Sweeping it under the rug: How government parties deal with deteriorating economic conditions

Abstract

Party competition in advanced industrial democracies is generally characterized as a two-dimensional space consisting of an economic and non-economic dimension. This study examines (a) the extent to which parties strategically place more emphasis on one of these dimensions vis-à-vis the other, something we coin relative emphasis, and (b) the extent to which voters perceive such shifts in relative emphasis. Our specific focus here is on government parties. We expect government parties to shift emphasis away from the economic to the non-economic dimension when economic conditions deteriorate. In doing so, they aim to reduce the importance voters attach to the economy and the degree to which voters attribute responsibility for the economy to the government. By combining expert data for 232 parties with survey data for roughly 30,000 individuals in 28 European countries in 2014, our analysis shows that while government parties generally pay more attention to the economic dimension, they shift attention to the non-economic dimension when economic conditions deteriorate. In contexts where government parties have shifted attention away from the economic to the non-economic dimension, voters overall attach less importance to the economy and attribute less responsibility to the government for the state of the economy.

Introduction

Why do political parties decide to prioritize some issues and downplay others? This question is at the heart of our understanding of party competition. While an infinite number of issue dimensions could theoretically define competition between political parties, there are practical limits to the amount of issues parties can mobilize. Most importantly perhaps are the cognitive limitations of voters (Downs, 1957; Nie et al., 1976; Kinder and Fiske, 1986). Voters use broad-based ideological labels, such as left and right, as heuristics to help them decide which party to vote for at election time (Hinich and Munger, 1996; Miller and Shanks, 1996). Reducing the number of issue dimensions is also beneficial to parties themselves as it helps them to deal with a collective action problem, namely the mobilization of the electorate (Aldrich, 1995).

Party competition, at least in advanced industrial democracies, is generally perceived to consist of two dimensions (for an overview see De Vries and Marks, 2012). Scholars use a variety of terms to denote these two dimensions, the materialism versus post-materialism (Inglehart, 1990), the old versus new politics (Dalton, 2002), the interest versus values (Tavits and Potter, 2015), the left/right versus liberitarian/authoritarian (Kitschelt, 1994) or the left/right versus Green, Alternative, Libertarian/Traditional, Authoritarian, Nationalist (GAL/TAN) dimensions (Hooghe et al., 2002) for example. What these different understandings have in common is that they rely on a general distinction between a largely economic and largely non-economic dimension.¹ The economic dimension relates to collective questions about state intervention in the economy, while the non-economic dimension refers to societal value divides based on identity, religion or the environment for example (see Rovny and Polk, this issue). While the question of why parties pay attention to specific issues or individual dimensions has received abundant scholarly attention, think of the important work on issue ownership for example (e.g. Budge and Farlie, 1983; Petrocik, 1996), the extent to which (a) parties aim to strategically shift attention from one dimension to another, something we coin relative emphasis, and (b) the extent to which voters perceive such shifts in emphasis has sparked off much less scholarly interest. What is more, our current understanding of the relative emphasis that parties place on the economic vis-á-vis the non-economic dimension largely stems from single country contexts or examinations of specific sets of parties (e.g. De Vries and Hobolt, 2012; Meguid, 2008; Pardos Prado and Sagarzazu, 2015, this issue; Vavreck, 2009).

We currently lack a clear comparative understanding of the mechanisms that explain how relative emphasis works, when parties employ it, and with which aim (for first attempts, see Hobolt and De Vries, 2015, Tavits and Potter, 2015). Although this study will not be able to address all these important issues, it attempts to provide insights into at least some of them. The specific focus here is why government parties, who generally pay more attention to the economy (Vavreck, 2009) and hold electoral advances on the economic dimension (Riker, 1982), sometimes shift emphasis away from the economic dimension. We suggest that the state of the economy is crucially important in this respect. The state of the economy is probably one of the easiest ways for voters to make an assessment about government performance and hold them accountable for their time in office (Holbrook, 1996; Lewis Beck, 1988). Economic voting is a cornerstone of electoral accountability (see Duch and Stevenson, 2008; Lewis Beck and Stegmaier, 2000, for excellent overviews). Against this backdrop, government parties have a clear incentive to shift the debate away from the economy when economic conditions deteriorate. The key strategic reason to do this, we suggest, is an attempt to shift the issue priorities of voters and dodge responsibility for the state of the economy.

We test these propositions in a comparative fashion. We rely on Chapel Hill Expert Survey (CHES) data for 232 parties with European Election Survey (EES) data from roughly 30,000 individuals in 28 European countries in 2014 and complement these sources with economic data from Eurostat. These data sources allow us to empirically test our argument in a large set of countries that provide ample variation in economic conditions. We report two main findings: (a) although government parties generally pay more attention to the economic dimension, they shift attention to the non-economic dimension when economic circumstances worsen; and (b) in contexts where government parties shift attention away from the economic to the non-economic dimension voters view the economy as less important and hold government parties less responsible for the state of the economy.

Due to a lack of time-series data, we cannot fully disentangle the causal direction of these relationships. Notwithstanding, it seems fair to expect that voters will pay more attention to the economic dimension vis-à-vis the non-economic one when faced with an economic downturn. The fact that we do not find this in contexts where government parties play up the importance of the non-economic dimension at the expense of the economic one suggests that the strategic activities of government parties may at least to some extent shape voter perceptions.

