Party mandate theory and time-series analysis: a methodological comment

In 1990 Ian Budge and Richard Hofferbert published an article in support of “the doctrine of the party mandate”, using evidence from regression analyses relating the content of postwar US party platforms and governmental outputs in terms of yearly expenditure rates. Their approach was severely criticized by Gary King and Michael Laver (1993) but has been maintained by the authors in a subsequent extension of their analysis to include data from Australia, Canada and seven European States. The present article takes issue with both the approach followed by Budge and Hofferbert and the alternative approach recommended by King and Laver. It is argued that the trend problem has not been adequately dealt with and the formalization of the mandate model lacks conceptual consistency. Three major suggestions emerge from the discussion: (1) the formal mandate model should be extended to include a “divergence term” designed to separate positive from negative mandate effects; (2) the analysis should pay closer attention to the parameter restrictions that follow from the theoretical model; and (3) the regression equations should be interpreted heuristically in terms of “cointegration” or “causal-trends” models.