of the null hypothesis that election platforms make no difference. In what follows, we review how different scholars have tested this null hypothesis. To conduct this review, we will ask two broad questions, one empirical, and one methodological:
To what extent are campaign pledges subsequently redeemed? In other words, what is the level of congruence between pledges and actions?
How do we know when a campaign pledge is redeemed or not? In other words, to what extent do the studies we review present results that are valid and replicable?
2. Method: How to Test Whether Campaign Pledges Are Redeemed
There exist at least three distinct methods to test empirically the predictive value of campaign promises. These can be classified along a continuum measuring the extent to which researchers must use their own judgement in assessing when a pledge is redeemed. At one end of the continuum we find correlational studies matching the variation in some objectively quantifiable measure of policy output (roll call votes) in one or more policy domains with quantitative measures of pre-election policy preferences by parties or candidates (David 1971, Ringquist and Dasse 2004). These studies require little or no subjective interpretation of the data. They maximize external validity and reliability of the data in both the independent and the dependent variables. But the method sacrifices the substance of the policies under analysis. In other words, one loses in terms of internal validity what one gains in terms of reliability. Moreover, since the variables that this methodology requires (roll call votes and surveys-based evidence of politicians’ pre-election stance on issues) are available only in the US, the methodology is not very useful in the context of comparative research.
Next along the continuum we find studies that try to combine the advantage of a correlational design while not sacrificing the substance of the policies under analysis. The method consists of correlating variations in some measure of government output (public expenditures in specific policy domains) with how much space has been devoted by the winning party to each domain in its election platform. This method has been used by scholars associated with the Comparative Manifesto Project (CMP), a standing research group of the European Consortium for Political Research (ECPR). The initial objective of the CMP project was (and still is) to record and analyze the contents of the election platforms in democratic countries since World War 2. These data are then used to position the parties in their respective national political space and track their evolution from one election to another (Volkens 2002). The election platforms are coded into a pre-established set of policy categories. The score of each category is then calculated to reflect the relative emphasis of each category in each party platform. The results of this coding have been used as a basis for the empirical tests of the predictive power of campaign pledges. Annual budget expenditures are measured in various policy sectors that match the categories used to code the election platforms. The actual test is conducted by correlating the amounts of public spending (in percentage of GDP) with election platform emphases
Empirical test of this type were carried out in a comparative volume (Klingemann et al. 1994) and in single-country studies in Germany (Hofferbert and Klingemann 1990), Canada (Petry 1988, 1995), the US (Budge and Hofferbert 1990), France (Petry 1991), Britain (Hofferbert and Budge 1992) with somewhat mixed results. Budge and Hofferbert (1990), Hofferbert and Budge (1992) and Klingemann et al. (1994) report high correlations between government spending and election promises for the most part. These positive results are suspect, however. King and Laver (1993), in their replication of the Hofferbert and Budge (1990) study of US data have pointed out major flaws in the Budge-Hofferbert methodology, including the failure to report the standard errors for the regression coefficients; the failure to measure the effect of the time trend on changes in public spending and the absence of test of autocorrelations. When these elements are included in the regression equations, there is no longer a significant correlation between public expenditures and the election platforms of the winning party. Thome (1999) goes further in his methodological critique by showing that the results by Budge and Hofferbert (1990, 1992) and by Klingemann et al. (1994) are irremediably tainted by their failure to include several additional parameter restrictions required by the theoretical model.
Another difficulty with the Budge-Hofferbert method is its reliance on the selective emphasis methodology which consists of only recording the space devoted by a party to a particular policy category in its platform, without differentiating negative from positive party attitudes toward issues. Hofferbert and Budge (1992) praise the selective emphasis methodology on the ground that stressing a particular priority implies an intent to take only positive action in the relevant policy domain, for example by increasing public expenditures or passing legislation. However this is not always the case. A party may very well emphasize (prime) a particular issue in its platform in order to discuss (frame) it negatively, and this may have a measurable effect on policy. Therefore, prudence would recommend that we separate positive from negative emphases (frames). The method has also been criticized for its narrow conception of public policy outputs. By focusing solely on the budgetary expenditures, we miss a large array of state activities and outputs —including laws, administrative decisions, speeches— that are directly relevant to the question of whether governments keep their election promises.
The method that we find at the other end of the continuum consists in counting specific pledges in election platforms and then examining the record of government actions in order to determine how many pledges have been redeemed. Scholars who use the pledge method are obviously going to be better able to control the internal validity of their research design. Royed (1996) and Royed and Borelli (1997, 1999) are strong advocates of this method which, unlike the other ones, does not sacrifice the substance of policy. This is part of the reason why the method has been more widely used by a larger circle of scholars than the other two methods. Another reason for the relative popularity of the pledge method among scholars is that it produces data in the form of percentages of pledges redeemed that are simpler to interpret than the regression coefficients produced by the two other methods. But there is always the risk that these data are invalid because they rely on a subjective interpretation of whether a promise is kept. This is why it is so important to establish whether and to what extent the studies that use the pledge method and present their results in the form of percentages of redeemed pledges contain valid and replicable proofs of what they pretend to demonstrate. This is what we set out to do in the next section.
3. Analysis
After an exhaustive bibliographic search, we have been able to identify 18 separate studies comprising 21 cases that present results in the form of percentages of redeemed campaign pledges.4 Table 5.1 reports the data. Each case is identified by the author and the year of publication of the study, the country (or countries), the period of analysis, and the average percentage of pledges redeemed. All the studies reported in the table calculate percentages of promises that are redeemed, but the validity of the method of calculation varies greatly from one study to the next. We have therefore added four criteria intended to give some idea of the validity of the calculation/demonstration in each study. The entries for these criteria are in the form of binary answers to four simple questions:
Does the study contain an operational definition of a campaign pledge? (Yes or No). A definition is considered operational here if it contains explicit criteria of exclusion of what a pledge is not. Another condition is that the documentary sources of campaign pledges are precisely referenced.
Does the study contain an operational definition of government action/output? (Yes or No). A definition is considered operational here if it contains explicit criteria of exclusion of what a relevant government action/output is not. Another condition is that the documentary sources of government actions are precisely referenced.
What is the extent of the documentation of government outputs with which campaign pledges are matched? (Large or Small). For example, are campaign pledges matched only with laws, or are they also matched with throne speeches, with budgets, with annual reports from various ministries?
How precise, replicable and valid is the demonstration that a pledge is fulfilled or not? Or put in more simple terms, how much room is left to a researcher’s own judgment?
Our objective is not to measure the overall quality of these studies. That will require an analysis which goes well beyond the scope of this chapter. This research is a first step to check out the availability of all the elements that are required to conduct a meta-analysis.
Table 5.1 about here
Only five out of the 21 studies reviewed meet the four methodological criteria. Not surprisingly, the more recent the study is, the better overall score it gets. The five studies that received a perfect score have all been published within the past twelve years. It is not until 1996, with the first study by Terry Royed, that explicit definitions of what a pledge is are systematically provided. The systematic occurrence of operational definitions of government action is even more recent. The usefulness of an operational definition of campaign pledges and government actions is obvious in any comparative exercise: The broader the definition of pledges, the smaller the expected proportion of pledges that will be fulfilled. Conversely, the broader the definition of government actions, the larger the proportion of pledges that are expected to be redeemed. Thus, one needs an operational definition if one is to validly compare the proportion of fulfilled pledges across case studies. It was not until the late 1980s that the documentary sources of campaign pledges and government actions were fully referenced. This criterion is important because it makes it possible to judge and compare the quality and the diversity of the documents used to measure pledge fulfillment in different studies. The final criterion of whether there is a valid proof that pledges are redeemed is the least frequently met. Aside of the work by David (1971), only the most recent studies satisfy this criterion enough to deserve a mark in the table.