Our finding that government parties aim to redirect attention away from the economic to the non-economic dimension when economic conditions deteriorate corroborates the results presented by Pardos Prado and Sagarzazu (this issue). Using longitudinal data from Spain, the authors convincingly show that government parties are more likely to stress nationalist conflicts at times of economic downturn. Moreover, our suggestion that government parties strategically aim to shift emphasis from the economic to the non-economic dimension nicely complements the work by Rovny and Polk (this issue) highlighting how the historical roots of religious conflicts shape the attention parties place on the non-economic dimension. While historical legacies clearly condition party emphasis of non-economic issues, government participation and fluctuations in the economy may be important for understanding variation across time and across countries.

This study locates itself at the intersection of party competition and economic voting and builds on insights from the political psychology literature regarding responsibility attribution. Our findings provide important implications for all three sets of literatures as well as inform our understanding of party strategy and the role parties play in securing representation. We discuss these implications in the concluding section. Overall, we proceed in four steps. First, we introduce our theoretical conjectures and hypotheses. In a next step, we elaborate the data, operationalizations and methods we employ in the empirical analysis. Third, we provide an overview of our empirical findings. Fourth and finally, we discuss the implications of our study.

Dimensionality, government parties and the economy: Four hypotheses

Scholarly work suggests that party competition is largely two-dimensional in nature. It consists of an economic dimension pitting parties that favour a more redistributive role for the state against those that favour a market-oriented role, and a non-economic dimension that places national conservative parties against cosmopolitan liberal parties (Benoit and Laver, 2006; Hooghe et al., 2002; Kitschelt, 1994; Kriesi et al., 2008). Why do parties place more emphasis on one dimension vis-à-vis the other? Recent work by Tavits and Potter (2015) highlights the importance of societal demand. The authors suggest that left-wing parties place more emphasis on the economic vis-á-vis the non-economic dimension when inequality rises. Inequality increases the proportion of the population falling in lower socioeconomic strata thus expanding the share of voters who could be receptive to the economic message of left-wing parties. As inequality rises, the constituency favouring less market intervention in the economy as espoused in right-wing party platforms is likely to shrink, and right-wing parties are likely to shift their emphasis to values in response.

These findings are important as they suggest that changes in economic conditions shape the societal demand for policy, and this in turn affects the relative emphasis parties place on the economic versus non-economic issues.

Yet, they also raise the question of the extent to which parties themselves may shape societal demand. This study contributes to this work by suggesting that changes in economic conditions or inequality may not only indirectly affect parties’ relative emphasis via societal demand but also directly based on parties’ strategic incentive to divert public attention away from their policy responsibilities. Specifically, we suggest that government parties have a strategic incentive to redirect attention away from economic to non-economic issues when economic conditions worsen.

Why is this the case? Economic performance is one of the most important issues for voters when deciding to re-elect an incumbent or not (Duch and Stevenson, 2008; Lewis Beck, 1988). The economic voting literature suggests that as economic conditions deteriorate and people evaluate the economy more negatively, the re-election prospects of incumbents decline (Lewis Beck and Stegmaier, 2000). One way in which government parties may wish to evade the punishment of the electorate is by affecting the way people perceive the economy. Extensive research shows that partisans tend to perceive the economy in a way that credits their own party, thus weakening the relationship between the real economy and individual perceptions thereof (e.g. Bartels, 2002; De Boef and Kellstedt, 2004; Evans and Andersen, 2006; Evans and Pickup, 2010; Lavine et al., 2012; Wlezien et al., 1997; Zaller, 1992). Yet, recent work suggests that incumbents often have limited success in changing people’s economic perceptions, although opposition parties are more successful (Pardos Prado and Sargazazu, 2015). This is largely the result of the direct experience voters have with the incumbent’s handling of the economy which limits the room for government parties to present a different version of reality (De Boef and Kellstedt, 2004). ‘While the incumbent has to deal with “real events,” the challenger can be judged on “only talk”’ (Pardos Prado and Sargazazu, 2015: 804). Although it might prove difficult to change people’s economic evaluations, especially when economic signals are strong (Parker Stephen, 2013; Pardos Prado and Sagarzazu, 2015), we suggest that government parties could attempt to change people’s issue priorities and responsibility attributions in order to increase their chances of re-election (Peffley, 1985; Rudolph, 2003; Tilley and Hobolt, 2011).

Important work in the field of political psychology stresses the centrality of responsibility attributions for retrospective voting generally and economic voting specifically (e.g. Feldman, 1982; Kinder and Mebane, 1983; Lau and Sears, 1981; Peffley, 1985; Rudolph, 2003). These authors have argued that economic perceptions only inform people’s decisions at the ballot box when responsibility for the economy is attributed to the incumbent government:

Before economic discontents take on political significance, people must either believe that the government produced them or that it is the government’s job to remedy them. These beliefs revolve around issues of responsibility […and] the impact of economic perceptions on political behavior is [conditioned] by judgments of accountability (Peffley, 1985: 192).

