Complexity Beyond Incentives: The Critical Role of Reporting Language
创建于 更新于
摘要
This paper shows that the complexity of multi-attribute preferences and the language used to report them critically shape the performance of strategy-proof assignment mechanisms [page::21]. The authors find that lexicographic (LEX) reporting yields the highest accuracy across treatments, while additive (SEP) and non-separable (COMP) structures reduce reporting accuracy with no robust difference between SEP and COMP [page::21][page::22]. Under direct serial dictatorship (SD-DIRECT), misreporting rises compared with a non-allocative accuracy benchmark (ACCURACY), indicating incentive- and framing-driven reporting frictions beyond mere list length [page::25]. Attribute-based simplifications (SD-LEX, SD-WEIGHT) do not improve accuracy relative to full ranking, due to substantial representational losses that offset any burden reduction; however, sequential choice (SD-CHOICE) achieves the highest accuracy and the lowest efficiency loss and justified envy, with part of its advantage attributable to menu reduction and part to strategy-proofness and simpler decision framing [page::27][page::28][page::30][page::32][page::34]. The findings imply that reducing the length of a report alone is not sufficient and that menu-pruning or staged/dynamic designs can meaningfully improve allocation outcomes [page::38].
