Bart J. Wilson

Bertrand-Edgeworth Competition, Demand Uncertainty, and Asymmetric Outcomes

Abstract We analyze investment and pricing incentives in a symmetric Bertrand– Edgeworth framework with uncertain demand. Firms choose production capacities before observing demand. Prices are chosen after demand is observed. If the extent of demand variation exceeds a threshold level then a symmetric equilibrium in pure strategies for capacities does not exist. A smaller firm has no incentive (ex ante) to expand its capacity because capacity expansion would reduce its expected revenue in the event that demand is lower than expected. Output prices are predicted to have positive variance when demand is low and zero variance when demand is high. Journal of Economic Literature Classification Numbers: D43, L13.

Firm-specific cost savings and market power

Summary. We report a policy experiment that illustrates a potential problem of using historical pass-through rates as a means of predicting the competitive consequences of projected firm-specific cost savings in antitrust contexts, particularly in merger analysis. The effects of cost savings on welfare can vary vastly, depending on how the savings affect the industry supply schedule. In a capacity-constrained price-setting oligopoly, we observe that cost savings can overwhelm behaviorally salient market power incentives when the savings affect marginal (high cost) units. However, cost savings of the same magnitude on an infra-marginal unit leave market power unchanged.

Interactions of automated pricing algorithms: an experimental investigation

ABSTRACT The advent of electronic commerce enables retailers to set prices via automated algorithms. This paper employs the method of experimental economics to examine human behavior in environments where multiple automated pricing algorithms are available for use. The results of the experiments reported here indicate that individuals prefer using automated pricing algorithms to setting prices manually. However, neither a price-beating nor a price-matching automated rule is dominant in terms of the frequency of usage. We find that the composite distribution of prices is a mean- and median-preserving spread for the game-theoretic distribution and for the previously reported distribution of manually set prices. Our primary finding is that individuals encounter difficulty in coordinating to achieve tacitly collusive outcomes. However, as indicated by one market, multiple algorithms do not necessarily preclude tacit collusion. Keywords Automated pricing, Internet posted offer markets, experimental economics.

Structural features that contribute to market power in electric power networks: some preliminary results

The authors report an experiment that examines a primary concern of policy makers: how a structural feature of electric power networks can contribute to the exercise of market power by well-positioned players in deregulated markets. One such feature is the distribution of ownership of a given set of generating assets. For example, two large firms could be allocated baseload and intermediate generators such that either firm would be willing to withhold unilaterally the capacity of its intermediate generators from the market, to benefit from the supra-competitive prices which would result from only selling its baseload units. Conversely, ownership of some of the intermediate generators from each of these firms could be transferred to two other firms, so that no one firm can unilaterally restrict output to spawn supra-competitive prices.

Mixed Strategy Nash Equilibrium Predictions as a Means of Organizing Behavior in Posted-Offer Market Experiments

Publisher Summary Market power arises in many posted-offer markets and drives a distinction between the competitive prediction and the Nash equilibrium for the market viewed as a stage game. Pricing patterns in such markets tend to be characterized by “Edgeworth cycles” that deteriorate as the sessions progress. The amplitude and frequency of the cycles are sensitive to design and procedural details, and vary considerably from experiment to experiment. Although persistent serial correlation in pricing proscribe any direct test of static Nash mixing predictions, rough correspondence between the central moments of predicted and observed densities has been observed in a variety of different instances. However, the persistent and very prominent deviations observed in an asymmetric design suggests that Nash mixing predictions do not uniformly organize behavior well. The circumstances under which mixing predictions may organize outcomes well merits further investigation.

Design comparisons for procurement systems

We use experimental techniques to compare first-price auctions to a common but previously unexamined exchange process that we term multilateral negotiations. Initially, we find that transaction prices are statistically indistinguishable in the two institutions with four sellers, but that prices are higher in multilateral negotiations than in first-price auctions with two sellers. Surprisingly, we find in two-seller environments that a history of multilateral negotiations leads to higher auction prices, which suggests that buyers may see little price effect by moving from negotiations to auctions.