In social science, we strive to replace names by variables. Contrary to e.g. historians, our primary interest is not to analyze what role Silvio Berlusconi plays in the disheartening erosion of democracy in Italy (… yes, I am on vacation in this terrible and wonderful country), but rather to investigate to what extent a particular configuration of media-ownership and –consumption, in combination with a weak state contribute to undermine the rule and legitimacy of law.
This effort of dealing with abstract concepts, however, entails a range of risks: most importantly, concepts may over time acquire a “life” of their own, meaning that the concept – and its operationalization – remains stable, while the world keeps turning. Thereby, the concept may eventually measure something new, while the empirical results are still interpreted with reference to the original meaning. A prominent example is the use of the variable “left power” (measuring vote shares or seat shares in parliaments and cabinets) in comparative political science research. In myriads of regression models, political scientists test, re-test and control for the effect of “left power” on almost everything, from welfare policies to state spending to political stability, corruption, macro-economic policies, etc. etc. What is more, the declining explanatory capacity of “left power” is one of the main insights of much comparative political-economic research.
But what do we measure, when we add the variable “left power” to our models? What does it mean when the “left power” coefficient in a regression explaining generous and interventionist government policies declines over time?
Two interpretations of this result have been given: on the one hand, analysts argue that the left and the right are converging in their policy positions. On the other hand, it is taken as evidence that the left has become unresponsive to its electorate. Both interpretations, however, may be premature, because they rely on untested assumptions as to who votes for left parties and what policies those parties – consequently- are supposed to advocate. Basically, they both assume that left-wing parties are “working-class” parties, advocating ever more expansive and big social and economic policies.
These assumptions have probably never been really accurate (see Bartolini’s essential work, which shows that left parties hardly ever mobilized more than half of their electorate among the working class), but they are certainly outdated today. In Switzerland, high-skilled service sector workers were already by the late 1990s more than twice as likely to vote for the Social Democrats than blue-collar workers. In France and Germany, both groups of voters are equally likely to support the social-democratic left. Moreover, these trends have amplified, with more and more middle class voters being drawn towards the “new left” (Green parties etc.) and workers drifting towards the (new) right*. So, why should we expect “left parties” to advocate the same policies they demanded in the 1960s and 1970s? Why should they ask for more and more redistribution, if the main beneficiaries of redistribution have been abandoning the left for almost two decades now?
Let me be clear: I am not arguing here that parties are still equally responsive to their electorate as they were in the past. Maybe their hands are indeed tied by external constraints and ideological shifts, and maybe they have really lost sight of the needs and demands of their voters. But so far, we just don’t know. The policy-responsiveness of political parties can only be tested if we first update our assumptions on their electoral constituencies and the preferences of these voters. This is what we attempt to do in a research project that has started in spring 2010 at the University of Zurich, and which will assess the chain of responsiveness from elections to policy outputs in seven European countries. With this project, we hope not least to fill our “old concepts” with new meaning.
*These results rely on ISSP Role of Government 1996 and 2006 data, as well as on analyses in a recent paper I have written.