Strategic Ambiguity in Formulation of Manifestos: Czech Parliamentary Parties ( )

Transcription

1 Petra Svačinová, Masaryk University Strategic Ambiguity in Formulation of Manifestos: Czech Parliamentary Parties ( ) Introduction Responsiveness of politicians is one of the basic assumptions of party mandate theory, which is in the pledge approach translated into an expectation of inherent willingness of parties to fulfill (as many as possible) definitive pledges given to voters in manifestos while facing obstacles limiting their attempts. Scholars rooted in the pledge approach have found a big amount of factors influencing the ability of parties to fulfill their pledges, and, regardless of the varied percentage of fulfilled pledges, the assumption of the willingness of parties to fulfill the pledges has not been challenged (Naurin 2013: 16). In the pledge approach, just definitive (Royed 1996) and/or objectively testable (Thomson 2001) pledges are usually extracted from manifestos and analysed (Schermann, Ennser-Jedenastik 2012). However, as we can see in everyday politics, candidates for public offices regularly make vague statements that leave voters uncertain about the policies they intend to pursue. (Tomz, van Houwelling 2009). In this paper, the formulation of vague pledges is seen as the rational strategy of at least some kinds of parties. This vision is competing to the traditional vision of party mandate model, that presents parties aiming to formulate specific pledges and striving to fulfill what they pledged. In this paper, the concept of strategic ambiguity is applied to the study of party competition in multiparty systems, and to study of pledges in electoral manifestos. In the theoretical chapter, I introduce the concept of strategic ambiguity and its application to the multiparty competition, connected to the saliency of policy topics for different parties. Following that, I formulate hypotheses concerning party strategies regarding the ambiguity of pledges in party manifestos. In the empirical part of the article, the hypotheses are tested in pledges from manifestos of Czech parliamentary parties ( ). Paper prepared for the IPSA Congress, July 23-28, 2016 Poznań, Poland. Preliminary version, please do not circulare without the permission of the author.

2 Theory Responsiveness as an inherent presumption of the party mandate model Party mandate model presumes that parties behave as policy-seekers and the reason is assumed to be office-seeking. Parties firstly formulate policy programs in order to take votes (and seats), and after election, they strive for fulfillment of their pledged policy programs because they seek for re-election and are afraid of punishment by voters (who are considered to be policy-seekers punishing the deviation of mandate) (Klingemann et al. 1994; Louwerse 2011). Thus, the party mandate model considers both, the policy programme of the party as the commitment given to voters and party responsiveness to voter s preferences. If parties and candidates care about their reputation, party manifesto shall constraint their later policies (Aragones, Neeman 2000). In the empirical research rooted in party mandate model, this assumption is expressed in the vision of parties presenting their truthful positions and clear-cut pledges in pre-electoral campaigns. Presentation of clear-cut pledges leads to justified voters perceptions and electoral choose of parties which best fit their policy preferences. The empirical research in the pledge approach to the party mandate model takes the pledge as its unit of analysis. The studies in the approach are mainly concerned with the question, what features of the pledge or of bargaining situation do determine the odds of pledge fulfillment. In countries with coalition governments, the established set of hypotheses is standardly tested and confirmed, such as the basic hypothesis, that government parties are able to fulfill more pledges than non-government parties (Royed 1996; Thomson 2001; Costello, Thomson 2008; Mansergh, Thomson 2007; Schermann, Ennser-Jedenastik 2012; Thomson et al. 2012). A lot of factors influencing the odds of pledge fulfillment have been found in these studies. 1 All studies rooted in the pledge approach limit the data collection to objectively testable pledges, drawing the definition from Royed (1996) with Thomson s (2001) specification. Royed identifies the pledge as commitment to carry out some action or produce some outcome, where an objective estimation can be made as to whether or not the action was indeed taken or the outcome produced. (Royed 1996: 79). In her operationalization, Royed divided pledges into three categories according to their specificity. Definitive pledge is clear-cut, it must be easy to identify how the promised action shall look like from the formulation of the pledge ( the action/outcome either happens or it does not ). In difficult definitive pledges, the action is promised and the fulfillment can be theoretically objectively determined, but, the testing is 1 Institutional variables cover some of mechanisms of coalition government (positive influence of holding corresponding portfolio, promise of Prime minister party, promise adopted into coalition agreement - Thomson 2001). The non-institutional variables are connected with features of promises and character of party competition (positive influence of promise saliency for the party, congruency of promise for coalition and parliamentary parties, status quo promise Thomson 2001, Costello, Thomson 2008, Mansergh, Thomson 2007, Royed 1996, Thomson et al. 2012, low policy distance Schermann, Ennser-Jedenastik 2012; explicit link to EU-topics in postcommunist country Kostadinova 2013).