Let us now turn to a more detail description of each study. We start with the study by Pomper (1968) —the oldest one on the list— and its updated version (Pomper and Lederman 1980). However innovative, Pomper’s work fails to provide fully operational definitions of pledges and government actions (at least based on our definition). The assessment of pledge fulfillment is based almost exclusively on a volume entitled Congress and the Nation: 1945-64 and the subsequent publications of Congressional Quarterly Service. The demonstration that a particular pledge is redeemed relies on a typology which involves five categories: full action (passage of a law), executive action, similar action (indirect action by the executive or legislative branch), negative fulfillment, defeated (the law did not pass), and no action (status quo). This typology finely separates unfulfilled from partially or fully fulfilled pledges. However, the criteria and method that Pomper uses to separate fulfilled from unfulfilled pledges are not fully explained. It is therefore difficult to replicate his work. The same diagnosis applies to the work by David (1971) who expands and rearranges Pomper’s 1968 results. Even though he remains entirely uncritical of Pomper’s methodology, David points out that his and Pomper’s work still contain many ambiguities and that it “must be regarded as the beginning rather than the end of the research that is needed” (1971, 311).
Bradley (1969) identified pledges having to do with social security in Democratic and Republican platforms from 1932 to 1964 and determined whether these were carried out in the form of law. The Bradley article gives a detailed account of which campaign promises were fulfilled and which were not. There is also a discussion of policy changes that had not been previously proposed in Democratic platforms. Although we are able to calculate the proportion of social security platforms planks that were fulfilled, the information discussed in this article is essentially qualitative, with little concern about operationalization and measurement of variables.
Elling’s (1979) research is an uncritical replication of Pomper’s work at the state level. Elling compares the way campaign pledges are redeemed in Illinois (where parties are said to be more pragmatic) and Wisconsin (more ideological parties) respective governments. His methodology is almost identical to Pomper’s. There are no truly operational definitions of campaign pledges and government actions, and no explicit criteria to decide which pledges are fulfilled and which are not. Elling finds that 50% of campaign pledges are redeemed in Illinois, and 45% in Wisconsin. These are the lowest percentages among the 21 studies reviewed here.
The next work is Rose’s study (1984) of how British governments fulfilled their election pledges in the 1970s. This work is probably the least sophisticated of the bunch in terms of our criteria, and therefore the hardest to assess and replicate. There is no attempt at providing explicit definitions of campaign pledges and government actions. There is not clear reference to the archival sources that were analyzed. Neither is there a demonstration of how a pledge is declared fulfilled either completely or partially. In Rose’s defense, his study of whether and how pledges are fulfilled is not the primary purpose of his book. The main goal of Rose was to analyze the influence of Britain’s political parties on politics, and measuring the fulfillment of election pledges was only a part of this objective.
Michael Krukones (1984) measures how US presidents have kept their electoral promises between 1912 and 1976. Krukones’ work is the first to exhaustively document its sources for government action. There are attempts at providing operational definitions of campaign pledges and presidential actions, although they fall short of our criteria. Krukones also innovates by weighting unfulfilled presidential pledges based on whether or not they represent “good faith” efforts to pass a policy that failed due to factors beyond presidential control. However, the methodology for deciding what constitutes a good faith effort and what does not is never explicitly presented. In his study of how US presidents fulfilled their campaign pledges from 1960 to 1984, Fishel (1985) goes one step further in measuring good faith efforts by assessing whether presidential promises need congressional approval to be fulfilled.
Rallings (1987) compares how British and Canadian governments have kept their campaign promises in the 1970s and 1980s. There are no clear operational definitions of pledges and government actions in this work. It is stipulated that to be counted as such, a pledge must anticipate some future action by the governments and not only intentions. However, how actions are distinguished from intentions is not specified. The documentary basis of government actions includes laws, budget speeches, and possibly other sources. But these sources are not precisely identified and referenced. As other studies that preceded it, Rallings’ work makes no reference to explicit criteria on which to decide that a campaign promise is carried out or not. The results for Britain give an average score of 63.7%, against 71.5% for Canada. However, these numbers are the result of a direct match of government actions with the content of throne speeches only. A direct match of government action with election pledges is nowhere to be found in Rallings’ work. This undermines somewhat the purpose of the study.
Next in chronological order comes the study by Monière (1988) about pledge fulfillment in the first mandate of the Conservative government of Brian Mulroney. Unlike Rallings or Rose, who match government output with pledges from party platforms only, Monière matches government output with party platforms and with campaign speeches by party leaders that are covered in national newspaper. Campaign pledges are matched with legislative documents only (laws and House of Commons Hansard). Monière concludes that 74% of campaign pledges are redeemed on average. The number climbs to 80% when non-verifiable pledges are excluded from the analysis. However, once again, there is little explicit discussion of the criteria for deciding which pledges are verifiable and which are not.
Kalogeropoulou’s (1989) study of Greece’s PASOK government is directly inspired by Rose’s work. Unlike Rose, however, Kalogeropoulou makes detailed references to the pre- and post-election documents he uses to assess if pledges are redeemed. These documents come in a large variety: partisan, legislative, executive, and administrative sources are consulted as well as media reports. Unfortunately, Kalogeropoulou does not explain clearly how he uses these documents and on what criteria he relies in order to decide if the government fulfills its pledges.
The studies by Royed (1996) and Royed and Borelli (1997, 1999) are considerably more sophisticated methodologically than the previous work. They are the first ones to provide a clear definition of pledges (although government actions are still left largely undefined). A large variety of documentary sources are consulted to determine whether a pledge is fully or partially redeemed. In the first article, Royed does not specify what documents had been used to operationalize government output, but she does it for the 1997 and 1999 articles with Borelli (CQ reports and statistical abstracts). These studies also go a long way toward an explicit demonstration of how some pledges are declared redeemed while others are not. But the demonstration remains incomplete and, one suspects, there is still a certain amount of subjective interpretation underlying the methodology of deciding whether a pledge is fulfilled or not. The 1996 study by Royed found that 85% of pledges were redeemed in Great Britain under Prime Minister Thatcher (the highest proportion of the 21 studies reviewed here) against only 52% in the US under President Reagan. (the second lowest proportion).
Thomson (2001) analyzed the platform-to-policy linkage in the Netherlands by measuring the fulfillment of election pledges on socio-economic policy. By all standards this article is well crafted. To dig out pledges from party manifestos Thomson used the same method as Royed did and came out with a rate of fulfillment of 61%. But contrary to Royed, whose definition of pledge is operationalized by the presence of an outcome, Thomson was more selective and restricted his definition to policy actions. To assess the fulfillment of election pledges, a CD-ROM database containing references to, and a short description of, all government decisions has been used as the main source. The dependant variable can take three values: not fulfilled, partially fulfilled and fully fulfilled, according to the degree of congruence between government decisions and pledges. A partially fulfilled pledge means that some policy may be taken in the direction indicated by the pledge, but falling short of full realization. One of the most interesting aspects of Thomson’s work is a counter-verification of pledge fulfillment by a panel of experts. A sample of 110 pledges was judged by area specialists and the inter-coder reliability, measured by a Cohen’s Kappa coefficient of 0.70, is quite good.
Petry (2002) finds that 75% of the pledges in the platform of the Parti québécois in the 1994 and 1998 Quebec elections were fulfilled over the period 1994-2000. As with the work by Royed (Royed 1996) and Thomson, Petry’s work provides operational definitions of election pledges and government actions. Government actions are based on a large variety of documentary sources (laws, internal party documents, annual reports by ministries, budget speeches, media reports). Unlike Thomson, there is no measure of inter-coder reliability. However, and this is a novelty, the method of linking pledges to government actions tries to be as neutral and objective as possible. It simply consists of reporting whether a documentary source declares that a pledge has been fulfilled or not. This leaves nothing, in theory, to the researcher’s own judgment. A similar method is used by Petry and Collette (2006) in their study of how the Liberal government of Jean Charest fulfilled its pledges after the 2003 Quebec election. Petry’s (2002) operational definition of pledges and government actions remain unchanged. However, the documentary basis has been extended considerably in the Petry and Collette study.
