The past decade has seen the growth of a considerable literature on the link between government popularity, as reflected by the proportion of the public indicating their intention to vote for the government in opinion polls, and the state of the economy, as represented by certain key variables. The work began in the early 1970s with articles by Goodhart and Bhansali, Mueller, and Kramer. It continued through the decade; some of the more recent contributions can be found in a set of readings edited by Hibbs and Fassbender. However, despite the amount and quality of this work, problems remain. Principal amongst these, as Chrystal and Alt have pointed out, is the inability to estimate a relationship which exhibits any degree of stability either over time or between researchers. Nearly all the studies have been successful in finding a significant relationship for specific time periods, but when these are extended, or when the function is used to forecast outside the original estimation period, the relationship appears to break down.
* School of Humanities and Social Sciences, University of Bath. Thanks are due to an anonymous referee for several valuable suggestions.