Seminars

We present a novel test to determine sparsity in characteristic-based factor models. Applying the test to industry and pseudo-random portfolios, we reject the null hypothesis that fewer than ten factors are sufficient to explain returns, and show that at least thirty factors are needed for the various subsample periods examined. We find that dense models outperform sparse ones in both pricing and investing. 

We study the ability of different classes of voting rules to induce agents to report their preferences truthfully, if agents want to avoid regret. First, we show that regret-free truth-telling is equivalent to strategy-proofness among tops-only rules. Then, we focus on three important families of (non-tops-only) voting methods: maxmin, scoring, and Condorcet consistent ones. We prove positive and negative results for both neutral and anonymous versions of maxmin and scoring rules. In several instances we provide necessary and sufficient conditions. We also show that Condorcet consistent rules that satisfy a mild monotonicity requirement are not regret-free truth-telling. Successive elimination rules fail to be regret-free truth-telling despite not satisfying the monotonicity condition. Lastly, we provide two characterizations for the case of three alternatives and two agents.

We consider the balanced exchange of bundles of indivisible goods. We are interested in mechanisms that only rely on marginal preferences over individual objects even though agents’ actual preferences compare bundles. Such mechanisms play an important role in two-sided matching but have not received much attention in exchange settings. We show that individually rational and Pareto-efficient marginal mechanisms exist if and only if no agent ever ranks any of her endowed objects lower than in her second indifference class. We call such marginal preferences trichotomous. In proving sufficiency, we define mechanisms, which are constrained versions of serial dictatorship, that achieve both desiderata based only on agents’ marginal preferences.

We then turn to strategy-proofness. An individually rational, efficient and strategy-proof mechanism—marginal or not—exists if and only if each agent’s marginal preference is not only trichotomous, but does not contain a non-endowed object in her second indifference class. We call such marginal preferences strongly trichotomous. For such preferences, our mechanisms reduce to the class of strategy-proof mechanisms introduced in Manjunath and Westkamp (2021). For trichotomous preferences, while our variants of serial dictatorship are not strategy-proof, they are truncation-proof and not obviously manipulable (Troyan and Morrill, 2020).