We close the list with a recent study by Ringquist and Dasse (2004) linking pledges to policy in environmental policy in the US. Instead of testing directly whether the pledges in party platforms are redeemed, they record individual Congress members’ scores in the National Political Awareness Test (NPAT).5 These are then matched with the results of congressional roll call votes on environmental issues. The next step was to integrate environmental policy promises into four different probit regression models, along with other independent variables such as campaign contributors, gender, race, etc. In the four models, the promise variable had a positive and significant effect on the dependent variable. Ringquist and Dasse’s decision to rely on NPAT survey results instead of party platform pledges and on Congressional roll call votes instead of the content of policy speeches or laws revives a tradition that was open thirty five years ago by David (1971).
4. Conclusion and discussion
In response to the first question: “do political parties keep their campaign promises once elected?” our review of 21 cases in 18 separate published studies reveals that parties fulfill 67% of their promises on average. Contrary to popular belief, political parties are reliable promise keepers. Why people underestimate the capacity of political parties to keep their election promises remains an open research question. But it is reasonable to conjecture that this is due in part to a bias in media coverage of how parties keep their promises. Stories of broken party promises on a few important issues have considerably more readership appeal and salience in the public than the coverage of pledges fulfilled on many less important issues.
Our positive finding does not go without important caveats. One is the wide variation in the rate of pledge fulfillment by political parties, from a minimum of 45% (Elling 1979, in the state of Wisconsin) to a maximum of 85% (Royed 1996, in Britain) and a standard deviation of 10.3 percentage points. Clearly, some parties tend to keep their electoral promises more than others. Although this chapter was intended as a literature review, not a meta-analysis, we cannot resist noticing two interesting regularities in the data of table 5.1 that shed light on the question of what factors determine the rate of pledge fulfillment. The first noticeable pattern is institutional. The pattern contrasts US cases with a low average rate of fulfilled promises (65%) and cases from Britain and Canada (Quebec included) with a significantly higher average rate of pledges fulfilled (74%). Although we cannot be sure in the absence of a multivariate statistical test, it is reasonable to conjecture that, other things being equal, parliamentary regimes like the Westminster systems of Britain and Canada, positively influence the likelihood that political parties keep their electoral promises once elected because they give the government the latitude to do so. By contrast, separation of powers in a presidential regime like the US limits the latitude of the executive to keep its promises, and would therefore have a negative effect on the rate of pledge fulfillment.6
The high rates of pledge fulfillment that we find are also limited somewhat by methodological caveats. In response to the question of what methodologies are used to demonstrate that parties keep their election promises, our review has uncovered wide methodological differences across the 18 studies. Recent studies are more sophisticated methodologically than studies conducted in the past, some of which fail to provide the information that would be necessary for even the most basic replication. One interesting pattern emerging from the data of table 5.1 suggests that the rate of pledge fulfillment varies in inverse proportion with the severity of the tests. Although we cannot be sure that the tendency would sustain a multivariate test, the seven cases at the bottom of table, those satisfying the most severe tests, have an average rate of fulfillment of 63%, more than ten percentage points lower than the 74% average for the rest of the data. This might reflect a tendency for political parties to be less reliable in recent years. But there is another likely explanation involving methodological aspects. Party promises in recent elections have tended to be more detailed and precise, and therefore better falsifiable, than in past elections. At the same time, due to an increase over time in the severity of the tests applied by researchers, the rate of pledges that are declared fulfilled has tended to decrease in recent studies. Thus, there probably is a relationship between the number of election promises that are kept and the methodology that researchers use to prove their case.
References
Bradley, J. P. 1969. Party Platforms & Party Performance concerning Social Security. Polity 1(3): 337-358.
Budge, I. and R. I. Hofferbert. 1990. Mandates and Policy Outputs: U.S. Party Platforms and Federal Expenditures. American Political Science Review 84(1): 111-131.
David, P. T. 1971. Party Platforms as National Plans. Public Administration Review 31(3): 303-315.
Davis, M. A. and M. Ferrantino. 1996. Towards a Positive Theory of Political Rhetorics: Why Do Politicians Lie? Public Choice 88(1-2): 1-13.
Downs, A. 1957. An Economic Theory of Democracy. New York: Harper.
Elling, R. C. 1979. State Party Platform and State Legislative Performance: A Comparative Analysis. American Journal of Political Science 23(2): 383-405.
Faure, A., G. Pollet, and P. Warin, ed. 1995. La construction du sens dans les politiques publiques : débats autour de la notion de référentiel. Paris: L 'Harmattan.
Fishel, J.
1985. Presidents and Promises: From Campaign Pledge to Presidential Performance. Washington: Congressional Quarterly Press.
Hofferbert, R. I. and I. Budge. 1992. The Party Mandate and the Westminster Model: Election Programmes and Government Spending in Britain, 1945-85. British Journal of Political Science 22(2): 151-182.
Hofferbert, R. I. and H.-D. Klingemann. 1990. The Policy Impact of Party Programmes and Government Declarations in the Federal Republic of Germany. European Journal of Political Research 18(277-304.
Kalogeropoulou, E. 1989. Election Promises and Government Performance in Greece: PASOK’s Fulfilment of its 1981 Election Pledges. European Journal of Political Research 17(289-311.
King, G. and M. Laver. 1993. Party Platforms, Mandates, and Governement Spending. The American Political Science Review 87(3): 744-750.
Klingemann, H.-D., I. Budge, and R. I. Hofferbert. 1994. Parties, Policies, and Democracy. Boulder: Westview Press.
Krukones, M. G. 1984. Promises and Performance: Presidential Campaigns as Policy Predictors. Lanham, Md: University Press of America.
Monière, D. 1988. Le discours électoral. Les politiciens sont-ils fiables ? Montréal: …show more content…
Québec/Amérique.
Petry, F. 1988. The Policy Impact of Canadian Party Programs: Public Expenditure Growth and Contagion from the Left. Canadian Public Policy-Analyse de politiques 14(4): 376-389.
Petry, F. 1991. Fragile Mandate: Party Programmes and Public Expenditures in the French Fifth Republic. European Journal of Political Research 20(149-171.
Petry, F.
1995. The Party Agenda Model: Election Programmes and Governement Spending in Canada. Canadian Journal of Political Science/Revue canadienne de science politique 28(1): 51-84.
Petry, F. 2002. La réalisation des engagements du Parti québécois : analyse d 'ensemble. In Le Parti québécois. Bilan des engagements électoraux 1994-2000, ed. Petry, F., 161-181. Québec: Presses de l 'Université Laval.
Petry, F. and B. Collette. 2006. Le gouvernement Charest a-t-il respecté ses promesses ? In Le Parti libéral. Enquête sur les réalisations du gouvernement Charest, ed. Petry, F., É. Bélanger and L. M. Imbeau, 83-101. Québec: Presses de l 'Université Laval.
Pomper, G. M. 1968. Elections in America. Control and Influence in Democratic Politics. Toronto: Dodd, Mead & Company.
Pomper, G. M. and S. S. Lederman. 1980. Elections in America. Control and Influence in Democratic Politics. New York: Longman.
Rallings, C. 1987. The Influence of Election Programmes: Britain and Canada 1945-79. In Ideology, Strategy, and Party Change: Spatial Analysis of Post-War Elections Programmes in Nineteen Democracies, ed. Budge, I., D. Robertson and D. Hearl, Cambridge: Cambridge University
Press.
Ringquist, E. J. and C. Dasse. 2004. Lies, Damned Lies, and Campaign Promises? Environmental Legislation in the 105th Congress. Social Science Quarterly 85(2): 400-419.
Rose, R. 1984. Do Parties Make a Difference? Chatam: Chatam House Publishers.
Royed, T. J. 1996. Testing the Mandate Model in Britain and in the United States: Evidence from the Reagan and Thatcher Eras. British Journal of Political Science 26(1): 45-80.