3 difficult and depends more on the enthusiasm of the scholar. 2 Judgemental/rhetorical pledges do not allow to determine the fulfillment of promised action because they assume value judgement of the reader. Royed recommended eliminating the second and the third group of pledges from the analysis. Robert Thomson (2001) followed Royed s methodological decision and specified the content needed for identification of the pledge as definitive. According to his definition, pledges are statements in which parties express unequivocal support for proposed government policy actions or non-actions that are testable. The demand of unequivocal support refers to hardness of the statement and is just rarely followed. On the other hand, the requirement of testability meaning that the criteria used to judge the fulfillment of pledges are in principle provided by the writers of election programmes, not by the researcher (Thomson 2001: 180) has been followed broadly in the succeeding studies. The restriction of the analysis on definitive and testable pledges was predominantly methodological, motivated by the strive for the reliability of coding (Royed 1996). 3 However, there are also theoretical arguments. From the prospective point of view, Rose (1984) suggests that voters punish unspecific pledges and force parties to formulate their manifestos specifically. The expectation of the stress to voters preference to clear-cut pledges present also Håkansson and Naurin (2014), who as the only authors in Western European pledge research deal with unspecific pledges and add the historical perspective to the thinking about their formulation. According to them, increasing volatility, the professionalization of campaigns and adaptation of parties strategies to the process of mediatization bear a need to create an image of straightforward politicians with clear-cut pledges, which shall be more likely to attract media. Formulation of vague pledges shall be disadvantageous in the electoral arena and seems to be a strategic fail of parties. The expectation of Håkansson and Naurin about increasing the specificity of pledges in time was confirmed at Swedish parliamentary parties during a long period ( ). Schermann and Ennser-Jedenastik (2012) brought the retrospective part of party mandate model and claim that only the adoption of a hard and objectively testable pledge in the coalition agreement can be considered a real policy gain in coalition-bargaining. (Schermann, Ennser-Jedenastik 2012: 8). Taking these arguments into account, unspecific pledges 4 are seen as non-important for parties and voters, or even disadvantageous and dangerous for parties and as such. Their elimination from the analysis in pledge approach thus seems to be reasonable. However, parties still keep formulating some parts of their manifestos unspecifically. The question for their reasons has not been answered yet and is still relevant. Hungarian researchers Szűcs and Pál (2012) theoretically focused on the function of non-testable rhetorical pledges, they state that even if some pledges lack the measurable policy content, they shall not be entirely ignored (Szűcs, Pál 2012: 5). The authors found three kinds of rhetorical pledges. So-called placebo pledges, which are very rare, and actually, are not pledges in their true meaning. They 2 Testing involves not just determining if legislation was or was not passed, but what the detailed provisions of legislation were, and whether they are likely to result in the outcome promised. (Royed 1996: 80) 3 If the main focus of the study is to identify fulfillment of promises, researchers need to deal just with pledges specific enough for reliable verification (Royed 1996). 4 Which, in the pledge approach means difficult definitive and rhetoric (Royed), soft or untestable (Thomson).

4 have the grammatical structure of pledges and can be measurable as the definition by Royed or Thomson requests, but, voters, as well as parties, do not expect their real fulfilment. 5 The placebo pledges, which are standard products of joke parties, are clearly unrealizable and funny. The second group of pledges contains tricky pledges, which are formulated in a way that makes the verification of fulfillment impossible. Tricky pledges confuse the content of mandate and weaken the binding power of voting. The third group consists of the politically informative pledges. These pledges, even when formulated vaguely, have some function for voters making a decision. They add the emphasis on some policy dimension and as such, they give the information about future policy intentions of parties. In this way, they can help voters to choose between candidates/parties. Differentiation between tricky and politically informative pledges is difficult and the authors conclude that it is rather the differentiation between two functions of the same pledge. Not just rhetorical (according to Royed), but all non-testable pledges can be considered as having these confusing functions. Their formulation does not need to be a strategic fail, but an expression of different and probably fruitful strategy in the formulation of party manifestos. Strategic ambiguity as a successful electoral strategy? The US tradition of rational choice theory responded to the empirical prevalence of vague statements in political campaigns by the constitution of the vision controversial to party mandate model. The obfuscation literature shows that for different reasons, at least some politicians may not be ready to commit themselves to particular policy plans. These politicians may need to weaken the commitment presented to voters to obfuscate the perceptions of voters about their planned policies. The strategy of ambiguity is expected mostly in topics, which are dangerous for parties. That means topics, in which parties and their electorates are the most distant. Clarity, on the other hand, is expected in topics, which are popular and parties and voters are congruent on them (Campbell 1983). The concept of strategic ambiguity assumes that formulation of policy statements is not random in politicians speeches and party documents, but it is rather a conscious decision of parties and candidates in order to hide their future policy plans. In practice, strategic ambiguity means formulating equivocal, vague (Shepsle 1972; Milita et al. 2014; Lo et al. 2014: 2) or emotional statements about future plans in office (Lingdren 2014), eventually keeping silence about planned policies in some areas (Milita et al. 2014; Rovny 2012; Lo et al. 2014). When talking about positioning of parties, the strategic ambiguity equals taking vaguely broad positions on issues or presenting a mixture of (possibly conflicting) positions on one issue (Rovny 2012: 271; Lo et al. 2014: 2) or across issues ( broad-appeal strategy according to Somer-Topcu 2014: 2). The research on voters responses to ambiguity in politicians statements started with the vision of risk-averse voters, who prefer strong commitments to ambiguous ones (Shepsle 1972, 5 As the example of placebo pledge, the authors show pledge of one Hungarian joke party We promise two eternities for everyone. (Szücz, Pál 2012: 8).

