Pareto efficient combinatorial auctions: dichotomous preferences without quasilinearity (Q1995288)

It is known that the Vickrey-Clarke-Groves (VCG) mechanism [\textit{W. Vickrey}, ``Counterspeculation, auctions, and competitive sealed tenders'', J. Finance 16, No. 1, 8--37 (1961; \url{doi:10.1111/j.1540-6261.1961.tb02789.x}); \textit{E. H. Clarke}, ``Multipart pricing of public goods'', Public Choice 11, No. 1, 17--33 (1971; \url{doi:10.1007/BF01726210}); \textit{T. Groves}, Econometrica 41, 617--631 (1973; Zbl 0311.90002)] satisfies two fundamental desiderata: it is dominant strategy incentive compatible (DSIC) and Pareto efficient. The authors' research is a contribution to the existing main research question: In combinatorial auction models, if agents have classical preferences, is it possible to construct a ``desirable'' mechanism -- a mechanism which inherits the DSIC, Pareto efficiency, IR, and no subsidy properties of the VCG mechanism? In this article a combinatorial auction model where preferences of agents over bundles of objects and payments need not be quasilinear is considered. However, the authors restrict the preferences of agents to be dichotomous. It is shown that if the domain of preferences contains all dichotomous classical preferences, there is no desirable mechanism. However, a natural generalization of the VCG mechanism to classical preferences, which we call the generalized VCG (GVCG) mechanism, is desirable if the domain contains only positive income effect dichotomous preferences.