Royed, T. J. and S. A. Borelli. 1997. Political Parties and Public Policy: Social Welfare Policy from Carter to Bush. Polity 29(4): 539-563.
Royed, T. J. and S. A. Borelli. 1999. Parties and Economic Policy in the USA. Pledges and Performance, 1976-1992. Party Politics 5(1): 115-127.
Schattschneider, E. E. 1942. Party Governments. New York: Rinehart & Winston.
Schedler, A. 1998. The Normative Force of Electoral Promises. Journal of Theorical Politics 10(2): 191-214.
Thome, H. 1999. Party Mandate Theory and Time-Series Analysis: A Methodological Comment. Electoral Studies 18(4): 569-585.
Thomson, R. 2001. The Programme to Policy Linkage: The Fulfilment of Election Pledges on Socio-Economic Policy in the Netherlands, 1986-1998. European Journal of Political Research 40(171-197.
Volkens, A. 2002. Manifesto Coding Instructions (Second Revised Edition). Berlin: Wissenschaftszentrum Berlin für Sozialforschung.
Table 5.1: Overview of Publications that Measure how Political Parties Keep their Promises
Authors
Year of Publication
Country
Time Period
Avg. % Pledges Fulfilled
Operational Definition of Pledge
Operational Definition of Government Action
Referenced Sources for Government Action
Explicit Proof that a Pledge is Redeemed
Pomper
1968
US
1944-1966
72
Bradley
1969
US
1932-1964
80
David
1971
US
1944-1966
72
X
Elling 1
1979
Illinois
1947-1971
50
Elling 2
1979
Wisconsin
1947-1971
45
Pomper & Lederman
1980
US
1944-1978
69
Rose
1984
Britain
1970-1979
80
Krukones
1984
US
1912-1976
71
X
Fishel
1985
US
1960-1980
61
X
Rallings 1
1987
Canada
1945-1979
72
X
Rallings 2
1987
US
1945-1979
64
X
Monière
1988
Canada
1984-1988
74
X
X
Karogelopoulou
1989
Greece
1981-1985
70
X
Royed 1
1996
Britain
1979-1988
85
Royed 2
1996
US
1980-1988
52
X
Royed & Borelli
1997
US
1977-1992
60
X
X
X
Royed & Borelli
1999
US
1976-1992
60
X
X
X
Thomson
2001
Netherlands
1986-1994
61
X
X
X
X
Petry
2002
Quebec
1994-2000
75
X
X
X
X
Ringquist & Dasse
2004
US
1997-1999
73
X
X
X
Petry & Collette
2006
Quebec
2003-2006
60
X
X
X
X
Average
67
7/21
7/21
12/21
4/21
Endnotes
The authors wish to thank Gerald Miller for his stimulating comments on an earlier version of this paper.
2 The two postulates of the mandate theory, although inspired by Downs’ (1957) economic theory of elections, are somewhat at odds with Downs’ theory. Unlike the mandate theory which assumes that party platform differ, Downs’ theory assumes that politicians offer platforms that converge toward the median voter equilibrium. In a context of identical (or very similar) party platforms, the information costs are very high, thereby precluding voters from making prospective assessments of what the winning party will do after the election. Rational voters do not differentiate between the policy priorities of the competing parties. They will support a party on the basis of heuristic shortcuts such as their retrospective assessment of the performance of the incumbent in office, or party ideology. Therefore, party manifestos should not be expected to be of much significance in elections.
3 Another positive theory of why politicians keep their campaign promises is Marxism. Marxism considers that the political agenda is controlled by a dominant elite working on behalf of capitalist interests. It is the same ruling elite that is at the origin of election programme discourse and of the decisions of the government. The congruence between the electoral discourse and the government action discourse is no coincidence according to Marxist theory: It should be expected because the two discourses are echoes of one another. According to Marxist theory, party platforms —at least the platforms of the more conventional non-extremist parties— are used by the ruling elite as a ploy to manipulate the popular classes into believing that they have a choice at election time. Ultimately, election promises are used to defuse potential popular uprising against the capitalist state’s chosen policy direction. From a Marxist perspective, the study of the correlation between election promises and government actions is scientifically futile. Marxist theory is therefore of no use to someone interested in studying whether governments do what they say.
4 The number of cases is greater than the number of studies because some studies display more than one case.
5 There are 12 questions about specific policy proposals in the NPAT. Ringquist and Dasse recoded the answers in a new binary variable. A score of 0 means an anti-environmental response and 1 for a pro-environmental response.
6 The conjecture is explicitly laid out by Royed (1996). Her study only applies to President Reagan in the US and Prime Minister Thatcher in Britain. Her conclusion that election promises are kept more often in a parliamentary regime than in a presidential regime cannot, therefore, be considered generally valid until tested across a larger sample of countries.
Index
Britain 8, 15, 16, 28, 32
Canada 8, 16, 20, 28
Campaign pledges …………………7, 12
Comparative Manifesto Project (CMP) 8
France 8
Germany 8
Greece 16, 23, 28
Mandate theory of elelctions ……... 5, 11
Netherlands 17, 25, 29
Quebec 18, 20, 29, 30
Reliability 7, 18
Roll call votes 7, 19
Selective emphasis 9
US 7, 8, 15, 19, 20, 27, 28, 29, 32
Validity 7, 10, 11
François Petry and Benoît Collette
November 2008
Abstract
This chapter addresses three questions about the relation between political discourse and action: Do political parties keep their promises once elected? What are the methodologies used by scholars to demonstrate that political parties keep (or do not keep) their campaign promises? Are these methodologies valid and reliable? We answer these questions based on a review of 18 journal articles and book chapters published in English and French over the past forty years that report quantitative measures of election promise fulfillment in North America and Europe. We find that parties fulfill 67 percent of their promises …show more content…
on average, with wide variation across time, countries, and regimes. Most studies have major methodological weaknesses (no operational definition, no mention of relevant documentation, flawed research design) although the more recent ones tend to show higher levels of methodological sophistication and a modicum of scientific transparency.
The extent to which government actions fulfill election promises as a theoretical issue has raised an important scholarly debate. It is also an empirical issue that raises methodological debates. Despite the relative pertinence of such a question in representative democracy from both normative and positive perspectives, there is surprisingly few studies addressing it. The objective of this paper is to contribute to these debates by examining how the relevant scholarly literature assesses the relationship between campaign promises and government actions. More specifically, we ask the following basic questions. Do political parties keep their campaign promises once elected? What are the methodologies used by different scholars to demonstrate that political parties keep (or do not keep) campaign promises? Are these methodologies valid and reliable? The literature review will focus on 18 journal articles and book chapters published in English and French over the past forty years that report quantitative measures of election promise fulfillment. From this review, we try to draw some general patterns from a variety of empirical sources. We determine which methodologies contribute to differences across studies, and we also identify areas that have been neglected and warrant further investigation.
1. Theory: Why should we expect political parties to keep their election promises?
Ask people around you if they think that political parties keep their electoral promises and you will probably obtain a high rate of negative answers. For example, an overwhelming majority of respondents in a survey recently carried out by one of the authors in a large undergraduate political science class thought that electoral promises have “little or no importance” when it comes to know what a party will do once elected in office. Only 10% thought that election promises are “very important”. Although not entirely scientific, these lay opinions seem to coincide well with the conventional academic interpretation that party platforms bear little relationship to what a government will do eventually, and, consequently that election promises are not a significant element in the democratic debate. American political scientist E.E. Schattschneider (1942) argued long ago that “party platforms are fatuities. They persuade no one, deceive no one, and enlighten no one.” More recently, British political scientist Anthony King asserted that party manifestos are “empty and meaningless” documents having “virtually random relationship” with what the party will do in office (cited in Rose 1984). Davis and Ferrantino (1996) have even developed a positive theory of political rhetoric which predicts that political candidates will lie because they are unable to transfer the value of their reputations as honest politicians.
