Research Note—Optimal Mechanism for Selling a Set of Commonly Ranked Objects

Juan Feng
DOI: https://doi.org/10.1287/mksc.1070.0290
2008-01-01
Abstract:This paper designs an optimal mechanism for selling a set of commonly ranked objects. Although buyers rank these objects in the same order, the rates at which their valuations change for a less-preferred object might be different. Four stylized cases are identified according to this difference: parallel, convergent, divergent, and convergent-then-divergent. In general, the optimal mechanism cannot be interpreted as a conventional second-price auction. A reserve price is imposed for each object. Depending on which of the four stylized cases is considered, a higher-value bidder may be allocated a higher-ranked or lower-ranked object. There is also a positive probability that a higher-ranked object is not allocated while a lower-ranked one is allocated. In a departure from the extant mechanism-design literature, the individual-rationality constraint for a mid-range type of bidder can be binding.
What problem does this paper attempt to address?