Rather than aiming to alter economic evaluations directly, government parties may be more successful at influencing voters’ perceptions of who is responsible for the economy and how important the economy is (Tilley and Hobolt 2011). Economic voting studies suggest that these responsibility attributions or issue priorities affect the extent to which voters hold government parties to account for the state of the economy at the ballot box (De Vries and Giger, 2014). When policy responsibility is clear, economic voting is more pronounced (e.g. Cutler, 2004; Powell and Whitten, 1993; Tilley and Hobolt, 2011). With this in mind, we suggest that an optimal strategy for incumbent parties facing deteriorating economic conditions is to shift emphasis away from economic to non-economic issues in order to try to influence the emphasis voters place on the economy and their judgements of government responsibility for the state of it. In her work on US presidential campaigns, Vavreck (2009) suggests that in bad economic conditions incumbents’ are best served by putting advantageous non-economic issues on the agenda. We test this proposition in a comparative context across a large set of countries as well as outside a campaign context, and hypothesize that:

• Relative Emphasis Hypothesis: Government parties that face deteriorating economic conditions will shift emphasis away from the economic to the non-economic dimension, while government parties in improving economic conditions will not.

Do all government parties have a strategic incentive to redirect emphasis away from the economic to the non-economic dimension when faced with worsening economic circumstances? Most likely not. Based on the existing literature, we can think of several sources of heterogeneity in relative emphasis of government parties, for example, issue ownership or holding the office of the prime minister. The literature on issue ownership suggests that parties should emphasize the policy issues they own (Petrocik, 1996). Ownership here refers to a situation in which the electorate, or at least large parts of it, views a party to be better and/or more credible at handling policy problems in a particular issue area. When it comes to the economy, left-wing parties in European countries for example are traditionally associated with the expansion of welfare state policies, while right-wing parties with economic growth (Budge et al., 1987).² Based on these differences in issue ownership, we would expect the responses to the deteriorating economic conditions to differ based on the ideological leanings of government parties. Specifically, we would expect right-wing parties in government to shift attention away from the economic dimension to the non-economic one when economic growth slows as this would allow them to hide bad performance on an issue they own. At the same time, we expect government parties on the left who are associated with more generous welfare spending to benefit electorally from highlighting the economic dimension over the non-economic one. Although voters across different countries might of course differ in the way they perceive right- and left-wing parties to have ownership of the economic dimension and variations across economic issues, such as growth, unemployment or inflation, might exist, recent work by Tavits and Potter (2015) underscores these general expectations. The authors suggest that due to different expectations about the societal demand for policy, left-wing parties will shift emphasis towards the economic dimension when faced with rising inequality, while right-wing parties will shift emphasis away from the economic to the non-economic dimension. Based on these different insights, we expect left-wing parties in government to devote more attention to economic issues when economic conditions deteriorate as they may wish to capitalize on their reputation for more generous welfare state policies, while right-wing parties facing the same economic circumstances should place less emphasis on the economic dimension, because they are more strongly associated with sound economic management. This leads us to our second hypothesis which states that:

• Issue Ownership Hypothesis: Right-wing government parties facing deteriorating economic conditions are likely to shift emphasis away from the economic to the non-economic dimension, while left-wing government parties facing the same conditions are likely to shift emphasis towards the economic dimension from the non-economic one.

A second source of heterogeneity among government parties might stem from holding the office of the prime minister or not. Evidence from the literature on economic voting suggests that the party of the prime minister might be rewarded or punished more strongly for economic up- or downturns compared to other government parties (Anderson, 2000; Lewis Beck, 1997). Anderson (2000) for example suggests that voters might use the party of the prime minister as a way to determine who is in charge and simplify the decision of who to blame or credit for the state of the economy. Because it is likely that the prime minister party would get most of the blame for bad economic performance, we would expect the incentives to shift attention away from the economic to the non-economic dimension when economic conditions worsen to increase for the prime minister party compared to other parties in the coalition. This leads us to the formulation of our third hypothesis which states that:

• Prime Minister Party Hypothesis: When faced with deteriorating economic conditions the party of the prime minister is more likely to shift emphasis away from the economic to the non-economic dimension compared to other parties in the government coalition.

We suggest that government parties aim to shift attention from the economic to non-economic dimension in order to change the emphasis people attach to the economy and the degree to which they hold the government responsible for the state of the economy. This expectation of course raises the question of the direction of the causal relationship, do people respond to the strategic changes in emphasis by government parties or the reverse? Although this is a difficult question to empirically deal with, most of the evidence to date suggests that it is a combination of both. Parties both respond to societal demands and are able to shape societal demand (Gabel and Scheve 2007; Steenbergen et al., 2007). Although this is important to acknowledge, it seems reasonable to expect that people facing an economic downturn will pay more attention to the economic vis-à-vis the non-economic dimension. If our results indicate that this is not the case in contexts where government parties pay more attention to the non-economic dimension at the expense of the economic one, this would be a noteworthy result. If we were to find that people residing in contexts where government parties pay more attention to the non-economic vis-à-vis the economic dimension view the economy as less important and hold the government less responsible for the state of the economy compared to people in contexts where the government pays more attention to the economic dimension at the expense of the non-economic one, it suggests that the strategic activities of government parties are important for voter perceptions. This leads to the formulation of our fourth and last hypothesis that states:

• Citizen Response Hypothesis: In contexts where government parties place less emphasis on the economic dimension vis-á-vis the non-economic one, citizens will view the economy as less important and the government as less responsible for the economy compared to contexts in which government parties place more emphasis on the economic dimension.