These negative pronouncements about the ability of governments to keep their election promises should not obscure the fact that several important theories hold the exact opposite, that politicians and political parties keep their campaign promises. There is first a normative theory of campaign promises. This theory states that political parties should follow the moral requirement of ‘decency’ which includes the following rules (Schedler 1998). First, avoid making promises which one knowingly cannot keep (realism criterion). Second, avoid making promises which one does not intend to keep (sincerity criterion). Third, avoid making contradictory promises (consistency criterion). The only exceptions to these moral rules are the occurrence of unforeseen events (natural disaster, economic crisis, war, unanticipated shifts public opinion) which allow a government to renege its moral obligation to keep its election promises.
Another theory of why politicians keep their campaign promises is the mandate theory of election. This theory comes from the positivist school of thoughts, and it is the one most frequently encountered in the literature. According to the theory political parties make specific pledges in their election platforms and they try to fulfill as many pledges as possible once elected in power. Hofferbert and Budge (1992) use the analogy of an architectural blueprint to illustrate this central idea of the theory. The theory is based on two postulates. First, the competing parties offer platforms that are quite distinct from each other; the differences between the platforms of the parties are sufficiently salient to allow the voters to make a rational non-arbitrary choice with a minimum of information. Party differences, and the consequences of electing one party rather than the other, are easily discernable by voters and the information costs of discovering these are small. Second, the voters prospectively compare the utilities provided by competing parties and they give their vote to the party which offers them the greatest utility.2 Based on these postulates, the theory predicts that the winning party carries through the platform on which it has been elected. The logic underlying this expectation is straightforward: The issues advocated by the party in government are the winning issues that contributed in getting the party elected in the first place. It is therefore rational for a utility-maximizing party to carry out its election promises. Another reason for keeping election promises is to avoid retaliation by disappointed voters at the next election.3
Although we have not found it in published form, it is also possible to build an explanation of why politicians keep their campaign promises following the constructivist approach. As with mandate theory, the constructivist approach would argue that there is a strong congruence between campaign pledges and subsequent government actions. However, this would be not because politicians find it rational to keep their promises but because the political discourse of campaign promises and of government actions goes through a same process of social construction of meaning (Faure et al. 1995). Unlike the mandate theory (and unlike positivist theories in general) which assumes that a true external reality can be discovered through the scientific method, constructivism postulates that we can only grasp different subjective constructions of reality produced by different people. Therefore, constructivism does not lead to empirically testable propositions. However, it is still a useful approach in that it challenges the positivist approach by raising important interrogations on the ambiguous nature of campaign pledges, on the relation of causality between campaign pledges and government actions, and on the meaning that the governing elites give to election promises in order to justify post hoc their political decisions. We will take up this thread in conclusion.
We therefore have a several theories that support the idea that politicians keep their promises, and that the pledges that parties offer in their election platforms do make a difference in subsequent policy making. This is against the background of the null hypothesis that election platforms make no difference. In what follows, we review how different scholars have tested this null hypothesis. To conduct this review, we will ask two broad questions, one empirical, and one methodological:
To what extent are campaign pledges subsequently redeemed? In other words, what is the level of congruence between pledges and actions?
How do we know when a campaign pledge is redeemed or not? In other words, to what extent do the studies we review present results that are valid and replicable?
2. Method: How to Test Whether Campaign Pledges Are Redeemed
There exist at least three distinct methods to test empirically the predictive value of campaign promises. These can be classified along a continuum measuring the extent to which researchers must use their own judgement in assessing when a pledge is redeemed. At one end of the continuum we find correlational studies matching the variation in some objectively quantifiable measure of policy output (roll call votes) in one or more policy domains with quantitative measures of pre-election policy preferences by parties or candidates (David 1971, Ringquist and Dasse 2004). These studies require little or no subjective interpretation of the data. They maximize external validity and reliability of the data in both the independent and the dependent variables. But the method sacrifices the substance of the policies under analysis. In other words, one loses in terms of internal validity what one gains in terms of reliability. Moreover, since the variables that this methodology requires (roll call votes and surveys-based evidence of politicians’ pre-election stance on issues) are available only in the US, the methodology is not very useful in the context of comparative research.
Next along the continuum we find studies that try to combine the advantage of a correlational design while not sacrificing the substance of the policies under analysis. The method consists of correlating variations in some measure of government output (public expenditures in specific policy domains) with how much space has been devoted by the winning party to each domain in its election platform. This method has been used by scholars associated with the Comparative Manifesto Project (CMP), a standing research group of the European Consortium for Political Research (ECPR). The initial objective of the CMP project was (and still is) to record and analyze the contents of the election platforms in democratic countries since World War 2. These data are then used to position the parties in their respective national political space and track their evolution from one election to another (Volkens 2002). The election platforms are coded into a pre-established set of policy categories. The score of each category is then calculated to reflect the relative emphasis of each category in each party platform. The results of this coding have been used as a basis for the empirical tests of the predictive power of campaign pledges. Annual budget expenditures are measured in various policy sectors that match the categories used to code the election platforms. The actual test is conducted by correlating the amounts of public spending (in percentage of GDP) with election platform emphases
Empirical test of this type were carried out in a comparative volume (Klingemann et al. 1994) and in single-country studies in Germany (Hofferbert and Klingemann 1990), Canada (Petry 1988, 1995), the US (Budge and Hofferbert 1990), France (Petry 1991), Britain (Hofferbert and Budge 1992) with somewhat mixed results. Budge and Hofferbert (1990), Hofferbert and Budge (1992) and Klingemann et al. (1994) report high correlations between government spending and election promises for the most part. These positive results are suspect, however. King and Laver (1993), in their replication of the Hofferbert and Budge (1990) study of US data have pointed out major flaws in the Budge-Hofferbert methodology, including the failure to report the standard errors for the regression coefficients; the failure to measure the effect of the time trend on changes in public spending and the absence of test of autocorrelations. When these elements are included in the regression equations, there is no longer a significant correlation between public expenditures and the election platforms of the winning party. Thome (1999) goes further in his methodological critique by showing that the results by Budge and Hofferbert (1990, 1992) and by Klingemann et al. (1994) are irremediably tainted by their failure to include several additional parameter restrictions required by the theoretical model.
Another difficulty with the Budge-Hofferbert method is its reliance on the selective emphasis methodology which consists of only recording the space devoted by a party to a particular policy category in its platform, without differentiating negative from positive party attitudes toward issues. Hofferbert and Budge (1992) praise the selective emphasis methodology on the ground that stressing a particular priority implies an intent to take only positive action in the relevant policy domain, for example by increasing public expenditures or passing legislation. However this is not always the case. A party may very well emphasize (prime) a particular issue in its platform in order to discuss (frame) it negatively, and this may have a measurable effect on policy. Therefore, prudence would recommend that we separate positive from negative emphases (frames). The method has also been criticized for its narrow conception of public policy outputs. By focusing solely on the budgetary expenditures, we miss a large array of state activities and outputs —including laws, administrative decisions, speeches— that are directly relevant to the question of whether governments keep their election promises.
The method that we find at the other end of the continuum consists in counting specific pledges in election platforms and then examining the record of government actions in order to determine how many pledges have been redeemed. Scholars who use the pledge method are obviously going to be better able to control the internal validity of their research design. Royed (1996) and Royed and Borelli (1997, 1999) are strong advocates of this method which, unlike the other ones, does not sacrifice the substance of policy. This is part of the reason why the method has been more widely used by a larger circle of scholars than the other two methods. Another reason for the relative popularity of the pledge method among scholars is that it produces data in the form of percentages of pledges redeemed that are simpler to interpret than the regression coefficients produced by the two other methods. But there is always the risk that these data are invalid because they rely on a subjective interpretation of whether a promise is kept. This is why it is so important to establish whether and to what extent the studies that use the pledge method and present their results in the form of percentages of redeemed pledges contain valid and replicable proofs of what they pretend to demonstrate. This is what we set out to do in the next section.
