Brett Williams
Publications
“Varieties of Risk Preference Elicitation” w/ Daniel Friedman, Sameh Habib, and Duncan James (Games and Economic Behavior) {Link}
Abstract: We explore risk preference elicitation via direct choice over lotteries. Our choice tasks differ incrementally, e.g., from choosing between two lotteries to selecting a portfolio from a continuous set of bundled Arrow securities, and from text to spatial presentation. Each subject completes multiple instances of five different tasks, and responses for each task are summarized in parametric (CRRA) and non-parametric (normalized risk premium) measures of risk preference. Variation in task attributes explains much of the observed wide variation in elicited preferences and in correlations across task pairs.
​
​
“Violations of First Order Stochastic Dominance” (Journal of the Economic Science Association) {Link}
Abstract: Leveraging the intricate design of Friedman et al. (2021), I explore inconsistent behavior in lottery choices across several elicitation trials, all through the lens of first order stochastic dominance. I find necessary and sufficient conditions for first order stochastic dominance (FOSD) violations for choices from a budget line of Arrow securities. Applying this characterization to existing data, I compare FOSD violation rates across a broad set of risk preference elicitation tasks.
​
Working Papers
“Zero Intelligence in an Edgeworth Box” (R&R Journal of Behavioral and Experimental Finance)
Abstract: This paper expands on the continuous double auction trader behavior literature, providing a general-equilibrium-suited version of Gode and Sunder’s (1993) “zero intelligence” (ZI) model. While the original authors of ZI (Gode, Spear, and Sunder, 2004) have proposed a venture into trader behavior in the Edgeworth box, the model proposed here generalizes to a much further extent and provides a lower level of “zero” intelligence. Agents participate as two-way traders with utility functions and allocations giving them a natural market-side of preference. Orders are chosen randomly upon entry, with a lattice of bundles of the market’s two goods providing the potential landing spots (upon potential future full acceptance of the order). I test the major assumptions of the model as well as many of the rules implicit in the CDA environment via simulation, providing a GE investigation analogous to the PE literature analyzing the original ZI model.
“Duality and Yquilibrium” with Jacob K. Goeree and Jason Tayawa (Under Review)
Abstract: We present a novel duality in vector optimization that provides an alternative characterization of Walrasian equilibrium. We demonstrate that equilibrium existence and the welfare theorems are a direct consequence of duality. By scalarizing the vector optimization problems, we further demonstrate that Walrasian equilibria are the maximizers, and roots, of a single function of allocation and prices – the economy’s potential. We illustrate the usefulness of the potential for computing equilibria.
​
"Paretian Equilibrium" with Jacob K. Goeree
Abstract: We introduce a simple quantity dynamic in which consumers are buyers (sellers) of a good if their marginal utility for that good is above (below) average. We demonstrate this dynamic converges to a Pareto optimal allocation in any exchange economy irrespective of the prices at which trade occurs. Together the Pareto optimal allocation and trade prices define a Paretian equilibrium. Walrasian equilibria, which subsume “correct” prices, are degenerate Paretian equilibria without any price dispersion. Data from a laboratory exchange economy exhibit the price variability predictedby Paretian equilibrium and refute the Walrasian hypothesis.
​
“LEAPE: Learning More-General Paretian Equilibrium” with Jacob K. Goeree
Abstract: We introduce LEAPE, a parameter-free model of trader behavior in multiple inter-linked markets. Transaction prices inform traders about the likelihood certain prices are acceptable. Traders submit limit orders accordingly or accept others’ limit orders. We demonstrate this simple learning model reproduces data patterns from prior general equilibrium experiments, whether observed prices converged, kept cycling, or led to odd outcomes. In each simulated replication, LEAPE predicts the more-general Paretian equilibrium outcomes recently proposed by Goeree and Williams (2023).
