Special Session Summary Strategic Behaviors in Competitive Games

Wilfred Amaldoss, Purdue University
Teck H. Ho, University of Pennsylvania
[ to cite ]:
Wilfred Amaldoss and Teck H. Ho (2001) ,"Special Session Summary Strategic Behaviors in Competitive Games", in NA - Advances in Consumer Research Volume 28, eds. Mary C. Gilly and Joan Meyers-Levy, Valdosta, GA : Association for Consumer Research, Pages: 126-127.

Advances in Consumer Research Volume 28, 2001     Pages 126-127

SPECIAL SESSION SUMMARY

STRATEGIC BEHAVIORS IN COMPETITIVE GAMES

Wilfred Amaldoss, Purdue University

Teck H. Ho, University of Pennsylvania

 

"NETWORK VERSUS NETWORK: THEORY AND EXPERIMENTAL EVIDENCE"

Wilfred Amaldoss and Amnon Rapoport

Collaborations are becoming increasingly popular. By being part of a network, a firm is likely to gain access to a larger resource base (e.g., Mahoney and Pandian 1992, Kraatz and Zajac 1999). Yet the conventional wisdom is that larger networks are more likely to fail (Gomes-Casseras 1994, p 9). There is a basis for this wisdom: As the number of partners in a network increases, its partners will commit fewer resources for the joint endeavor. In agreement with the conventional wisdom, the large Mips network, in contrast to the smaller PowerPc network, is a failure in the microprocessor business. On the other hand, the Visa network with over 23,000 member banks is thriving in the face of competition from American Express, a fully integrated firm. Similarly, the Star Alliance, a nine-member airline network established in 1997, is growing stronger in the competitive airline business. In light of this contradictory evidence what is the effect of number of partners in a network on fostering collaboration? The answer is not clear. There may be circumstances when the positive effect of network size may even override the free-riding problems, making it optimal for firms to forge larger networks.

It is possible that the type of investments required for the joint effort may moderate the effect of network size. Contrary to intuition, our game-theoretic model predicts that, if the investments are recoverable, the joint investment increases as the size of network increases. This is because the incremental investment from the new partners more than compensates for the total loss in investment due to under investment by current partners. However, if the investments are nonrecoverable, our model predicts that the joint investment does not depend on he network size. This is because the total decrement in investment, on account of the under-investment by current network partners, is precisely offset by the incremental investment made by new members.

We report the results of a series of 12 experiments designed to examine the effect of type of investment and network size on fostering innovation. When the investments were recoverable, our results show that individual network partners invested less as the network size increased. Yet, in keeping with the model prediction, the joint investment of the network increased as the network size increased. When the investments are nonrecoverable, theory predicts that the joint investment should not be affected by network size. In reality, the joint investment increased as the network size increased. This is because network partners made less than predicted reductions in investments, as the network size increased.

On the average, our subjects invested more that the equilibrium predictions. What could have caused such a behavioral deviation from the theoretical norm? We examine the implications of three competing explanations for the deviation: risk attitude, warm glow, and altruism. On generalizing our model to accommodate different risk attitudes in addition to risk neutrality, we observe that risk-seeking behavior on the part of all the members increases joint investment, while risk-aversion has the opposite effect. Allowing for warm glow preferences increase the mean joint investment. However, it doesn’t cause the joint investment to vary with network size, if the investments are nonrecoverable. On the other hand, altruism causes the joint investment to vary with network size, besides increasing the mean joint investment, even if the investments are nonrecoverable.

There is evidence of altruism in the investment decisions of our subjects. This suggests that a supply-side externality is at play in larger networksBpartners in larger networks may well invest more in their joint endeavor merely out of altruistic regard for their partners. The altruistic regard increases as the network size increases, and, in part, helps to reduce the desire to free ride on the efforts of partners.

Later with the aid of the Experience Weighted Attraction (EWA) Learning model (Camerer and Ho 1999), we attempt to discern the adaptive learning mechanism that can account for the investment behavior of our subjects. Our analysis suggests that the investment decisions were not exclusively guided by forming beliefs about other players. Rather, our subject’s strategy choices were guided by reinforcement-based learning, either completely or predominantly.

