Rhetoric on the economy: have European parties changed their economic messages?

ABSTRACT

This study analyses the determinants of dispersion of economic issue mentions in European party manifestos. We examined three main economic domains (governmental control of the economy, free market capitalism and support for the welfare state) as consequences of globalization forces, economic conditions, partisanship and electoral turnout. Employing aggregate-level Comparative Manifesto Project (CMP) data from legislative elections in 15 European countries from 1970 to 2010, we confirm that parties hold a common view of the salience of economic control of the state as a consequence of globalization pressure and economic growth levels. Partisanship of the cabinets (regardless of the political orientation) counteracted issue salience concentration in the welfare domain. Government size favoured dispersion in the free market realm. Our results do not indicate clear homogenization of parties’ economic messages in elections over the last 40 years.

KEYWORDS:

• Globalization
• political parties
• economic conditions
• electoral promises

JEL CLASSIFICATION:

• F68
• D72
• C23

Acknowledgements

The authors wish to acknowledge the attention of two anonymous referees of a previous version of this paper. Any remaining errors are the authors’ exclusively.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ These data are not used to reconstruct the parties’ policy positions but, more precisely, the salience of their policy objectives (Elff 2009).

² This category excludes education.

³ Relying on simple, unweighted measures may convey potentially misleading information because it ignores the positioning of the parties relative to the electorate given that party policies may be correlated with the vote (or seat) share (Alvarez and Nagler 2004; Ezrow 2007). For example, in the 1979 UK elections, the unweighted issue positions concerning welfare expansion were 8.4 for the Labour Party, 4.1 for the Liberal Party and 1.7 for the Conservative Party. The weighted values for welfare expansion emphasis were 3.10, 0.57 and 0.75 for the Labourists, Liberals and Conservatives, respectively. Compared to the unweighted scores, the emphasis devoted by the Labour Party and by the Liberals diminished by ~5 and 3.5 percentage points, respectively. The Conservatives exhibited lower variation (around 1 percentage point), thanks to their high vote share. The overall WPISD was quite high (4.235), indicating that in 1979 the UK parties were widely dispersed in this issue space.

⁴ We have also simply added all of the quasi-sentences from the CMP, summing both the positive and negative statements (Tavits and Potter 2014), yielding similar results to those commented upon here. Further details are available upon request.

⁵ Rae’s (1971) index of legislative fractionalization is highly correlated with the number of competing parties. Laakso and Taagepera (1979) developed an alternative measure (which is a better measure than the simple number of parties); this measure is the inverse of Rae’s formula. Thus, we also have estimated our empirical equation considering legislative fractionalization (following Laakso and Taagepera 1979) and electoral disproportionality. However, given the statistical nonsignificance of the estimates related to these variables, we opted to omit references to these variables so as to omit the presentation of their descriptive statistics and their estimated coefficients. Full details are available upon request.

⁶ The details of the weights for the different variables composing the KOF index can be found in http://globalization.kof.ethz.ch/media/filer_public/2015/03/04/variables_2015.pdf.

⁷ The unemployment rate and real GDP per capita were also included in previous regressions, but these variables were not significant.

⁸ Named CG in the original data set, Penn World Table Version 7.1, Centre for International Comparisons of Production, Income and Prices at the University of Pennsylvania, https://pwt.sas.upenn.edu Accessed 02/09/2013.

⁹ We used the Durbin–Hu–Hausman test for Equation 2 to analyse the risk of endogeneity. Our results led us to accept the exogeneity of the explanatory variables listed in Table 3 using the unemployment rate and real GDP per capita as instruments. Full details are available upon request.

¹⁰ Given possible reverse causality between voter turnout and our dependent variables, we conducted tests of Granger causality for panel data following Holtz-Eakin, Newey, and Rosen (1988) and Nair-Reichert and Weinhold (2001). Following these procedures, we tested the hypothesis that voter turnout does not cause (i.e. anticipate) the dependent variables assumed in our empirical equation. All of the respective p-values are <10%, allowing us to reject the hypothesis that voter turnout does not cause the dependent variables. Conversely, we obtained values allowing us to accept the hypothesis that the dependent variables do not anticipate (i.e. cause) the voter turnout. This combined evidence (voter turnout anticipating the dependent variables and the dependent variables not anticipating voter turnout) leads us to conclude that the hypothesis of reverse causality between voter turnout and the dependent variables can be rejected.