Public Choice 83: 159-172, 1995. © 1995 Kluwer Academic Publishers. Printed in the Netherlands. A rational-voter explanation of the cost of ruling* MARTIN PALDAM PETER SKOTT Department of Economics, Aarhus University, DK-8000Aarhus C Accepted 4 May 1994 Abstract. It is well known that the average government loses votes - the so-called cost of ruling. We show that the loss can be explained as a perfectly rational demand for change in a median voter model, once the model is amended to let the two parties be visibly different. I. The problem and four old explanations The literature on vote and popularity functions (from now VP-functions) has demonstrated that the average government in a developed democracy loses votes. Consider an election where the incumbent government receives V t percent of the valid vote and let V t_ I be the vote, similarly calculated for the same party(ies) at the last election. The government's gain is AV t - V t_ 1- Empirical work consistently shows that on average AV is negative: as a simple rule of thumb the average government suffers a loss of about 1.65% of the vote. t This erosion of support is often referred to as the cost of ruling, or, using an analogy from the theory of capital, a depreciation of the stock of popularity. These different labels, however, do not explain the empirical observation and, on the face of it, the cost of ruling is a paradox within a rational voter paradigm. The average government should rule exactly as well or as badly as the rational voter expects. Hence, it seems irrational that the voters punish the average government by voting against it. An explanation of this irrationality is needed, and four hypotheses can be found in the literature: (1) The coalition of minorities hypothesis suggested by Mueller (1970). Society is seen as a (large) collection of groups. To rule, a party must put together * We wish to thank Douglas Hibbs, Manfred Holler, Peter Nannestad, Arthur Schram and the referee for helpful comments. We are also grateful to participants at the Valencia Meeting of the European Public Choice Society for comments, that could easily have doubled the length of the paper.