Veto Players and Policy Stability: A Regression Discontinuity Approach

• In Event: Class and Inequality & Legislative Politics & International Security Virtual Poster Session
  In Virtual Poster Session: Legislative Politics Virtual Poster Session

Abstract

What determines with what ease current policy can be changed? In veto-player theory, pioneered by Tsebelis, the answer given is that the rate of policy change is inversely related to the ideological dispersion and the number of actors that can exercise veto power over the policy decision at hand. Intuitively, for each additional actor with such power – for example, a party in a coalition government – the probability that all of them prefer an alternative policy position over the status quo will fall, and policy stability will increase.

Since the turn of the millennium, veto-player theory has occupied a central role in comparative politics. As such, there are today numerous quantitative studies testing its predictions. Doing so in a classical regression framework is although challenging due to potential endogeneity problems. Since neither the number nor the ideological dispersion between parties in a ruling coalition is randomly assigned the empirical method would, in general, be unable to isolate the causal effect of veto players from omitted variables such as an unobserved voter preference for policy stability. This paper's main contribution is an attempt to overcome these difficulties by exploiting the quasi-experimental nature of close elections.

More precisely, it adds to the previous literature by using a regression discontinuity strategy to elicit an exogenous variation in the number and dispersion of veto players to test tax policy stability in Swedish municipalities. Specifically, the paper uses the precise number of additional (fewer) votes the largest party would have needed to win (hold) the absolute majority of the seats in the local council as its forcing variable. In doing so, the analysis leverages a distinct difference in the number of veto players and their ideological dispersion between two groups of local governments; those being formed after the largest party narrowly wins and those being formed after the largest party narrowly loses the absolute majority.

To illustrate the empirical strategy, we can consider the example of Sorsele municipality in northern Sweden. In the election of 1994, the Social Democratic Party fell just 102 votes short of winning the overall majority in the 31-seat local council, instead prompting the formation of an ideologically diverse coalition government with five partisan veto players. When four years later they instead actually secured that 16th seat, they did so with a mere two-vote margin.

The results of the paper do not support the predictions from veto-player theory. Rather than lending evidence to the hypothesis that more ideologically diverse governments become gridlocked, I find that larger ideological dispersion among veto players has a weak negative or no effect on tax policy stability. In other words: such local governments push through larger – not smaller – absolute changes to the municipal tax rate.

The empirical design is verified by the results being insensitive to conventional robustness test and to specifications using the number of veto players instead of their ideological dispersion. Additionally, the results are robust when adjusting for oversized coalitions and counting only veto players that form a minimum winning coalition; thus aligning the empirical set-up with the more restrictive hypothesis that veto players are relevant only when they can exercise their power as a member of such a minimum winning coalition.

Author

• Erik Scheller, Uppsala University