5 Aragones, Neeman 2000; the same argument used Rose 1984). In that environment, ambiguity can be seen just as a marginal and painful strategy suitable just for parties who mostly prefer the freedom in office (Aragones, Neeman 2000). The first empirical research on electoral effects of ambiguity in presidential elections (Campbell 1983) showed no impact of ambiguity on voters decisions. However, the more recent empirical studies showed that ambiguous appealing can be electorally successful. Berinsky and Lewis (2007) and Morgenstern and Zechmeister (2001) showed that voters are rather risk-neutral or even risk-acceptant. Moreover, psychological research showed that people tend to be optimistic in perceptions of congruence between their position and position of their preferred party. Tomz and Houwelling experimentally showed, that in a partisan environment, voters react positively to uncertainty about their preferred parties. This means that ambiguity can be used to maintain support of party s own supporters without suffering from loss of opposition supporters (Tomz, Houwelling 2009). The strategy of ambiguity can be seen even as attracting new voters. If the potential electorate is distant from the candidate, voters who have a clear information about the distance, are not willing to vote for such a party (Milita et al. 2014). The broad-appeal strategy can enable parties to attract new voters by expanding the party s appeal while not abandoning previous positions and discourage the previous supporters (Somer-Topcu 2014). Ambiguous position can be beneficial even in the retrospective perspective. Specificity of statements plays its role in defining the commitment: more specific the platform, stronger the commitment to specific policies in future. With the ambiguous platform, parties and candidates are less committed to their policy statements and thus, they are able to avoid pledges, which can be attributed to them later. Strategic ambiguity, thus, should help to free politicians from worrying about citizens holding them accountable for broken pledges (Alesina, Cukierman 1987, Aragones, Neeman 2000). Thomson (2011) showed that voters recognize the non/fulfilment of pledges by parties. Alas, to the best of my knowledge, there is no study on the influence of ambiguous statements of parties on the evaluation of the government performance among voters. Who is predisposed to blur future plans? As mentioned above, the research of strategic ambiguity considers the use of ambiguous statements as a possibly fruitful electoral strategy, whether voters are prospective or retrospective. But, not all parties are seen as equally predisposed to employ the strategy of ambiguity. Some parties, and under some conditions, are seen to have more incentives to adopt the strategy of ambiguity. Milita et al. (2015) connect the strategy of ambiguity with three characteristics of promising parties. Parties have incentives to take ambiguous strategy, if they have something to hide, somewhere to run or something to lose. In this chapter, I present the main features of parties increasing the incentives for strategic ambiguity and formulate the hypotheses about parties strategies in these situations, regarding their formulation of pledges in party manifestos.

6 a. Something to hide In the unidimensional competition, candidates and parties know the median position of the electorate. If the party s and electorate s visions are congruent, there is no incentive for the party to blur its position. When plans of the party are distant from the median voter, such party has something to hide and the fruitful strategy is to choose ambiguous statements in a given policy space (Alesina, Cukierman 1987). The assumption about parties having something to hide is also present in the theory of party salience and issue ownership. The theories consider that in multidimensional policy space parties rarely confront each other in specific policy issue. More likely, they emphasize policy areas, which they perceive as primary and beneficial for them 6 and try to highlight the general saliency of their strong topics among voters. They also try to de-emphasize those policy areas, which they perceive as secondary or electorally dangerous (Rovny 2012). The strategy of emphasizing topics is twofold. Firstly, parties pay more attention to their salient topics and speak more about them (in speeches and party documents). And, their declarations are expected to be more clear-cut in their primary policy areas. When party s policy plans are different from electorate s, or the electorate is dispersed in the particular topic, or if the party is perceived as incompetent or has not defined its position on the respective issue (Aragones, Neeman 2000) 7, the party is expected to blur the position on that topic. In the US research on the strategy of ambiguity and saliency, the early work of Campbell (1983) did not reveal the link between saliency of topic and ambiguity of speeches of candidates. But the newer research on the US (Milita et al. 2014) showed that candidate s strategies are highly responsive to issue-salience. When the topic is salient in the constituency, candidates speak very clearly about it. When we look at manifestos as materials created by parties to express the saliency of some topic, we can conclude the hypothesis, that saliency of policy domain (determined by the party in the manifesto) decreases the use of ambiguity in the respective domain. In pledge approach, the hypothesis is connected to pledges in party manifestos: H1: Bigger the individual saliency of policy area for the party, bigger the odds of a pledge to be testable in this area. b. Somewhere to run If the individual party was the only actor, just her salient issues would be discussed, and her dangerous issues would be neglected in the campaign. However, parties have rarely monopolist agenda control about policy debates and the party that hopes to de-emphasize its individually 6 Parties are considered to be the most competent on them (Petrocik 1996) 7 When party elites are in conflict about position of party on the issue; it is also expected for parties which do not know the position of electorate on respective issue and are afraid of taking dangerous position (Aragones, Neeman 2000; Glazer 1990), because of their inexperience in political competition or because of their single-issue targetting.

