Survey-based Horizons and Government Formation

Abstract

The main challenge to testing the policy horizon hypothesis is that it necessarily involves the measurement of a very elusive entity, the policy horizons of parties. In Chapter 3, two measurement methods were introduced to meet this challenge. Both methods have significant limitations, of which the most important relates to the use of coalition behaviour in the estimation procedure. Contrary to what might be supposed, the risk of circular explanation is not the prime concern. Although false or accidental horizon effects can be generated with these methods, we have seen that they can be readily distinguished from genuine horizon effects. A much more significant concern is that the behaviour-based estimation procedures are devoid of information on the sources of policy horizons. In this chapter, the horizon hypothesis is submitted to a testing process in which horizons are measured with direct reference to the constraints imposed on parties by their supporters, rather than indirectly via the coalitional consequences of those constraints.