Data, operationalizations and model specification

In order to test these four hypotheses, we rely on two rich data sources that allow us to empirically capture our key theoretical concepts in all 28 European Union (EU) member states, namely the 2014 CHES and the 2014 EES. The joint timing of these surveys not only allows us to connect party-level with individual-level data in a hierarchical model but also provides us with ample variation in economic conditions, something that is crucial for the empirical examination of our argument. Ideally, we would have extended the time-frame of investigation, but previous rounds of the CHES and EES do not include all measures required for the empirical analysis. Another alternative data source for party’s emphasis could have been the Comparative Manifesto Project (CMP, Volkens et al., 2012). This incredible valuable data source is based on party manifestos that are released only at election time and designed to capture long-terms shifts in a party’s policy position. This makes it more challenging to pick up short-term change that we are interested here, and at times hard to link the time-point of observation to actual changes in economic conditions. CHES data include estimates of the emphasis a party places on a variety of issue dimensions at a specific time-point for all parties included in the data set (Bakker et al., 2015a) and allow for a linkage to voter survey data covering the same year. Moreover, studies have cross-validated these CHES measures against the CMP for example and concluded that expert surveys outperform the other salience measures in terms of explanatory power (Netjes and Binnema, 2007).

Our empirical strategy has two parts. The first part examines our party-level hypotheses, namely the relative emphasis, issue ownership and prime minister party hypotheses. In order to do so, we construct a database where we combine the CHES data on emphasis, government and prime minister party status as well as ideology (including controls) with economic data from Eurostat. In the second part, we turn to the individual-level analysis to explore if the relative emphasis by government parties is correlated with the emphasis of citizens and the extent to which they hold these parties responsible for the state of the economy. In order to do so, we add the score of our party-based relative emphasis measure averaged across all government parties based on the CHES data to an individual model aiming to estimate people’s economic emphasis and responsibility attributions using the EES data. Let us elaborate the operationalizations used in each step in turn.

In order to examine if government parties shift their emphasis away from the economic to the non-economic dimension when economic conditions deteriorate, we need empirical measures of relative emphasis, government status and economic deterioration. We construct our relative emphasis measure by subtracting the emphasis a party places on the non-economic dimension from the emphasis they place on the economic left/right dimension as measured in CHES 2014. In the 2014 round of the survey, country experts were asked to evaluate the importance of issue dimensions to individual parties on a scale ranging from (0) ‘no importance’ to (10) ‘great importance’ (Bakker et al., 2015b). In terms of the economic dimension experts were first told that ‘[p]arties can be classified in terms of their stance on economic issues. Parties on the economic left want government to play an active role in the economy. Parties on the economic right emphasize a reduced economic role for government: privatization, lower taxes, less regulation, less government spending, and a leaner welfare state’, second prompted ‘to think about the salience of economic issues for a party’, and finally asked ‘how important economic issues [were] to the parties in their public stance [over the course of 2014]?’ (Bakker et al., 2015b: 7–8) Regarding the non-economic dimension, the same country experts were first told ‘[p]arties can be classified in terms of their views on democratic freedoms and rights. “Libertarian” or “postmaterialist” parties favor expanded personal freedoms, for example, access to abortion, active euthanasia, same-sex marriage, or greater democratic participation. “Traditional” or “authoritarian” parties often reject these ideas; they value order, tradition, and stability, and believe that the government should be a firm moral authority on social and cultural issues’, and then asked ‘how important were libertarian/traditional issues to the parties in their public stance [over the course of 2014]?’ (Bakker et al., 2015b: 8–9) Our relative emphasis measure can potentially vary between (−10) indicating that a party adds great importance to non-economic issues and no importance to economic issues to (+10) indicating that a party adds great importance to economic issues and no importance to non-economic issues. A value of zero indicates that a party attaches equal weight to the economic and non-economic dimension.

Our measure of government status captures if a party was in government or part of a governing coalition in 2014 and takes on a value of 1 if a party was in government for the entire year or at least 6 months of the year, 0 if it was in opposition for the entire year or at least 6 months of the year. To explore a possible heterogeneity between the prime minister party and other government parties, we coded if a government party held the office of the prime minister (1) or not (0) based on the ParlGov database (Döring and Manow, 2016). We examine the possible heterogeneity between government parties based on issue ownership by differentiating between left- and right- wing government parties based on expert evaluations of their positions on an 11-scale economic left/right dimension (0 indicates extreme left and 10 extreme right). Specifically, we coded parties holding positions below the mean party position in a country as left, and those above as right. Our measure of economic deterioration captures the change in the GDP growth rate between 2013 and 2014 based on data from Eurostat (2015). Specifically, we subtract the 2014 growth rate from the 2013 one. Since we are only interested in a deterioration of economic conditions versus not, we code stable or negative changes in growth rates between 2014 versus 2013 as (1), and positive changes as (0). As a robustness check, we also examined different levels of economic deterioration and growth using a continuous measure. This analysis yields similar results (see Table A.2 in the Supplemental Appendix). We use growth rates rather than inflation or unemployment rates for three reasons. First, due to the fact that 19 of 28 EU member states studied here are part of the Eurozone and have a common currency, inflation is largely determined by the interest rates set at the Eurozone level by the European Central Bank. Second, in 2014, growth rates provide us with the much more variation in economic performance across countries than unemployment, and thus allows for the best empirical test of our hypothesis. Third, with the exception of deep economic shocks unemployment rates change rather slowly and thus do not allow us to capture much variation from one year to the next. Economic growth levels fluctuate much more. In addition to these variables, our models also control for the ideological leaning of parties and their size. A party’s ideological leaning is expected to shape the emphasis parties place on political issues and is measured by their position on a general left-right scale ranging from (0) extreme left to (10) extreme right. Party size is measured as the percentage of votes obtained in the latest European parliamentary elections. This measure is included as it may be argued that larger parties with more personnel can pay attention to more issues simultaneously (e.g Netjes and Binnema, 2007).