7 dangerous issues can be unable to do so if other parties have incentives to emphasize them. At the same time, it can be difficult for the party to emphasize an issue to the general level if other parties refuse to do so (Scott, Steenbergen 2004). So, the topic dangerous for an individual party can be discussed broadly because of its general saliency. According to Milita et al. (2015), it is not possible to blur the position in generally salient issues, because these issues are discussed by media and parties have nowhere to run. In generally salient topics, parties are forced to take clear positions. But, if there is the space to run somewhere which means, to blur the position in the generally secondary policy area, parties try to do it. Milita et al. (2015) confirmed the analysis on speeches of US House candidates about the Iraq War. When the topic was salient in the constituency, candidates were more likely to speak clearly and less likely to speak ambiguously. Contrary, Rovný (2012) suggests that parties firstly try to keep silence about their dangerous topics. However, when it is impossible to keep silence because of the general saliency of topic, the expected strategy is to speak ambiguously. These two arguments state as base for two contrasting hypotheses. H2a: Bigger the general saliency of the issue, bigger the odds of promise to be clearcut in that issue for all parties. H2b: In the generally salient issues, bigger the odds of promise to be clear-cut with increase of individual saliency. c. Something to lose The third characteristic is connected to electoral expectations of parties in electoral competition, and can be considered as the reaction of differently defined actors to the structure of competition. According to Alvarez (1998), ambiguity is the reaction of candidates to the structure of competition. Thus, use of ambiguity varies across candidates as a function of incumbency, previous experience, and national prominence. Ambiguity shall be seen as characteristics rather than strategy, not a subject of candidate or party agency but rather an agency of exogenous context that can change only slowly and independently from the candidate or party (Rovny 2012). Current incumbents, governing parties, are seen to be high-quality candidates. The non-incumbents are seen as low-quality candidates. Low-quality candidates can be divided into two groups: high-quality non-incumbents (opposition candidates earlier successful in elections) and low-quality non-incumbents (previously unsuccessful or running for their first time). While the incumbents are very well-known and aiming at re-election, the low-quality non-incumbents are not known and their electoral expectations are the worst. The high-quality non-incumbents are an intermediary category. High-quality candidates have bigger incentives to cover the real plans because revealing dangerous plans can discourage voters from casting them vote. Their situation is characterized as they have something to lose. Otherwise, low quality parties and candidates without previous experience with office, and with little ambitions to take seats are more open to risk because they have nothing to lose. Revealing their plans can be seen as the only possibility how to take some votes (Milita et al. 2014: 425).

8 1) Government versus opposition According to Milita et al. (2015), current government parties possess, given to their function as a government, the valence characteristics. They are generally perceived as more experienced. Because of this advantage, incumbent parties are perceived predisposed to blur their position and let voters decide just on the basis of valence characteristics. That means government parties shall be more ambiguous in their statements than opposition parties. On the other hand, according to numerous authors (Berger, Munger and Potthof 2000; Downs 1957; Hayes et al. 2008), incumbent plans are already known, which limits the ability of government parties to make real ambiguous positions. Government parties have clearer reputation regarding their policy plans. Blurring becomes futile strategy, particularly the possibility to blur salient (the economic) issues seems to be limited (Rovny 2012). However, as Rovny (2012) states, the probable acknowledgment of government party s plans does not mean that party won t try to mist its position. Thinking in that way, government party shall try to use the strategy of ambiguity and attract voters attention to its experience and leader abilities. This argument is in accordance with Alesina and Cukiermann (1987) that incumbent politicians may use office and decisive institutional procedures to fuzz their true positions. H3: Government parties have bigger odds of a pledge to be ambiguous in comparison to opposition parties (high- and low- quality non-incumbents). 2) High versus low-quality non-incumbents All challengers are typically at the disadvantage when compared to incumbents. But not all nonincumbents are equal. Candidates who had previously held elected office should be less disadvantaged than those who have not held elected office. Having previously held office gives the non-incumbents some minimal level of experience, policy expertise and connections. New parties entering the arena are probably inexperienced and do not know the preferences of the electorate (Hayes et al. 2008). This general inexperience shall lead to the use of strategic ambiguity. However, as Milita et al. (2015) state, low-quality non-incumbents are typically electorally disadvantaged, because they are unknown to the public. They cannot run on valence and are predisposed to gamble and make clear-cut appeals. Their hope for victory is that their issue position will resonate in public opinion. Because the low-quality non-incumbents have nothing to lose, they shall take clear positions regardless on the congruence between them and the constituency. The high-quality non-incumbents are an intermediary category between highquality incumbents and low-quality non-incumbents. H4: Low-quality non-incumbents have bigger odds to be clear-cut than parliamentary parties (high-quality incumbents and high-quality non-incumbents).

