Political parties as drivers of post-crisis social spending in liberal welfare states

Abstract

In the aftermath of the global economic crisis, the challenges facing welfare states are unprecedented. While government leaders have been in broad agreement that the severity of the recession called for decisive actions to limit the costs of the crisis, national responses have differed significantly. This article seeks to explain these divergent patterns and answer the critical question: how has the crisis affected the politics of social spending across liberal welfare states? While political conflict over social spending may increase across all countries in the wake of an economic crises, partisanship is expected to have a stronger effect in liberal welfare states, due to weak automatic stabilizer effects and a reliance on discretionary spending. This research tests the effects of political parties on social spending across nine liberal welfare states (Australia, Canada, Ireland, Japan, New Zealand, South Korea, Switzerland, the UK, and the USA) during the pre-crisis (1990–2007) and post-crisis (2008–2013) periods. It also provides in-depth analysis of the USA and the UK, two representative liberal welfare states who adopted highly dissimilar post-crisis social spending. The findings demonstrate that while political parties were not correlated with social spending during the pre-crisis period, after the global economic crisis they were significant in influencing social spending levels. This indicates an important shift in political dynamics across liberal welfare states over time that has not been fully accounted for by the existing literature.

Appendix

The estimating equation for the quantitative model is:

$$Si,t = (1 + \beta^{\text{CR}} *{\text{CR}} + \beta^{\text{C}} *C + \beta^{\text{PR}} *{\text{PR}} + + \beta^{M} * \, M + \beta^{\text{EU}} *{\text{EU}})*(\varSigma \delta t*Dt) + \varSigma y^{k} X^{k} i,t + \lambda + \alpha i \, + \varepsilon i,t$$

Si, t = S refers to government spending in some policy area, i indexes countries, t time period, k = k a set of control variables (Xi,t), Dt = Annual time dummy variable used to signify common economic shocks, CR = conservative party, C = centre party, PR = proportional representation, M = mixed electoral system, EU = EU membership, β = the key parameters are the betas because they capture the extent to which political-institutional differences mediate the effects of common unobserved shocks on spending. For example, if there are no institutional effects, then $\beta^{\text{PR}} = \beta^{\text{CR}} = 0$ and policies are entirely a function of the control variables plus the set of time and country-specific effects. αi = unobserved case specific effects, εi,t = unobserved random error term.

The model in this analysis uses random-effects panel data which permits individual effects to be measured across countries and over time, differentiating the direction and strengths of the effects of political variables before and after economic shocks. Robust standard errors are used to account for heteroskedasticity and autocorrelation. Several factors contribute to the use of a random-effects model. First, the dataset includes matched time-year country units, which do not fit standard OLS analysis. Second, many of the independent variables, such as EU membership and electoral system, are time invariant, eliminating the option of fixed-effects modelling. A Hausman test was run to test the appropriateness of using a random-effects model versus a fixed-effects alternative. Hence, random-effects modelling becomes the optimal choice for this panel data analysis (Stock and Watson 2011) (Table 2).

Table 2 Independent variable descriptions