In order to examine our citizen response hypothesis stating that people will place less emphasis on the economy and hold government parties overall less responsible for the economy in contexts where government parties place less emphasis on the economic dimension versus the non-economic one, we rely on individual-level data from the 2014 EES. The 2014 EES is a post-election survey with representative samples in each of the 28 EU member states (Schmitt et al., 2015). This EU-wide survey allows us to examine both people’s views of the importance of the economy and the government responsibility for the state of it in all 28 EU member states using identical questions. In the case of importance, we combine people’s responses to two questions, namely asking them what the most and second most important issues were when voting in the European elections.³ On the basis of people’s responses, we created a variable that takes on the value of (1) when respondents viewed unemployment or economic growth as the first or second most important, and a value of (0) when they viewed other issues as most important. In the case of government responsibility judgements, we rely on the following item:

Now I would like to ask you some questions about how much responsibility the [government has on] the current economic situation in [country]. Please use a scale from 0 to 10, where 0) means that you think they have ‘no responsibility’ and 10) means that they have ‘full responsibility’.

To explain variation in the importance people attach to the economy of the way they attribute responsibility for the state of it, we must account for the variation in relative emphasis placed by government parties. We capture the relative emphasis placed on the economic vis-á-vis the non-economic dimension by government parties in a country by taking the average relative salience of all government parties within the same country. As a robustness check, we also created a relative emphasis measure based on a weighted average based on the vote share of government parties. Table A.3 in the Supplemental Appendix provides the results based on this measure and shows that they are similar to the non-weighted measure. Finally, we add some individual-level controls, such as government partisanship,⁴ retrospective economic evaluations,⁵ political sophistication⁶ and a set of demographic variables (gender, age, education,⁷ occupation and unemployment⁸).

In the party-level analysis, our unit of observation is a party who places emphasis on the economic versus non-economic dimension in 2014 in a given country. In the individual-level analysis, our unit of observation is an individual who attaches importance to economy or responsibility for the state of it to the incumbent in 2014 in a given country. Due to the natural hierarchies in the data, both parties and individuals are nested in countries, we employ a hierarchical (or multilevel) linear model (HLM). Snijders (2011) suggests that using a HLM technique allows for a single model that incorporates the different levels of data without assuming a single level of analysis and for the modelling of random intercepts at both levels of analysis, the party/individual and country level. While the intercept in traditional linear regression represents the value taken by the dependent variable when all explanatory variables are held at zero, the random intercepts that are estimated in a HLM regression indicate that for any given country and any given party/individual, there are different baseline propensities that modify this traditional observation-level intercept. This allows us to account for the fact that parties and individuals within a country may be more likely to share common characteristics (and errors) compared to parties and individuals in other countries. Overall, HLM regression models facilitate the exploration of the generalizability of findings across different contexts. As a robustness check, we also employed a fixed effects model to deal with country clustering and results are reported in Table A.1 in the Supplemental Appendix. The results based on fixed effects or HLM models yield similar results.

Empirical results

The three models presented in Table 1 provide an empirical examination of the three party-level hypotheses, namely the relative emphasis, issue ownership and prime minister party hypotheses. They aim to explore (a) if government parties that face a deterioration of economic conditions place less emphasis on the economic and more on the non-economic dimension, (b) if we find differences between government parties based on their issue ownership (parties on the left versus right) or (c) the fact that they have supplied the prime minister or not. The reference category in all analyses is being a party in opposition. The dependent variable in all three models presented in Table 1 is our relative emphasis measure, namely the emphasis placed by a party on the non-economic dimension subtracted from the emphasis that same party places on the economic dimension. The results do not only provide coefficient estimates accompanied by standard errors in parentheses but also measures of the variance in relative emphasis at both levels of analysis, the party and country level. There is substantial variance in relative emphasis at the country level which at least in part reflects the heterogeneous economic conditions found within the 28 member states of the EU (see Stiglitz, 2016 for a discussion).

Table 1.

Hierarchical models testing government reaction to worsening of economic conditions.

	Model 1	Model 2	Model 3
Government party	1.78** (0.32)
Left-wing		1.00** (0.39)
Right-wing		2.63** (0.40)
Prime Minister			1.95** (0.50)
Other government			1.73* (0.35)
Economic change t −(t−1)	0.78 (0.85)	0.81 (0.81)	0.79 (0.85)
Government* economic deterioration	−3.27** (0.68)
Left-wing* economic deterioration		−2.63** (0.73)
Right-wing* economic deterioration		−3.86* (1.13)
Prime Minister* economic deterioration			−2.93** (0.99)
Other government* economic deterioration			−3.43** (0.77)
Left-right position	−0.22** (0.06)	−0.29** (.06)	−0.22** (0.06)
Party size	0.04** (0.01)	0.04** (.01)	0.04* (0.02)
Intercept	1.45** (0.48)	1.77** (.47)	1.48** (0.48)
${\widehat{\sigma }}_{pc}$	1.87** (0.09)	1.84** (.09)	1.87** (0.09)
${\widehat{\sigma }}_c$	1.52** (0.25)	1.43** (.24)	1.52** (0.25)
N pc	232	232	232
N c	28	28	28

Note: The dependent variable is the difference between the level of emphasis a party places on the economic dimension versus the non-economic dimension. Standard errors appear in parentheses below coefficient estimates. The reference category is being an opposition party.

