Abstract: Facing public projects with consequences of difficult to assess with markets, as for instance in environmental conflicts the stakeholders react according to two types of values: (i) their direct vested interests that can be in certain case compensated in economic transactions and (ii) their moral attitudes which is not prone to any monetary valuation. In the later case, the usual way to get preferences expression in society is voting which ignores the strength of preferences. The economic and political literature has shown since long that the range of strategy-proof, non-manipulable, mechanisms, as well in voting as in mechanisms allowing for transfers is very slim[sg1] . However in the special case of the decision over a public project, we propose here a set of mechanisms and we show that there are no other strategy-proof mechanisms out of this class. This class is much wider than the classical Groves-mechanisms and, contrarily to Green & Laffont who have precluded feasible solutions with non-linear utility and pareto optimality. We establish a general necessary condition for strategy-proofness requiring a different definition of Pareto-optimality. The mechanisms are based on the properties of commutative groups in algebra and can be simply implemented. In particular a pure voting procedure (i.e. a quasi-referendum) belongs to this class.

The interest of this approach is to design systems of penalties in pivot-mechanisms in which the role of money is mitigated and the focus is made on the non-monetary change in welfare, suffered or benefited by the stakeholders. We think that this can open a way to formalize systems of compensations in environmental issues, used in some countries to exhibit and internalize the liability of pivot stakeholders in decisions, without jeopardizing preference revelation.