Conferences/Seminars: 2023 ANZWEE, 2023 European ESA Meeting, Sydney Experimental BrownBag (SEBB 2024)
​
​
“Opening the Book: Price Information’s Impact on Market Efficiency in the Lab” {PDF}
Abstract: Human trader behavior and market convergence are studied in a general equilibrium two good setting through the use of the continuous double auction. The set of active bids and asks shown to traders in a session is either (1) only the best bid and ask in the market or (2) the full set of active orders; similarly, the set of visible transactions spans the full history of the trading period to only a trader’s own transactions. Laboratory markets reveal no improvement in allocative efficiency and price convergence when improving price/order accessibility symmetrically across the set of bids and asks, as well as the transaction history. Improvements in accessibility in only one factor leads to reductions in efficiency. Effects on price discovery related outcomes appear to be asymmetric across transaction history and orderbook improvements, with the order of increased visibility impacting market outcomes. Results are compared against standard simulated benchmarks.
Conferences: 2022 ANZWEE, 2021 ESA Global, 2021 Second ZIMI, 2021 ESA NA
Awards: Best Student Paper Award at The Second Conference on Zero/Minimal Intelligence Agents (2021) (co-awarded)
​
“Minimal Intelligence in the Double Auction: Logit-Choice and Reservation Utility” {PDF}
Abstract: I propose a new boundedly-rational trader-behavior model for the continuous double auction, fit in the general equilibrium framework. Traders’ order placement and acceptance strategies are driven by GE-adaptions of two classic mechanisms: within-period reservation prices (here utilities) (Friedman (1991)), and price-acceptability beliefs (Gjerstad and Dickhaut (1998)). Reservation utilities are considered from both a buying and selling standpoint. Reservation utility associated with the side of entry is used to gauge immediate acceptability of contra-side orders while contra-side reservation utility informs the trader of admissible orders for placement. Traders place logit choice probabilities over their set of admissible orders, based on beliefs formed about
the acceptability of each order. The logit choice parameter allows the precision of trader order placement to range from uniformly random to perfectly maximizing given beliefs (mapping the range of trader behavior from Gode and Sunder (1993) to Gjerstad and Dickhaut (1998)). Simulations report market performance for this model, finding promising measures of efficiency and evidence of convergence to nearly Pareto optimal allocations.
Conferences: 2020 ESA Global
*Formerly titled "Expanding minimal intelligence in the double auction"
​
“Efficiency in Queuing Under Decentralized Mechanisms” with Kristian López Vargas and Shuchen Zhao {Link}
Abstract: We study, theoretically and in the laboratory, three simple decentralized mechanisms to reallocate positions in a queuing problem. In our environment, players have heterogeneous values for time and arrive in random order to a queue before service starts. While waiting for service, they can switch positions, with different rules depending on the mechanism in place. We mainly focus on three institutions: voluntary swapping upon request, take-it-or-leave-it (TIOLI) monetary offers, and a non-fungible reputation point system (``social token''). Compared to the initial order of the lines, we find modest efficiency gains when swapping and the social token are in place. In contrast, we find a sizable efficiency improvement in queues with monetary transfers. Although TIOLI improves efficiency, this gain comes at the cost of higher inequality. We also study these mechanisms with and without unilateral communication from the switch requester. This form of cheap talk has no impact on the main studied outcomes.
Conferences/Seminars: SEBB 2023, DUFE Seminar 2023, ESAM 2023
​
“Disappointment Aversion Across Risk Elicitation Tasks” with Sameh Habib
Abstract: The choice of functional form and parametrization of an agent’s utility function is crucial when investigating questions around risk preferences experimentally. We adapt the estimation of CRRA preferences from laboratory data generated in Friedman et al. (2021) to estimate a second parameter for disappointment aversion (a la Gul (1991)), making use of a sufficient statistic, L, for prices and probabilities developed in Habib et al. (2017). Evidence is shown, suggesting minimal or no gains in estimation when amending constant relative risk aversion with an extra parameter for disappointment aversion. The two parameter model does provide small improvements in estimation for subject’s with less stable behavior.
​
Works In Progress
“Visual Markets” with Daniel Friedman and Vivian Zheng
Abstract: We introduce a pure point and click user interface for n human traders in a two-good general equilibrium economy. Compared to standard continuous double auction with text entry, we conjecture that our video-game-like Visual Market will converge much faster to a neighborhood of the Pareto set, and will generate final allocations that are at least as efficient.
​​
​
"Designing Markets" with Jacob K. Goeree and Jason Tayawa
​
​