* Significant at the p ≤ 0.05 level.

** Significant at the p ≤ 0.01 level.

Turning to the results of model 1, exploring the relative emphasis of government parties when economic circumstances are deteriorating, we would expect the coefficient for being a government party to be positively signed as research suggests that incumbents pay more attention to economic issues (Vavreck, 2009). The interaction between being a government party and facing an economic deterioration compared to last year should be negatively signed, indicating that government parties will de-emphasize the economic dimension compared to the non-economic dimension when economic circumstances are worsening. Note that we aim to address possible endogeneity concerns by taking a negative change in economic conditions between t (2014) and t − 1 (2013). The results in model 1 in Table 1 support these expectations. Not only is the coefficient for being a government party positively signed and statistically significant indicating that overall government parties emphasize the economic dimension more than opposition parties, the interaction is negatively signed and statistically significant. In order to examine if there is a significant difference in relative emphasis between government parties facing an economic deterioration versus government parties that do not, we perform a post estimation hypothesis test.⁹ The difference in relative emphasis of government parties in deteriorating economic circumstances versus government parties in improving economic circumstances is −2.49 and statistically significant (p value = 0.01). This result indicates that government parties in bad economic times shift emphasis away from the economic versus non-economic dimension compared to their counterparts in good economic times. Given that our relative emphasis measure runs from (−10) full emphasis on the non-economic and no emphasis on the economic dimension and (+10) full emphasis on the economic and no emphasis on the non-economic dimension, this 2.49-point negative change is both statistically and substantially significant.

Model 2 presents the same analysis, but now splits government parties based on their ideological left/right leaning. Building on the issue ownership literature, we expect left-wing government parties to shift emphasis towards the economic away from the non-economic dimension when economic conditions deteriorate as the left expects to be able to capitalize on their reputation of providing welfare, while right-wing government parties are expected to shift attention away from the economic dimension. The results presented in model 2 of Table 1 suggest that this expectation is not borne out by the data. The coefficients for a left-wing or right-wing party in government are both positively signed and statistically significant. This suggests that both left-wing and right-wing parties in government place more emphasis on the economic dimension than their counterparts in opposition. The coefficients for the interaction term with an economic deterioration are both negatively signed and statistically significant. This already suggests that there might be little difference in relative emphasis between the left-wing and right-wing parties in an economic downturn. In order to examine this more rigorously, we performed a post estimation hypothesis test to explore if the relative emphasis of a left-wing government party in an economic downturn is different from the relative emphasis of a right-wing party in government during the same circumstances. The difference in the relative emphasis between left-wing and right-wing government parties facing an economic deterioration is small, −0.39, and not statistically significant (p value = 0.71). When we explore the difference in relative emphasis between left-wing and right- wing parties in government during good economic times by comparing the coefficient for left-wing and right-wing government party, we find that left- wing parties pay considerably less attention to the economic versus non-economic dimension when economic conditions are good, −1.63, compared to right-wing government parties, and that this difference is statistically significant (p value = 0.001). That said, we find little empirical support for the idea that right-wing parties in government are more likely to shift emphasis away from the economic to the non-economic dimension during economic downturns compared to the left-wing parties in government.

We explore differences in relative emphasis between prime minister parties and other government parties. We know from the economic voting literature that the prime minister party will be blamed more for deteriorating economic performance compared other parties in the government coalition (Anderson, 2000). As a result, we expect the prime minister party to have more incentive to shift attention away from the economic to the non-economic dimension when economic conditions worsen compared to other parties in the coalition. The results of this analysis are presented in model 3 of Table 1 and we find little empirical support for this expectation. The coefficients for being a prime minister party or simply another party in government are both positively signed and statistically significant. This suggests that both prime minister parties and other parties in government parties place more emphasis on the economic dimension than their counterparts in opposition. The coefficients for the interaction term with an economic deterioration are both negatively signed and statistically significant. This provides us with a first hint that there might not be much of a difference in relative emphasis between the prime minister party and other government parties in an economic downturn. In order to examine this more in-depth, we perform another post estimation hypothesis test to explore if the relative emphasis of a prime minister party facing an economic deterioration is different from the relative emphasis of other parties in government when economic conditions worsen. The difference in the relative emphasis of a prime minister versus other government parties when faced with an economic deterioration is about half a point, 0.72, but fails to reach statistical significance (p value = 0.47). When we explore the difference in relative emphasis between a prime minister party and other parties in government in good economic times by comparing the coefficient for prime minister party and other government party, we find that prime minister parties pay slightly more attention to the economic versus non-economic dimension in economic good times, 0.22, compared to other government parties, but that this difference fails to reach statistical significance (p value = 0.68). Overall, we find little empirical support for the prime minister party hypothesis.