3. Analysis
After an exhaustive bibliographic search, we have been able to identify 18 separate studies comprising 21 cases that present results in the form of percentages of redeemed campaign pledges.4 Table 5.1 reports the data. Each case is identified by the author and the year of publication of the study, the country (or countries), the period of analysis, and the average percentage of pledges redeemed. All the studies reported in the table calculate percentages of promises that are redeemed, but the validity of the method of calculation varies greatly from one study to the next. We have therefore added four criteria intended to give some idea of the validity of the calculation/demonstration in each study. The entries for these criteria are in the form of binary answers to four simple questions:
Does the study contain an operational definition of a campaign pledge? (Yes or No). A definition is considered operational here if it contains explicit criteria of exclusion of what a pledge is not. Another condition is that the documentary sources of campaign pledges are precisely referenced.
Does the study contain an operational definition of government action/output? (Yes or No). A definition is considered operational here if it contains explicit criteria of exclusion of what a relevant government action/output is not. Another condition is that the documentary sources of government actions are precisely referenced.
What is the extent of the documentation of government outputs with which campaign pledges are matched? (Large or Small). For example, are campaign pledges matched only with laws, or are they also matched with throne speeches, with budgets, with annual reports from various ministries?
How precise, replicable and valid is the demonstration that a pledge is fulfilled or not? Or put in more simple terms, how much room is left to a researcher’s own judgment?
Our objective is not to measure the overall quality of these studies. That will require an analysis which goes well beyond the scope of this chapter. This research is a first step to check out the availability of all the elements that are required to conduct a meta-analysis.
Table 5.1 about here
Only five out of the 21 studies reviewed meet the four methodological criteria. Not surprisingly, the more recent the study is, the better overall score it gets. The five studies that received a perfect score have all been published within the past twelve years. It is not until 1996, with the first study by Terry Royed, that explicit definitions of what a pledge is are systematically provided. The systematic occurrence of operational definitions of government action is even more recent. The usefulness of an operational definition of campaign pledges and government actions is obvious in any comparative exercise: The broader the definition of pledges, the smaller the expected proportion of pledges that will be fulfilled. Conversely, the broader the definition of government actions, the larger the proportion of pledges that are expected to be redeemed. Thus, one needs an operational definition if one is to validly compare the proportion of fulfilled pledges across case studies. It was not until the late 1980s that the documentary sources of campaign pledges and government actions were fully referenced. This criterion is important because it makes it possible to judge and compare the quality and the diversity of the documents used to measure pledge fulfillment in different studies. The final criterion of whether there is a valid proof that pledges are redeemed is the least frequently met. Aside of the work by David (1971), only the most recent studies satisfy this criterion enough to deserve a mark in the table.
Let us now turn to a more detail description of each study. We start with the study by Pomper (1968) —the oldest one on the list— and its updated version (Pomper and Lederman 1980). However innovative, Pomper’s work fails to provide fully operational definitions of pledges and government actions (at least based on our definition). The assessment of pledge fulfillment is based almost exclusively on a volume entitled Congress and the Nation: 1945-64 and the subsequent publications of Congressional Quarterly Service. The demonstration that a particular pledge is redeemed relies on a typology which involves five categories: full action (passage of a law), executive action, similar action (indirect action by the executive or legislative branch), negative fulfillment, defeated (the law did not pass), and no action (status quo). This typology finely separates unfulfilled from partially or fully fulfilled pledges. However, the criteria and method that Pomper uses to separate fulfilled from unfulfilled pledges are not fully explained. It is therefore difficult to replicate his work. The same diagnosis applies to the work by David (1971) who expands and rearranges Pomper’s 1968 results. Even though he remains entirely uncritical of Pomper’s methodology, David points out that his and Pomper’s work still contain many ambiguities and that it “must be regarded as the beginning rather than the end of the research that is needed” (1971, 311).
Bradley (1969) identified pledges having to do with social security in Democratic and Republican platforms from 1932 to 1964 and determined whether these were carried out in the form of law. The Bradley article gives a detailed account of which campaign promises were fulfilled and which were not. There is also a discussion of policy changes that had not been previously proposed in Democratic platforms. Although we are able to calculate the proportion of social security platforms planks that were fulfilled, the information discussed in this article is essentially qualitative, with little concern about operationalization and measurement of variables.
Elling’s (1979) research is an uncritical replication of Pomper’s work at the state level. Elling compares the way campaign pledges are redeemed in Illinois (where parties are said to be more pragmatic) and Wisconsin (more ideological parties) respective governments. His methodology is almost identical to Pomper’s. There are no truly operational definitions of campaign pledges and government actions, and no explicit criteria to decide which pledges are fulfilled and which are not. Elling finds that 50% of campaign pledges are redeemed in Illinois, and 45% in Wisconsin. These are the lowest percentages among the 21 studies reviewed here.
The next work is Rose’s study (1984) of how British governments fulfilled their election pledges in the 1970s. This work is probably the least sophisticated of the bunch in terms of our criteria, and therefore the hardest to assess and replicate. There is no attempt at providing explicit definitions of campaign pledges and government actions. There is not clear reference to the archival sources that were analyzed. Neither is there a demonstration of how a pledge is declared fulfilled either completely or partially. In Rose’s defense, his study of whether and how pledges are fulfilled is not the primary purpose of his book. The main goal of Rose was to analyze the influence of Britain’s political parties on politics, and measuring the fulfillment of election pledges was only a part of this objective.
Michael Krukones (1984) measures how US presidents have kept their electoral promises between 1912 and 1976. Krukones’ work is the first to exhaustively document its sources for government action. There are attempts at providing operational definitions of campaign pledges and presidential actions, although they fall short of our criteria. Krukones also innovates by weighting unfulfilled presidential pledges based on whether or not they represent “good faith” efforts to pass a policy that failed due to factors beyond presidential control. However, the methodology for deciding what constitutes a good faith effort and what does not is never explicitly presented. In his study of how US presidents fulfilled their campaign pledges from 1960 to 1984, Fishel (1985) goes one step further in measuring good faith efforts by assessing whether presidential promises need congressional approval to be fulfilled.
Rallings (1987) compares how British and Canadian governments have kept their campaign promises in the 1970s and 1980s. There are no clear operational definitions of pledges and government actions in this work. It is stipulated that to be counted as such, a pledge must anticipate some future action by the governments and not only intentions. However, how actions are distinguished from intentions is not specified. The documentary basis of government actions includes laws, budget speeches, and possibly other sources. But these sources are not precisely identified and referenced. As other studies that preceded it, Rallings’ work makes no reference to explicit criteria on which to decide that a campaign promise is carried out or not. The results for Britain give an average score of 63.7%, against 71.5% for Canada. However, these numbers are the result of a direct match of government actions with the content of throne speeches only. A direct match of government action with election pledges is nowhere to be found in Rallings’ work. This undermines somewhat the purpose of the study.
Next in chronological order comes the study by Monière (1988) about pledge fulfillment in the first mandate of the Conservative government of Brian Mulroney. Unlike Rallings or Rose, who match government output with pledges from party platforms only, Monière matches government output with party platforms and with campaign speeches by party leaders that are covered in national newspaper. Campaign pledges are matched with legislative documents only (laws and House of Commons Hansard). Monière concludes that 74% of campaign pledges are redeemed on average. The number climbs to 80% when non-verifiable pledges are excluded from the analysis. However, once again, there is little explicit discussion of the criteria for deciding which pledges are verifiable and which are not.
Kalogeropoulou’s (1989) study of Greece’s PASOK government is directly inspired by Rose’s work. Unlike Rose, however, Kalogeropoulou makes detailed references to the pre- and post-election documents he uses to assess if pledges are redeemed. These documents come in a large variety: partisan, legislative, executive, and administrative sources are consulted as well as media reports. Unfortunately, Kalogeropoulou does not explain clearly how he uses these documents and on what criteria he relies in order to decide if the government fulfills its pledges.