 

"ARE POSTED PRICES 'FAIR’? AN EXPERIMENTAL ANALYSIS OF DYNAMIC BUYER-SELLER INTERACTIONS"

Darryl Banks, J.Wesley Hutchinson, and Robert J. Meyer

This paper examines "fairness effects" in posted-price markets. Two explanations for the phenomena have been offered. By way of an experimental investigation we seek evidence of the empirical validity of these explanations. The persistent observation of "fairness effects" in posted-price markets is strong evidence that the classic economic model, in which buyers are price-takers and sellers choose prices to maximize profits against the distribution of reservation values, given that buyers are price-takers, is incomplete.

The dominant explanation for these effects posits that the perceived fairness of a prospective transaction affects the buyer’s assessment of the transaction’s utility, and that sellers take this into account when setting prices. Fairness is related to the division of gains from trade, a 50/50 split being the fairest of them all. Consequently, the perceived fairness of a posted price is a function of the perception of the seller’s cost, as, for any given price, the lower is the seller’s cost the smaller is the buyer’s share of the gains. This explanation is consistent with a great deal of existing evidence that shows that buyers are more reistant to high prices when they believe sellers’ costs are low, and sellers try to justify high prices with claims of high costs.

In a recent paper Banks et al propose a different explanation for "fairness effects," arguing that they are "rational" behaviors of forward-looking buyers and sellers. They argue that the phenomena are due to the parties’ intertemporal incentives and to the existence of private information on both sidesBonly buyers know their reservation values and only sellers know their costs. A buyer’s willingness to reject prices that are no higher than his/her reservation value is a strategy designed to make the seller post low prices now and/or in the future. This strategy can be effective only if the seller is uncertain about the buyer’s reservation value. The seller anticipates this and posts prices that, if accepted, reveal the truth about the buyer.

Banks et al’s results are consistent with the existing empirical evidence but yield a striking prediction about the effects of the parties’ initial beliefs. They predict that a buyer’s initial beliefs about a seller’s cost will not affect his/her responses to the seller’s prices but will affect the prices that the seller posts. Moreover, while the buyer’s own initial beliefs will not affect his/her responses to the seller’s prices, the seller’s initial beliefs will affect the buyer’s responses to the posted prices. This prediction differs sharply with what we would expect to observe if perceived fairness is an argument in the utility function, providing a basis for a test of the two theories. We find evidence supporting the Banks et al result.

 

"AN EXPERIMENTAL STUDY OF SEVERAL ELECTRONIC MARKET INSTITUTIONS"

Teck H. Ho and Steve Hoch

The internet has opened up a whole host of opportunities for sellers and buyers to experiment with different market institutions. Some experts believe, for example, that the future of business-to-business and business-to-customer e-commerce will be dominated by auctions (InfoWorld, 1998) and the stock market success of several new market makers (e.g., ebay.com and priceline.com) demonstrates investors’ beliefs that these innovative market institutions will significantly increase in revenue in the future. Thus the study of different internet market institutions has tremendous practical relevance.

It is also intellectually challenging to study electronic market institutions. Most market institutions outside of the well-developed financial markets and a limited number of auction markets rely almost exclusively on posted prices by a limited number of sellers. With significant reductions in e-search and transactions, the internet already has spurred the development of numerous new market institutions where not only sellers post prices (amazon.com) but so do buyers, both statically (priceline.com) and interactively (ebay.com). Enthusiasts of e-commerce argue that these new market mechanisms can dramatically increase efficiency by appropriately matching buyers’ willingness-to-pay with sellers’ willingness-to-sell.

There are 3 critical dimensions along which one can evaluate a market institution:

$Which market institution has the highest average price of transaction?

$Which market institution generates the most social surplus?

$Which market institution has the most equitable way of dividing the social surplus between the buyers and sellers?

We reported experimental results from two commonly used electronic market institutions and evaluate them along te 3 dimensions mentioned above.

Dipankar Chakravarti was the discussant for this special session.

----------------------------------------