While we find evidence for the relative emphasis hypothesis, we find no empirical support for the issue ownership or prime minister party hypotheses. These findings suggest that although other studies have shown that ideological leanings may affect the emphasis parties place on different dimensions, also in the context of economic changes, when parties are in government and can expect to be held accountable for their economic performance, they all have a strategic incentive to downplay economic issues in favour of non-economic issues. Although prime minister parties may be punished or rewarded more for government actions compared to other coalition partners, when it comes to shifting emphasis away from the economic dimension when economic conditions are bad government parties act similarly. This evidence supports Vavreck’s proposition about incumbent behaviour in presidential campaigns, namely that the optimal strategy for incumbents facing bad economic conditions is to emphasize advantageous non-economic issues. We find evidence of this strategy in a large set of countries and within as well as outside campaign contexts. All these results hold when we employ a fixed effects model to deal with country clustering (see Table A.1 in the Supplemental Appendix).

After having established that government parties respond to worsening economic conditions by shifting their emphasis away from the economic to the non-economic dimension, we now turn to the extent to which these shifts are perceived by ordinary citizens. Of course, there are a variety of ways to capture citizen responses. Here we rely on two measures: (a) the importance people attach to the economy, and (b) the degree to which they attribute responsibility to the government for the state of the economy. By relying on both an economic importance and attribution measure, we aim to capture the extent to which people distinguish between deteriorating economic conditions and the government’s response to it. We expect that people who reside in contexts where government parties shifted emphasis away from the economic to the non-economic dimension view the economy as less important and hold government parties less responsible for the economy compared to those residing in contexts where government parties place more emphasis on the economic vis-á-vis the non-economic dimension. As we already alluded to earlier, our data does not allow us to disentangle the causal direction of these relationships. While empirical evidence suggests that voters respond to parties as well as the reverse (Gabel and Scheve 2007; Steenbergen et al., 2007), it seems reasonable to assume that when economic conditions deteriorate, citizens will most likely pay more attention to the economic vis-à-vis the non-economic dimension. So if we find that this is less the case in contexts where government parties shifted importance away from the economic dimension, this would be largely consistent with the notion that government parties try to strategically place emphasis on issues to change voter perceptions.

To explore this possibility, we create a relative emphasis score by taking the average relative emphasis of all government parties in a country and use it as a country-level predictor of importance attached to the economy and responsibility attributions for the state of the economy. Higher scores indicate that government parties on average attach more salience to the economic dimension compared to the non-economic dimension. If we weigh the average relative emphasis of government parties by vote share this yields similar results (see Table A.3 in the Supplemental Appendix). We expect the importance that citizens attach to the economy and the degree to which they hold the government responsible for economic conditions to be higher in contexts where government parties place more emphasis on the economic dimension compared to the non-economic one and lower in contexts where government parties place less emphasis. Accordingly, we expect the coefficients for Relative Emphasis of Government Parties to be positive and statistically significant as higher values of relative emphasis indicate that government parties on average place more emphasis on the economic vis-à-vis non-economic dimension. We control for identification with the government, retrospective economic evaluations, political sophistication and varying socio-demographic profiles of individuals.

The results of these analyses are reported in models 1 and 2 in Table 2. The results presented in model 1 in Table 2 suggest that people who reside in contexts where government parties on average pay more attention to the economic dimension at the expense of the non-economic one view the economy as more important compared to people residing in contexts where government parties on average pay less attention to the economic compared to the non-economic dimension. The effect is statistically significant, but with 0.02 rather small. This might not be entirely surprising given that on average most people view issues such as unemployment and economic growth as extremely important already. Given this ceiling effect, it is noteworthy that we still find that people residing in contexts where government parties emphasize the economic dimension think that the economy is more important. We also find that people in the same contexts hold the government more responsible for the economy. The increase of 0.12 is statistically significant. These results hold while controlling for government party identification and people’s retrospective economic evaluations.

Table 2.

Hierarchical models testing citizen response to relative emphasis of government parties.

	Economic importance Model 1	Economic attribution Model 2
Relative emphasis of government parties	0.02* (0.01)	0.12** (0.06)
Retrospective economic evaluations	−0.01* (0.004)	−0.01 (0.02)
Government party identifier	0.07** (0.01)	−0.03 (0.04)
Education	−0.01** (0.005)	0.13** (0.02)
Working-class identification	0.03* (0.01)	−0.22** (0.05)
Professional worker	−0.01 (0.01)	0.13* (0.06)
Unemployed	0.04** (0.01)	−0.01 (0.05)
Self employed	−0.02 (0.02)	0.03 (0.08)
Age	−0.003* (0.0002)	0.001 (0.001)
Gender	0.005 (0.01)	0.05 (0.03)
Political sophistication	0.02* (0.01)	0.21** (0.03)
Intercept	0.76** (0.04)	6.78** (0.19)
${\widehat{\sigma }}_{ic}$	0.46** (0.002)	2.55** (0.01)
${\widehat{\sigma }}_c$	0.11** (0.01)	0.58** (0.08)
N ic	16,931	29,859
N c	28	28

Note: The dependent variable in model 1 is if a respondent thinks that unemployment or economic growth are the most important issues, and in the model 2 the degree to which a respondent thinks that the government is responsible for the state of the economy. Standard errors appear in parentheses below coefficient estimates. Table A.4 in the Supplemental Appendix estimates a hierarchical logit model for Economic Importance which yields similar results.

* Significant at the p < 0.05 level.

** Significant at the p < 0.01 level.

Overall, our empirical results suggest that (a) government parties generally pay more attention to the economic dimension; (b) shift attention to non-economic dimension when economic conditions deteriorate; and (c) that this pattern is uniform across government parties. Moving our attention from the party to the individual level, we find that in contexts where government parties on average place more emphasis on the non-economic dimension at the expense of the economic one, (d) people view the economy as less important and (e) hold government parties less responsible for the state of the economy compared to contexts in which government parties place more emphasis on the economic dimension vis-à-vis to the non-economic one.