The studies by Royed (1996) and Royed and Borelli (1997, 1999) are considerably more sophisticated methodologically than the previous work. They are the first ones to provide a clear definition of pledges (although government actions are still left largely undefined). A large variety of documentary sources are consulted to determine whether a pledge is fully or partially redeemed. In the first article, Royed does not specify what documents had been used to operationalize government output, but she does it for the 1997 and 1999 articles with Borelli (CQ reports and statistical abstracts). These studies also go a long way toward an explicit demonstration of how some pledges are declared redeemed while others are not. But the demonstration remains incomplete and, one suspects, there is still a certain amount of subjective interpretation underlying the methodology of deciding whether a pledge is fulfilled or not. The 1996 study by Royed found that 85% of pledges were redeemed in Great Britain under Prime Minister Thatcher (the highest proportion of the 21 studies reviewed here) against only 52% in the US under President Reagan. (the second lowest proportion).
Thomson (2001) analyzed the platform-to-policy linkage in the Netherlands by measuring the fulfillment of election pledges on socio-economic policy. By all standards this article is well crafted. To dig out pledges from party manifestos Thomson used the same method as Royed did and came out with a rate of fulfillment of 61%. But contrary to Royed, whose definition of pledge is operationalized by the presence of an outcome, Thomson was more selective and restricted his definition to policy actions. To assess the fulfillment of election pledges, a CD-ROM database containing references to, and a short description of, all government decisions has been used as the main source. The dependant variable can take three values: not fulfilled, partially fulfilled and fully fulfilled, according to the degree of congruence between government decisions and pledges. A partially fulfilled pledge means that some policy may be taken in the direction indicated by the pledge, but falling short of full realization. One of the most interesting aspects of Thomson’s work is a counter-verification of pledge fulfillment by a panel of experts. A sample of 110 pledges was judged by area specialists and the inter-coder reliability, measured by a Cohen’s Kappa coefficient of 0.70, is quite good.
Petry (2002) finds that 75% of the pledges in the platform of the Parti québécois in the 1994 and 1998 Quebec elections were fulfilled over the period 1994-2000. As with the work by Royed (Royed 1996) and Thomson, Petry’s work provides operational definitions of election pledges and government actions. Government actions are based on a large variety of documentary sources (laws, internal party documents, annual reports by ministries, budget speeches, media reports). Unlike Thomson, there is no measure of inter-coder reliability. However, and this is a novelty, the method of linking pledges to government actions tries to be as neutral and objective as possible. It simply consists of reporting whether a documentary source declares that a pledge has been fulfilled or not. This leaves nothing, in theory, to the researcher’s own judgment. A similar method is used by Petry and Collette (2006) in their study of how the Liberal government of Jean Charest fulfilled its pledges after the 2003 Quebec election. Petry’s (2002) operational definition of pledges and government actions remain unchanged. However, the documentary basis has been extended considerably in the Petry and Collette study.
We close the list with a recent study by Ringquist and Dasse (2004) linking pledges to policy in environmental policy in the US. Instead of testing directly whether the pledges in party platforms are redeemed, they record individual Congress members’ scores in the National Political Awareness Test (NPAT).5 These are then matched with the results of congressional roll call votes on environmental issues. The next step was to integrate environmental policy promises into four different probit regression models, along with other independent variables such as campaign contributors, gender, race, etc. In the four models, the promise variable had a positive and significant effect on the dependent variable. Ringquist and Dasse’s decision to rely on NPAT survey results instead of party platform pledges and on Congressional roll call votes instead of the content of policy speeches or laws revives a tradition that was open thirty five years ago by David (1971).
4. Conclusion and discussion
In response to the first question: “do political parties keep their campaign promises once elected?” our review of 21 cases in 18 separate published studies reveals that parties fulfill 67% of their promises on average. Contrary to popular belief, political parties are reliable promise keepers. Why people underestimate the capacity of political parties to keep their election promises remains an open research question. But it is reasonable to conjecture that this is due in part to a bias in media coverage of how parties keep their promises. Stories of broken party promises on a few important issues have considerably more readership appeal and salience in the public than the coverage of pledges fulfilled on many less important issues.
Our positive finding does not go without important caveats. One is the wide variation in the rate of pledge fulfillment by political parties, from a minimum of 45% (Elling 1979, in the state of Wisconsin) to a maximum of 85% (Royed 1996, in Britain) and a standard deviation of 10.3 percentage points. Clearly, some parties tend to keep their electoral promises more than others. Although this chapter was intended as a literature review, not a meta-analysis, we cannot resist noticing two interesting regularities in the data of table 5.1 that shed light on the question of what factors determine the rate of pledge fulfillment. The first noticeable pattern is institutional. The pattern contrasts US cases with a low average rate of fulfilled promises (65%) and cases from Britain and Canada (Quebec included) with a significantly higher average rate of pledges fulfilled (74%). Although we cannot be sure in the absence of a multivariate statistical test, it is reasonable to conjecture that, other things being equal, parliamentary regimes like the Westminster systems of Britain and Canada, positively influence the likelihood that political parties keep their electoral promises once elected because they give the government the latitude to do so. By contrast, separation of powers in a presidential regime like the US limits the latitude of the executive to keep its promises, and would therefore have a negative effect on the rate of pledge fulfillment.6
The high rates of pledge fulfillment that we find are also limited somewhat by methodological caveats. In response to the question of what methodologies are used to demonstrate that parties keep their election promises, our review has uncovered wide methodological differences across the 18 studies. Recent studies are more sophisticated methodologically than studies conducted in the past, some of which fail to provide the information that would be necessary for even the most basic replication. One interesting pattern emerging from the data of table 5.1 suggests that the rate of pledge fulfillment varies in inverse proportion with the severity of the tests. Although we cannot be sure that the tendency would sustain a multivariate test, the seven cases at the bottom of table, those satisfying the most severe tests, have an average rate of fulfillment of 63%, more than ten percentage points lower than the 74% average for the rest of the data. This might reflect a tendency for political parties to be less reliable in recent years. But there is another likely explanation involving methodological aspects. Party promises in recent elections have tended to be more detailed and precise, and therefore better falsifiable, than in past elections. At the same time, due to an increase over time in the severity of the tests applied by researchers, the rate of pledges that are declared fulfilled has tended to decrease in recent studies. Thus, there probably is a relationship between the number of election promises that are kept and the methodology that researchers use to prove their case.
References
Bradley, J. P. 1969. Party Platforms & Party Performance concerning Social Security. Polity 1(3): 337-358.
Budge, I. and R. I. Hofferbert. 1990. Mandates and Policy Outputs: U.S. Party Platforms and Federal Expenditures. American Political Science Review 84(1): 111-131.
David, P. T. 1971. Party Platforms as National Plans. Public Administration Review 31(3): 303-315.
Davis, M. A. and M. Ferrantino. 1996. Towards a Positive Theory of Political Rhetorics: Why Do Politicians Lie? Public Choice 88(1-2): 1-13.
Downs, A. 1957. An Economic Theory of Democracy. New York: Harper.
Elling, R. C. 1979. State Party Platform and State Legislative Performance: A Comparative Analysis. American Journal of Political Science 23(2): 383-405.
Faure, A., G. Pollet, and P. Warin, ed. 1995. La construction du sens dans les politiques publiques : débats autour de la notion de référentiel. Paris: L 'Harmattan.
Fishel, J.
1985. Presidents and Promises: From Campaign Pledge to Presidential Performance. Washington: Congressional Quarterly Press.
Hofferbert, R. I. and I. Budge. 1992. The Party Mandate and the Westminster Model: Election Programmes and Government Spending in Britain, 1945-85. British Journal of Political Science 22(2): 151-182.
Hofferbert, R. I. and H.-D. Klingemann. 1990. The Policy Impact of Party Programmes and Government Declarations in the Federal Republic of Germany. European Journal of Political Research 18(277-304.
Kalogeropoulou, E. 1989. Election Promises and Government Performance in Greece: PASOK’s Fulfilment of its 1981 Election Pledges. European Journal of Political Research 17(289-311.