9 3) New party types Taking the changing party organization and goals into account, we can also conclude that not all low-quality non-incumbent parties are the same, especially in post-communist countries. Changing goals of parties and accommodation of their functioning lead to the rise of new party types, who have goals different from expectations of party mandate model. Especially the political entrepreneur parties or business-firm parties (Hopkin, Paolucci 1999) are the case. The first goal of such parties is to satisfy the requirements of leaders of these parties for positions and power. Their manifestos shall show the new party types as a standard party concerned on the fulfillment of pledges, even if the main goal of them is kept secret. Because parties founded by businessmen don t standardly have problems with a political campaign and general popularity, they are not forced to take clear positions. That means, low-quality non-incumbents belonging to new party types are expected to use the tools of the strategy of ambiguity more than the rest of low-quality non-incumbents. H5: Low-quality non-incumbents belonging to new party types have bigger odds to be clearcut than rest of non-quality non-incumbents (containing traditional parties). Czech political parties included into analysis ( ) Sixteen manifestos of Czech political parties (parliamentary and non-parliamentary) are included into the analysis 8. Chosen parties took place in three parliamentary elections and contain examples of all four party types described above (high-quality incumbents government parties, high-quality non-incumbents opposition parties, low-quality nonincumbents of new party type and non-quality non-incumbents of tradition party type) (categories are visualized in table 1). As high-quality incumbents were treated parties, which acted as government parties for the longest time of the latest election period 9 (that means, for election 2006, government parties of the period were recognized as high-quality incumbents, opposition parties from this period were recognized as high-quality non-incumbents. As low-quality non-incumbents were treated parties, which had not been present in the legislative body before, but were successful in given election (for election 2006 parties, which had not been in Parliament in period , but took parliamentary seats in 2006). 10 This group was divided into two sub-groups 8 Because of the lack of time for analysis, I had to neglect three parties: KDU-ČSL, ČSSD and SZ in 2010, ČSSD in Czech political system sometimes products the caretaker governments after breakdown of party government (once in given period government of Jan Fischer in 2009) parties, which took part in this government were not treated as government parties. 10 I did just one exclusion from this proces KDU-ČSL, has been parliamentary party continuously since There was just one parliamentary period without KDU-ČSL ( ). Because of its long presence in

10 according to the type of parties. Parties that are considered as entrepreneur parties in the Czech context (Hloušek 2012; Kopeček, Svačinová 2015) were treated as low-quality non-incumbents of a new type. The rest of parties was treated as low-quality non-incumbents of the traditional type. Table 1: Parties 11 divided according to their type Elections High-quality incumbents High quality nonincumbents Low-quality nonincumbents (trad.) 2006 KDU-ČSL ODS, KSČM SZ ODS, KDU-ČSL KSČM, ČSSD TOP 09 VV Low-quality nonincumbents (new) 2013 ODS, TOP 09 KSČM, ČSSD, KDU-ČSL - ANO, Dawn Data and Methods The source of data are manifestos of parties in three election periods. The manifestos were earned at web pages of these parties. In the case of more types of the manifesto in one election, the longest was used for the analysis. Unit of analysis The pledge approach defines as its unit of analysis the pledge in the manifesto. Definition of the pledge has been developed since the 1960s (Pomper 1968, Railings 1987) from the broad and possibly unclear definition to more narrow for its purpose that is the identification of pledge fulfilment among political parties (Pétry and Collete 2009 for a deeper review of early definitions of pledge and problems of their application). The first broadly accepted definition was formulated by Terry J. Royed (1996), part of scholars, nevertheless, use the definition by Robert Thomson (2001). In Royed s view, the pledge is the commitment to carry out some action or produce some outcome, where an objective estimation can be made as to whether or not the action was indeed taken or the outcome produced (Royed 1996: 79). Thomson specified the definition by the requirement of unequivocal support, his definition of pledge requires unequivocal support for proposed government policy actions or non-actions that are testable (Thomson 2001). The broad acceptation of one of those two definitions does not mean that the methodology for extracting pledges from manifestos is only twofold in the majority of texts (Naurin 2013 for discussion), but there are similar requirements for the unit of analysis. parliamentary body I think that the party was in 2013 sufficiently well-known and was not in the situation of lowquality non-incumbents. So, I treated it as high-quality non-incumbent although the party has not been in the parliament in preceeding period. 11 KDU-ČSL (Christian Democratic Union Czechoslovak Peoples party), ODS (Civic Democratic Party), KSČM (Communist party of Bohemia and Moravia), SZ (Green party), ČSSD (Czech Social Democratic Party, TOP 09 (Tradition, Responsibility, Prosperity 2009), VV (Public Affairs), ANO (Movement Yes 2011), Dawn (Dawn of Direct Democracy).

11 The definition has been further improved, a good example of such a beneficial improvement of the definition made researchers of the AUTNES study 2008 (Dolezal et al. 2014). As can be seen from both, Royed s and Thomson s definition, the extraction of pledges from manifestos is in the majority of texts limited by the requirement of verifiability, to real, testable or hard pledges. For achieving my goal (to detect possible use of the strategy of ambiguity in formulation pledges in manifestos), I needed to find testable and non-testable pledges as well. To do so, I used just the shortened version of the definition (definition of potential pledge by Terry Royed) and defined the pledge as broadly as possible. I used the procedures for differentiation between testable and untestable pledges in the second phase, in the process of formulating the coding scheme. I was inspired by the approach of the AUTNES study (2008) and discussed the problems with Laurenz Ennser-Jedenastik, one of the researchers of the project. In the AUTNES project, all grammatical sentences are divided into minimal core sentences consisting subject ( we, the party ), the noun phrase (the subject of the pledges, standardly the noun and some extensions adjectives, descriptions of the subject) and the verbal phrase (a verb linked to the object and the subject). Simultaneously, if there are more areas of action/result or more addresses of the action/result, these situations were seen as criteria for identification of more pledges in one grammatical sentence. Dependent variable The dependent variable is the specificity of the pledge ( non-definitive or non-testable pledge was coded as 0, testable pledge was coded as 0. For determination of testable and non-testable pledge, I used detailed coding instructions proceeding from Thomson (2001) and his followers. The testability shall be identified according to the policy content of every pledge (action or result, which is at the core of the pledge). The testable pledge is specified in sufficient detail for the researcher to find out if the action or result were realized. Fulfilment/non-fulfilment of the pledge shall be recognized at the end of the parliamentary period. The criteria used to judge the fulfilment of pledges is in principle provided by the writers of election programmes, not by the political scientist. The criteria shall ideally be objectively measurable, the example of the testable pledge is the creation of the new office, the new law, increasing taxes etc. The pledge shall be identified as non-testable, if the researcher must add his own value judgement for assessing of the fulfilment of the pledge (for example the pledge of improvement, making something more fairly, financially viable, easement etc.). If the researcher needs to guess how authors of the manifesto imagined the fulfilment, if from the formulation of pledge can be concluded more than one form of fulfilment, the pledge shall be coded as non-testable.