Concluding remarks

This study suggests that the emphasis parties place on the two dimensions that generally characterize party competition in advanced industrial democracies, the economic and non-economic one, is not a stable equilibrium, but constantly subject to change. In line with previous work, most notably Tavits and Potter (2015), we suggest that partisan responses to changing economic conditions are crucial for understanding the variation in the relative emphasis parties place on one vis-á-vis the other dimension. Yet, unlike existing work we suggest that parties may not only respond to changing economic conditions because of the way these shape societal demand for policy but also to strategically affect the way in which the public responds. This study theoretically argues and empirically substantiates that parties in government respond to deteriorating economic conditions by shifting the emphasis they place on the economic dimension towards the non-economic one. A key strategic reason to do so is to try to alter the extent to which citizens view the economy as the most important issue and hold government parties responsible for the state of the economy.

While research suggests that it is difficult for government parties to alter the way people evaluate economic conditions, this study suggests that by shifting emphasis to non-economic issues they may attempt to dodge their responsibility for a worsening economy in the eyes of voters. Although these latter findings are intuitive, it is important to highlight that we provide evidence of a correlation between a shift in emphasis of parties and people’s views about the importance of the economy and who is responsible for the state of it. This study thus provides an important first step in gauging if the strategic redirection of emphasis away from the economic to the non-economic by government parties generates a voter response. In a next step, it is important to aim to causally identify this effect as issues of endogeneity remain. The findings presented here will hopefully spark off future work to carefully examine the success of relative emphasis strategies through experimental work. Moreover, future work should consider the consequences of strategic issue emphasis on party strategies more broadly. For example, might changes in party emphasis of government parties also spark off a repositioning on the economic to the non-economic dimension? Surely, governing parties have to take a position on economic dimension even when they attach less emphasis to it. So what do they do? Do they stay put, approach the main opposition party, or try to carve a niche position?

Our findings have important implications for the literature on political dimensionality, party competition and economic voting. When it comes to the literature on political dimensionality, our study suggests that possible differences in the emphasis that parties place on policy issues due to ideological leanings, that have been documented in the literature (e.g. Tavits and Potter, 2015), may at times be overridden by the strategic incentives parties face when they are in office. This intuition is also corroborated in the study of Pardos Prado and Sagarzazu (this issue) that reports similar dynamics using longitudinal data on party emphasis on the issue of nationalism in Spain. Recent work has highlighted the importance of government status for the introduction of new policy issues or wedge issues that split parties on the dominant dimension of party competition (De Vries and Hobolt, 2012; Hobolt and De Vries, 2015; Van der Wardt et al., 2014). This study suggests that government status may be an important factor more generally as it also conditions the attention parties attach to the two dominant issue dimensions that characterize party competition.

When it comes to party competition, our results indicate that parties are not merely vessels carrying societal divisions based on economic or value- based concerns but also actively structure and determine the content of societal conflict (e.g. Riker, 1982). Although party competition in Europe gravitates towards economic and value-based issues reflecting long-lasting societal divisions (Lipset and Rokkan, 1967), political parties can strategically choose to emphasize some concerns over others in an attempt to gear up public attention to controversies that benefit them. This suggests that parties not only represent the public but actively shape the public’s demand for representation.

Finally, the results presented in this study have implications for the study of economic voting. Recently, authors have highlighted the importance of responsibility attribution for the extent to which voters hold governments accountable for the economy (e.g. Rudolph, 2003, Tilley and Hobolt, 2011). This study suggests that government parties can, at least to some extent, manipulate these judgements. While existing research suggests that the ability of incumbents to change people’s evaluations of the state of the economy is rather limited as these are heavily influenced by people’s direct experiences or economic internationalization (e.g. De Boef and Kellstedt, 2004; Pardos Prado and Sagarzazu, 2015; Whitefield and Rohrschneider, this issue), parties in governments are not entirely powerless when economic conditions deteriorate. When faced with worsening economic circumstances, government parties are able to influence, at least in part, the extent to which people hold them responsible for these circumstances.

Author biographies

Catherine E De Vries is professor and holds the Westerdijk Chair of Political Behaviour in Europe in the Department of Political Science and Public Administration at the Vrije Universiteit Amsterdam. She is also an affiliated professor at the University of Essex and an associate member of Nuffield College at the University of Oxford. She acts as the scientific advisor for eupinions, an independent platform for European public opinion funded through the Bertelsmann Foundation. Over the years, she has published extensively on European Union politics, public opinion, corruption and remittances in leading journals such as the American Political Science Review, Journal of Politics, British Journal of Political Science and Comparative Political Studies. She recently published her book Euroscepticism and the Future of European Integration with Oxford University Press (2018).

Hector Solaz is a lecturer in the John Stuart Mill College at the Vrije Universiteit Amsterdam. He previously held positions at the University of Oxford and University of Essex. His main research interests lie within behavioural science and political economy. He mainly uses laboratory experimental methods to study the economic and political implications and foundations of behavioural phenomena. In particular, his research focuses on the determinants of cooperation, the emergence and enforcement of social norms, corruption and remittances. He has published extensively on these topics in leading general science journals, such as PLoS One, or leading political science journals such as the American Political Science Review and Comparative Political Studies.