Résumé
Do income inequalities aect the provision of public goods? Warr established in 1983 a theorem of neutrality: under some conditions, a marginal redistribution of endowments among agents does not aect the amount of public good provided by their voluntary contributions. Generalizations of this result by Bergstrom et al. (1986), helped to better understand this phenomenon: neutral redistributions are those of "low" amplitude, so that agents whose income decreases can maintain their consumption of private goods, and adjustments of individual contributions leave unchanged the aggregate contribution to the public good. Itaya et al. (1997) have focused on the consequences of a non-neutral redistribution on welfare. In the rst two chapters of this thesis we test these predictions in the laboratory using a public good game with quadratic utility functions. The rst chapter considers a redistribution of a "low" amplitude that should not modify the amount of public good supplied. However in Chapter 2, we run a redistribution of a "high" magnitude so that it aects the amount of public good provided and the social welfare. Although some theoretical predictions are found in the laboratory, such as the modication or not of the amount of public good and of the welfare, predictions on individual behaviors and payos are rarely veried. In particular, we note that following a modication of their endowment, some subjects decrease or increase their contribution less than theory predicts and that poor agents over-contribute. It further appears that the emergence of inequalities does not aect behaviors in the same way than when these inequalities preexist and thus that the direction of the redistribution, depending on whether it creates or decreases inequalities, matters. In Chapter 3 we study more precisely a redistribution that creates inequalities in a linear public good game and we test wether men and women respond similarly to the modication of their endowment and what are the consequences on the supply of the public good. We show that when women become rich, the quantity of public good provided decreases. It also appears that behaviors are modied when the rich gender is common knowledge.