| Jean-Francois Laslier, Laboratoire d'Econometrie de l'Ecole Polytechnique Karine van der Straeten, Universite de Cergy-Pontoise |
| Voting Under Ignorance of Unemployed Skills: The Bias Towards Overtaxation |
| Session: C-12-15 Wednesday 16 August 2000 by Straeten, Karine van der |
| Usual models on voting over basic income - flat tax schedules rest on the assumption that voters take decisions under complete information, in a static framework. In particular, voters know the whole distribution of skills even if at equilibrium some individuals choose not to work. If individuals' productivity remains unknown until they have chosen to work, it may be more convincing to assume that voters only have beliefs about the distribution of skills and that a dynamic game takes place. At each period, individuals vote according to their beliefs, beliefs that they update when getting new information from the job market. We show that this dynamic process converges towards some equilibrium which depends on the true distribution, but also on initial beliefs. It is sometimes said that theoretical voting models have little empirical support since some countries with very similar distributions of wages choose quite different tax schedules. If these countries differ with respect to their initial beliefs about the distribution of skills among unemployed individuals, our model predicts that the tax schedules eventually chosen may also be quite different.\qquad \qquad\ Second it is shown that, under this kind of incomplete information, the tax rate and unemployment rate at equilibrium are always higher than (or equal to) the rates which would be chosen if all the parameters of the economy were known. The economic intuition supporting this result is quite simple. If we assume that individuals' productivity is discovered by observing the labor market, highest productivities are more easily discovered. So, at equilibrium, there is no way to overvalue the number of highly productive individuals whereas there is some chance of overestimating the thickness of the lower tail of the \ productivity distribution. Microsimulations on French data are provided, illustrating that this bias towards overtaxation and overunemployment introduced by incomplete information may be of significant size in real economies. |