King, G. and M. Laver. 1993. Party Platforms, Mandates, and Governement Spending. The American Political Science Review 87(3): 744-750.
Klingemann, H.-D., I. Budge, and R. I. Hofferbert. 1994. Parties, Policies, and Democracy. Boulder: Westview Press.
Krukones, M. G. 1984. Promises and Performance: Presidential Campaigns as Policy Predictors. Lanham, Md: University Press of America.
Monière, D. 1988. Le discours électoral. Les politiciens sont-ils fiables ? Montréal: …show more content…
Québec/Amérique.
Petry, F. 1988. The Policy Impact of Canadian Party Programs: Public Expenditure Growth and Contagion from the Left. Canadian Public Policy-Analyse de politiques 14(4): 376-389.
Petry, F. 1991. Fragile Mandate: Party Programmes and Public Expenditures in the French Fifth Republic. European Journal of Political Research 20(149-171.
Petry, F.
1995. The Party Agenda Model: Election Programmes and Governement Spending in Canada. Canadian Journal of Political Science/Revue canadienne de science politique 28(1): 51-84.
Petry, F. 2002. La réalisation des engagements du Parti québécois : analyse d 'ensemble. In Le Parti québécois. Bilan des engagements électoraux 1994-2000, ed. Petry, F., 161-181. Québec: Presses de l 'Université Laval.
Petry, F. and B. Collette. 2006. Le gouvernement Charest a-t-il respecté ses promesses ? In Le Parti libéral. Enquête sur les réalisations du gouvernement Charest, ed. Petry, F., É. Bélanger and L. M. Imbeau, 83-101. Québec: Presses de l 'Université Laval.
Pomper, G. M. 1968. Elections in America. Control and Influence in Democratic Politics. Toronto: Dodd, Mead & Company.
Pomper, G. M. and S. S. Lederman. 1980. Elections in America. Control and Influence in Democratic Politics. New York: Longman.
Rallings, C. 1987. The Influence of Election Programmes: Britain and Canada 1945-79. In Ideology, Strategy, and Party Change: Spatial Analysis of Post-War Elections Programmes in Nineteen Democracies, ed. Budge, I., D. Robertson and D. Hearl, Cambridge: Cambridge University
Press.
Ringquist, E. J. and C. Dasse. 2004. Lies, Damned Lies, and Campaign Promises? Environmental Legislation in the 105th Congress. Social Science Quarterly 85(2): 400-419.
Rose, R. 1984. Do Parties Make a Difference? Chatam: Chatam House Publishers.
Royed, T. J. 1996. Testing the Mandate Model in Britain and in the United States: Evidence from the Reagan and Thatcher Eras. British Journal of Political Science 26(1): 45-80.
Royed, T. J. and S. A. Borelli. 1997. Political Parties and Public Policy: Social Welfare Policy from Carter to Bush. Polity 29(4): 539-563.
Royed, T. J. and S. A. Borelli. 1999. Parties and Economic Policy in the USA. Pledges and Performance, 1976-1992. Party Politics 5(1): 115-127.
Schattschneider, E. E. 1942. Party Governments. New York: Rinehart & Winston.
Schedler, A. 1998. The Normative Force of Electoral Promises. Journal of Theorical Politics 10(2): 191-214.
Thome, H. 1999. Party Mandate Theory and Time-Series Analysis: A Methodological Comment. Electoral Studies 18(4): 569-585.
Thomson, R. 2001. The Programme to Policy Linkage: The Fulfilment of Election Pledges on Socio-Economic Policy in the Netherlands, 1986-1998. European Journal of Political Research 40(171-197.
Volkens, A. 2002. Manifesto Coding Instructions (Second Revised Edition). Berlin: Wissenschaftszentrum Berlin für Sozialforschung.
Table 5.1: Overview of Publications that Measure how Political Parties Keep their Promises
Authors
Year of Publication
Country
Time Period
Avg. % Pledges Fulfilled
Operational Definition of Pledge
Operational Definition of Government Action
Referenced Sources for Government Action
Explicit Proof that a Pledge is Redeemed
Pomper
1968
US
1944-1966
72
Bradley
1969
US
1932-1964
80
David
1971
US
1944-1966
72
X
Elling 1
1979
Illinois
1947-1971
50
Elling 2
1979
Wisconsin
1947-1971
45
Pomper & Lederman
1980
US
1944-1978
69
Rose
1984
Britain
1970-1979
80
Krukones
1984
US
1912-1976
71
X
Fishel
1985
US
1960-1980
61
X
Rallings 1
1987
Canada
1945-1979
72
X
Rallings 2
1987
US
1945-1979
64
X
Monière
1988
Canada
1984-1988
74
X
X
Karogelopoulou
1989
Greece
1981-1985
70
X
Royed 1
1996
Britain
1979-1988
85
Royed 2
1996
US
1980-1988
52
X
Royed & Borelli
1997
US
1977-1992
60
X
X
X
Royed & Borelli
1999
US
1976-1992
60
X
X
X
Thomson
2001
Netherlands
1986-1994
61
X
X
X
X
Petry
2002
Quebec
1994-2000
75
X
X
X
X
Ringquist & Dasse
2004
US
1997-1999
73
X
X
X
Petry & Collette
2006
Quebec
2003-2006
60
X
X
X
X
Average
67
7/21
7/21
12/21
4/21
Endnotes
The authors wish to thank Gerald Miller for his stimulating comments on an earlier version of this paper.
2 The two postulates of the mandate theory, although inspired by Downs’ (1957) economic theory of elections, are somewhat at odds with Downs’ theory. Unlike the mandate theory which assumes that party platform differ, Downs’ theory assumes that politicians offer platforms that converge toward the median voter equilibrium. In a context of identical (or very similar) party platforms, the information costs are very high, thereby precluding voters from making prospective assessments of what the winning party will do after the election. Rational voters do not differentiate between the policy priorities of the competing parties. They will support a party on the basis of heuristic shortcuts such as their retrospective assessment of the performance of the incumbent in office, or party ideology. Therefore, party manifestos should not be expected to be of much significance in elections.
3 Another positive theory of why politicians keep their campaign promises is Marxism. Marxism considers that the political agenda is controlled by a dominant elite working on behalf of capitalist interests. It is the same ruling elite that is at the origin of election programme discourse and of the decisions of the government. The congruence between the electoral discourse and the government action discourse is no coincidence according to Marxist theory: It should be expected because the two discourses are echoes of one another. According to Marxist theory, party platforms —at least the platforms of the more conventional non-extremist parties— are used by the ruling elite as a ploy to manipulate the popular classes into believing that they have a choice at election time. Ultimately, election promises are used to defuse potential popular uprising against the capitalist state’s chosen policy direction. From a Marxist perspective, the study of the correlation between election promises and government actions is scientifically futile. Marxist theory is therefore of no use to someone interested in studying whether governments do what they say.
4 The number of cases is greater than the number of studies because some studies display more than one case.
5 There are 12 questions about specific policy proposals in the NPAT. Ringquist and Dasse recoded the answers in a new binary variable. A score of 0 means an anti-environmental response and 1 for a pro-environmental response.
6 The conjecture is explicitly laid out by Royed (1996). Her study only applies to President Reagan in the US and Prime Minister Thatcher in Britain. Her conclusion that election promises are kept more often in a parliamentary regime than in a presidential regime cannot, therefore, be considered generally valid until tested across a larger sample of countries.
Index
Britain 8, 15, 16, 28, 32
Canada 8, 16, 20, 28
Campaign pledges …………………7, 12
Comparative Manifesto Project (CMP) 8
France 8
Germany 8
Greece 16, 23, 28
Mandate theory of elelctions ……... 5, 11
Netherlands 17, 25, 29
Quebec 18, 20, 29, 30
Reliability 7, 18
Roll call votes 7, 19
Selective emphasis 9
US 7, 8, 15, 19, 20, 27, 28, 29, 32
Validity 7, 10, 11