Disentangling Preferences for Multi-Dimensional Policies: A Conjoint Experiment

A large literature in political science examines citizens’ economic policy preferences, typically in a uni-dimensional framework: citizens can prefer left-wing over right-wing policies, more or less social spending, or more or less generous pensions. However, in reality social policies are multi-dimensional and it has been long acknowledged that their distributive consequences depend on the specific configuration of policies as much as on the overall level of spending (Esping-Andersen 2006). The ways to expand or retrench a social policy are manifold. Policy-makers may change the funding source, the scope of recipients, or the balance between cash, and in-kind components of policies. Policy-makers also often propose package deals, in which they compensate retrenchment on one dimension by expanding on another (Bonoli 2005). Preferences for the precise configuration of social policies is all the more relevant in contemporary societies characterized by increasing heterogeneity among potential recipients along dimensions such as ethnicity, family structure, or work status in dualized labor markets. In spite of the importance of the specific features of social policies, previous research on citizens’ social-policy preferences has largely failed to take into account their multi-dimensionality. Consequently, we know little about which dimensions receive more support in the population, and among whom.

This paper focuses in one relevant social policy, unemployment programs, and asks two questions: Which elements of policy change gain wide acceptance in the population? Are there systematic differences in the dimensions that are relevant for different kinds of publics? To address these questions, we conduct a conjoint experiment asking respondents to choose between sets of alternative reforms of unemployment benefits. The study is run in Spain, a country where unemployment programs are very relevant because of the high unemployment rate, and which is characterized by intense labor market dualization and substantial ethnic heterogeneity among the recipients of benefits. We present respondents with reforms that vary in who they target (long vs short-term unemployed, nationals, the poor, or all citizens), in the provision of training, their economic cost, and how they are funded.

Our results first show that respondents care mostly about the distributive effects of policies: programs that increase protection for the most needy are preferred to those that target other publics. The source of funding is also a relevant driver of choices. Regarding the question of which individual-level characteristics affect preferences, we find that income, labor market, position and risk situation have a surprisingly small influence on the importance given to characteristics of the unemployment insurance. This finding runs counter to political economy theories based on economic self-interest. By contrast, ideology and economic values turn out to affect the policy priorities of respondents. This evidence provides strong support for the growing literature that finds that ideology dominates interests when predicting attitudes towards redistributive and insurance programs.