Partisan ideology, veto players and their combined effect on productive and protective welfare policy: how the left still matters after all?

ABSTRACT

Quantitative analysis of the question whether ‘parties still matter’ has largely focused on the dynamics of aggregated expenditure-based dependent variables or protective welfare policies such as unemployment, sickness and family benefits. This article develops a series of pooled time-series cross-section regression specifications predicting changes in disaggregated protective welfare policies alongside productive welfare policies, namely family services, active labour market programmes and public education, across 17 Western democracies (1971–2010). In so doing, it employs the latest Comparative Manifesto Project [Volkens, A., P. Lehmann, N. Merz, S. Regel, and A. Werner. 2014. The Manifesto Data Collection. Manifesto Project (MRG/CMP/MARPOR). Version 2014b. Berlin: Wissenschaftszentrum Berlin für Sozialforschung (WZB). Accessed January 2015. https://manifestoproject.wzb.eu/information/information] and Veto Player [Jahn, D., T. Behm, N. Düpont, and C. Oberst. 2012. Parties, Institutions & Preferences: Veto Player (Annual), Version 2012–02. Accessed January 2015. http://comparativepolitics.uni-greifswald.de/data.html] data to explore the effect of the ideological position of prime ministers’ parties, veto players and their combined effect on these welfare policy areas. The article confirms that Left and Right governments ceased to make any substantive difference for protective welfare policies from the early 1980s onwards. Yet, it also finds that positive Left partisan effects have largely persisted for productive welfare policies. In the era of global competition, Left party ideology has continued to be an important factor in realising a social investment perspective in practice; in terms of the expansion of family services, its effects have been contingent on the veto power Left prime ministers faced.

KEYWORDS:

• Partisan difference
• veto players
• productive and protective welfare policies
• social investment
• global competition
• pooled TSCS analysis

Acknowledgement

I am indebted to John Hudson, Nick Ellison and Antonios Roumpakis as well as the anonymous reviewers of this journal for their invaluable feedback on how to improve earlier drafts of this article. Special thanks are also due to the attendants of the White Rose Doctoral Training Centre Workshop, Operationalising Historical Institutionalism: How to Capture Institutional Continuity and Change (University of York, May 15th 2015) and the Comparative Research Symposium, Methodology of Institutional Continuity and Change, at the 2015 Social Policy Association Annual Conference (Belfast Metropolitan College, July 6th 2015) for offering helpful questions and suggestions.

Notes

1 Focussing on job placement services, training and job creation as opposed to a social disciplining workfarist approach (see e.g. Dingeldey 2007).

2 Rather than simply adding an arbitrary ‘break-point’ (e.g. 1980), this article follows Allan and Scruggs (2004, 505) coding for individual countries according to the last negative (or the lowest) economic growth year in the first half of the 1980s (all years before that event were coded ‘0’): 1981: Austria, Belgium, Denmark, Finland, Sweden and United Kingdom; 1982: Canada, Germany, Italy, Netherlands, Norway and United States; 1983: Australia, France, Ireland and Japan, 1986: New Zealand.

3 A number of studies argued that moderate Christian democratic parties have historically been associated with welfare state expansion (Van Kersbergen 1995). A relatively recent argument in the literature suggests that the distinction between socialism on the Left and liberalism/conservatism on the Right necessitates spatial analysis (Jahn 2010). This, however, is beyond what this article can deliver.

4 The post-recession period was also split into two separate models (1981–1996; 1997–2010) but this did not alter the findings significantly.

5 In the context of relatively small sample sizes, t-statistics above an absolute value of one (‘1’) point towards ‘statistically relevant’ coefficients. This is because such point estimates would normally surpass common significance thresholds if larger sample sizes were available.

Additional information

Notes on contributors

Stefan Kühner

Stefan Kühner is Assistant Professor in the Department of Sociology and Social Policy at Lingnan University. His main research interest centres on comparative and international social policy, with particular emphasis on the policies and politics of welfare state change. He is co-editor of the book series Research in Comparative and Global Social Policy (Bristol: Policy Press) and Executive Committee Member of the East Asian Social Policy Research Network (www.welfareasia.org).