12 Independent variables Individual saliency of policy domain every pledge was categorized into one of 13 policy domains 12 (the coding scheme of AUTNES study accommodated to the Czech case by adding some specific Czech topics was used). Individual saliency for party means percentage of pledges from entire manifesto which was dedicated to given policy domain (when, for example, a party dedicated 20% of pledges to the domain of Economics, every pledge of this party devoted to Economics was coded with 20). As can be seen from the figure 1 (showing the percentage of pledges in policy domain of every party), not all domains were equally salient for all parties, but some domains seem to be more important than the others (for example the topic Immigration was not very important at all). All policy domains created the content of entire manifesto maximally of 25 %. Figure 1: Percentage of policy domain in manifestos of parties General saliency of policy domain this concept anticipates that there is a broad discussion of some topics among more parties in the electoral campaign. It is very difficult to find an appropriate measurment for topic, which parties could have perceived as generally salient while formulating their manifestos. It is also difficult to answer if parties these perceptions had. The operationalization shall include the general saliency of topic in time-period before election. That is the result why I decided to estimate the general saliency not from the post-election voter surveys, but also from manifestos of parties (even when I realize that this decision can be criticised). I took three most salient topics from each party manifesto and counted them up (table 2) for every election. From this procedure, I chose 3 (4 in case of 2013) topics, which repeated the most frequently. These topics include the topics which were discused broadly by more parties and it can be deduced that more parties attracted the general audience to the topic. Thus, pledges dedicated to Economics, Welfare state and Infrastructure in 2006 and 2010; and pledges dedicated to Economics, Welfare state, Education and Culture and Reform of 12 Economics, Welfare state, Budget, Education and Culture, Security, Army, Foreign Affairs, Europe, Infrastructure, Society, Environmental protection, Reform of Institutions, Immigration.

13 institutions in 2013 were coded as generally salient by value 1 (the rest of pledges was coded 0). Table 2: Distribution of three the most individually salient pledges among parties in given electoral years Economics Welfare state Budget Education and Culture Security Army Foreign affairs Europe Infrastructure Society Environment protection Reform of Institution Immigration Type of the party variable containing four categories of parties: high-quality incumbents government parties (1), high-quality non-incumbents opposition parliamentary parties (2) and low-quality non-incumbents of traditional (3) and new (4) type. Year of elections 2006 (1), 2010 (2), 2013 (3). Analysis Descriptives As can be seen from the table 3, the number of pledges decreased with time. It can be an interesting result of changing campaigning in time. While in 2006 the average number of pledges per manifesto was 841, the number decreased to 549 in 2010 and to 353 in Also interesting is very small variance in the percentage of specific pledges in manifestos. On average, 25 % of pledges are specific per manifesto. We can say that the average percentage of specific pledges per manifesto slightly increases (24 % for 2006, 25 % for 2010 and 28 % for 2013), but the differences in different elections are very small. The most specific manifesto was of ODS in 2013 (44%), the most ambiguous were of KSČM in 2006 (19%).

14 Table 3: Specificity of manifesto and number of pledges in manifestos in different years party % of specific pledges N of pledges KDU-ČSL 28% 818 ODS 26% 945 KSČM 19% 750 SZ 22% mean 24% 841 KSČM 23% 457 ČSSD 21% 513 TOP09 22% 510 VV 31% 678 ODS 25% mean 25% 549 ODS 44% 294 TOP09 25% 381 KSČM 21% 387 ČSSD 24% 452 KDU-ČSL 27% 392 ANO 31% 396 Dawn 42% mean 28% 353 Total mean 25% 536 When looking at a specificity of manifestos according to the type of party (figure 3), there are probably more promising differences. Manifestos of government parties seem to be more specific than manifestos of parliamentary opposition (but, the difference can be caused by the most specific manifesto of ODS in 2013). There is also a visible difference between traditional and new-type low-quality on incumbent's manifestos. This can be seen as a surprise, because manifestos of low-quality non-incumbents were expected to be the most specific at all. The manifestos of entrepreneur parties are the most clear-cut in the sample (33 % of clear-cut pledges on average). That is some surprise, too, because I expected that new party type- nonincumbents try to mist their mandate in the pre-electoral period in comparison to their traditional colleagues.

