Reference

Peter Bossaerts (University of Melbourne), Asset Pricing under Computational Complexity, Fédération des Banques Françaises Seminar, TSE, September 24, 2018, 12:30–14:00, room MF 323.

Abstract

We study asset pricing in a setting where correct valuation of securities re- quires market participants to solve instances of the 0-1 knapsack problem, a computationally \hard" problem. We identify a tension between the need for absence of arbitrage at the market level and individuals' tendency to use me- thodical approaches to solve the knapsack problem that are effective but easily violate the sure-thing principle. We formulate hypotheses that resolve this ten- sion. We demonstrate experimentally that: (i) Market prices only reveal inferior solutions; (ii) Valuations of the average trader are better than those reflected in market prices; (iii) Those with better valuations earn more through trading; (iv) Price quality is unaffected by the size of the search space, but decreases with instance complexity as measured in the theory of computation; (v) Participants learn from market prices, in interaction with trading volume.

Research partnership

Fédération des Banques Françaises Research Initiative (sustainable)