Forecasting and the outcome of the 2001 British general election

Impressive progress has been made in forecasting election outcomes in a variety of institutional settings. In this issue of the journal, Lewis-Beck, Nadeau and Belanger suggest a plausible and parsimonious model to forecast the outcome of the 2001 British general election. This model has the advantage of generalizability, but it has the disadvantage of producing increasingly inaccurate predictions of the governing party’s vote share over the course of the 1990s. This comment speculates on why this might be the case and on how the model might be re-specified to overcome this problem. It focuses on the macroeconomic measures it employs, on its definition of the governing party’s vote share and on the possible distorting effects of turnout decline.