15 ODS 2013 TOP ODS 2010 KDU-ČSL 2006 mean KSČM 2013 ČSSD 2013 KDU-ČSL 2013 KSČM 2010 ČSSD 2010 ODS 2006 KSČM 2006 mean TOP SZ 2006 mean ANO 2013 Dawn 2013 VV 2010 mean mean Figure 3: Percentage of specific pledges in manifestos according to party types 50% 45% 40% 35% 33% 30% 25% 27% 23% 22% 25% 20% 15% 10% 5% 0% High-quality incumbents High-quality non-incumbents Low-quality nonincumbents (traditional parties) Low-quality nonincumbents (new party type) Total Relationship between testability and saliency of domain Because I expect that presence of testable pledges is influenced by saliency of given policy domain, I illustrated the relationship between saliency of the domain (percentage of pledges formulated in given policy domain) and percentage of testable pledges in this domain. The scatterplot shows that with an increase in salience of the domain for a party the percentage of testable pledges slightly increases (Pearson correlation coefficient for this relationship is 0,219).

16 Party level Pledge level Figure 4: Scatterplot for the relationship between saliency of policy domain in manifesto and percentage of testable pledges in given policy domain. All variables were included into models of multilevel binary logistic regression (command xtmelogit in stata). Random effects are estimated on the level of party. Odds ratios and standard errors are reported. Three models were established, in the first model, the dummy for highquality incumbents was used (to test the third hypothesis), in the second model, the dummy for low-quality non-incumbents was used (to test the fourth hypothesis). In the third model, the categorical variable containing all four distinguish groups of parties was used to test the fifth hypothesis. Model 1: Multilevel binary logistic regression for testability of the pledge Odds Ratio Odds Ratio Odds Ratio (Std. Err.) (Std. Err.) (Std. Err.) Individual saliency of policy domain for party 1.030*** (0.008) 1.023*** (0.008) Generally salient policy domain 1.784** 1.789** (0.315) (0.316) Individual saliency * general saliency 0.963** 0.963** (0.012) (0.012) High-quality incumbent: government party (dummy) (0.133) Low-quality non-incumbent (dummy) (0.147) High-quality incumbent: government party (ref.high-quality non-incumbent) 1.030*** (0.008) 1.77** (0.311) 0.963** (0.012) 1.213** (0.113)

17 Low-quality non-incumbent (trad.) (0.128) Low-quality non-incumbent (new) 1.680*** (0.192) Year: 2010 (ref 2006) (0.159) (0.140) (0.090) Year: (0.186) (0.168) (0.110) Const *** (0.030) 0.206*** (0.026) 0.207*** (0.023) N obs N groups Log likelihood From the odds ratios presented in all models we can see that individual saliency slightly, but significantly increases the odds of pledge to be testable (increase in 1 % of individual saliency of policy domain increases the odds of pledge to be testable in 3 percent). This enables to support the first hypothesis (Bigger the individual saliency of policy area for the party, bigger the odds of a pledge to be testable in this area.). Looking at the effect of general saliency, we can see that the pledge treated as generally salient increases the odds of specificity in 77 %. This clearly supports the hypothesis 2a (Bigger the general saliency of the issue, bigger the odds of promise to be clearcut in that issue for all parties). The interaction between general and individual saliency was used for testing of the hypothesis 2b. The model shows than increase in individual saliency decreases the odds of testability of pledge in generally salient topics. This rejects the hypothesis 2b (In the generally salient issues, bigger the odds of promise to be clearcut with increase of individual saliency). The rest of hypootheses was concerned on features of party types. The first model was estimated to test the third hypothesis (Government parties have bigger odds of a pledge to be ambiguous in comparison to opposition parties (high- and low- quality non-incumbents). Although there is slightly incresasing effect of the dummy variable, it is not statistically significant, so, the third hypothesis is not supported by the data. To test the fourth hypothesis, I estimated the model with dummy for both sub-groups of low-quality non-incumbents. As it is seen in the second model, this group is also not significantly increasing the odds of testability of the pledge. So, the fourth hypothesis (Low-quality non-incumbents have bigger odds to be clear-cut than parliamentary parties (high-quality incumbents and high-quality non-incumbents) cannot be maintained. For testing of the last hypothesis, the third model was estimated. The odds of pledge of the low-quality non-incumbents belonging to new party types are (in comparison to reference category) bigger than of the traditional low-quality non-incumbents. This leads to necessity to reject the fifth hypothesis (: Low-quality non-incumbents belonging to new party types have bigger odds to be clear-cut than rest of low-quality non-incumbents (containing traditional parties). Simmultaneously, it is seen that odds of pledges of government parties to be testable is higher in comparison to the reference category (opposition parties). This means that government parties, while having bigger odds oof testable pledges in comparison the the entire group of opposition parties, are nevertheless not the most ambiguous among four party types.

18 Conclusion The paper is rooted in the pledge approach to the party mandate model. I tried to look at the pledge approach from the other side. I looked not at the fulfilment of testable pledges into coalition agreements or into reality, as it is usual in this field of study. I looked at the testability of the pledge as the dependent variable. In the theoretical part, the arguments why to deal with the non-testable pledges were introduced. The arguments of the theory of strategic ambiguity and issue ownership were used to deduce the hypotheses about the effect of individual and general saliency of policy domain on the testability of pledge in these domains. These expectations were confirmed individual as well as general saliency of policy domain do influence the odds of testable pledges. As I thought, individual features of parties their in/experience with a presence in government or legislative body can influence the pre-electoral campaign. I expected that government parties can be more than the other parties motivated to obfuscate the mandate, on the other hand, lowquality non-incumbents will more probably produce clear-cut manifestos. I also focused on subgroup among low-quality non-incumbents and expected that entrepreneur parties try to mist the mandate more than other low-quality non-incumbents. The models allowed maintaining just some of hypotheses. Government parties were not more clear-cut than the broad group of the rest of parties, however, in more detailed comparison, pledges of government parties had bigger odds to be testable than parliamentary opposition. Pledges of low-quality non-incumbents as a complex group, on the other hand, had bigger odds to be testable (in comparison to the rest of parties). However, when this group was divided into two sub-groups (traditional and new party types), the expectation regarding the difference between them was totally opposite. Pledges of entrepreneur parties had bigger odds to be testable.this result is interesting and leads to further study about the substance of clear-cut pledges in manifestos of new parties (are these pledges populist, unrealizable?) as well as to study of electoral strategies of non-parliamentary parties at all. The paper probably contributed to the study of strategic ambiguity in multiparty systems. I studied the hypotheses about using ambiguity traditionally operationalized through the opinions of citizens or experts on manifestos and their testable and non-testable pledges. Further attention shall be paid to the reliability of the coding process, adding missing manifestos into the analysis and probably extending the dataset on other election years and non-parliamentary parties, which were not successful in elections.

19 Literature Alesina, A. F. Cukierman, A The politics of ambiguity. Alvarez, R. M Information and elections. University of Michigan Press. Aragones, E. Neeman, Z Strategic ambiguity in electoral competition. Journal of Theoretical Politics 12, no. 2, Berger, M. M. - Munger, M. C. Potthoff, R. F The Downsian model predicts divergence. Journal of Theoretical Politics 12, no. 2, Berinsky, A. J. Lewis, J. B An estimate of risk aversion in the US electorate. Quarterly Journal of Political Science 2, no. 2, Campbell, J. E The electoral consequences of issue ambiguity: An examination of the presidential candidates issue positions from 1968 to Political Behavior 5, no. 3, Costello, R. - Thomson, R Election pledges and their enactment in coalition governments: A comparative analysis of Ireland. Journal of Elections, Public Opinion and Parties, Vol. 18, no. 3, s Dolezal, M. Ennser Jedenastik, L. Müller, W. C. Winkler, A. K How parties compete for votes: A test of saliency theory. European Journal of Political Research 53, no. 1, Downs, A. (1957): An Economic Theory of Democracy, New York: Harper and Row. Håkansson, N. Naurin, E Promising ever more An empirical account of Swedish parties pledge making during 20 years. Party Politics. Hayes, Tom, et al Strategies of ambiguity: Modeling rhetoric in primary election campaigns. Annual Meeting of the Southwestern Political Science Association. Las Vegas, March Hloušek, V Věci veřejné: politické podnikání strany typu firmy. Politologický časopis 19, no. 4, Hopkin, J. Paolucci, C The business firm model of party organisation: Cases from Spain and Italy. European Journal of Political Research 35, no. 3,

20 Klingemann, H. R. Hofferbert, R. Budge, I. (1994). Parties, policies, and democracy. Boulder: Westview Press. Kopeček, L. Svačinová, P Kdo rozhoduje v českých politických stranách? Vzestup nových politických podnikatelů ve srovnávací perspektivě. Středoevropské politické studie 17, no. 2, Kostadinova, P Democratic performance in post-communist Bulgaria: election pledges and levels of fulfillment, East European Politics 29, no. 2, Lindgren, E What words can tell: Effects of emotive and vague words on voters interpretation and evaluation of election campaign proposals. VIM Conference at McMaster University, Ontario, May 20-22, Lo, J. - Proksch, S-O. Slapin, J. B Ideological clarity in multiparty competition: A new measure and test using election manifestos. British Journal of Political Science, Louwerse, T Political Parties and the Democratic Mandate Comparing Collective Mandate Fulfilment in the United Kingdom and the Netherlands. Dissertation Thesis, Universiteit Leiden. Mansergh, L. Thomson, R Election pledges, party competition, and policymaking. Comparative Politics, s Milita, K. Ryan, J. B. Simas, E. N Nothing to hide, nowhere to run, or nothing to lose: Candidate position-taking in congressional elections. Political Behavior 36, no. 2, Morgenstern, S. Zechmeister, E Better the devil you know than the saint you don t? Risk propensity and vote choice in Mexico. Journal of Politics 63, no. 1, Naurin, E Is a promise a promise? Election pledge fulfilment in comparative perspective using Sweden as an example. West European Politics 37, no. 5, Petrocik, J. R Issue ownership in presidential elections, with a 1980 case study. American Journal of Political Science, Pétry, F. Collette, B Measuring how political parties keep their promises: A positive perspective from political science. In L.M. Imbeau (ed.) Do They Walk Like They Talk? Springer: New York, Pomper, G. M. (1968). Elections in America: Control and Influence in Democratic Politics. New York: Dodd, Mead and Company.