Special Session Summary Internet Market Institutions

Rami Zwick, The Hong Kong University of Science and Technology
[ to cite ]:
Rami Zwick (2002) ,"Special Session Summary Internet Market Institutions", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 487-488.

Advances in Consumer Research Volume 29, 2002     Pages 487-488

SPECIAL SESSION SUMMARY

INTERNET MARKET INSTITUTIONS

Rami Zwick, The Hong Kong University of Science and Technology

SESSION OVERVIEW

A causal observation indicates that the diversity in shopping on the Internet lies not in the difference in selection but in the multiplicity of methods of purchasing and in particular in multitude of pricing mechanisms. Many of these methods are not prevalent in the off-line environment at the consumers level. It’s not a matter of what you want to buy, but how you want to buy it. One can, for example, buy at auction through eBay, name your price at Priceline, haggle at HaggleZone, tell the 'market’ what you want and wait for its best offers at Imandi, group buy at MobShop, search for the lowest price at MySimon and check prices with RUSure, wait for the price to drop at OutletZoo, or learn if you received the best deal in time to cancel or return a purchase and buy from a store offering a better deal at Nosaleisfinal.com.

These different models offer ways to buy that are outside the regular, offline experience of the typical shopper.

The early enthusiasm for these novel dynamic pricing mechanisms is echoed in a PricewaterhouseCoopers’ 'E-Business Technology Forecast’ (1999) report predicting that the use of dynamic pricing will grow and that special-purpose technology available today to implement dynamic pricing will be integrated into enterprise and specialty Web sites.

Why do we observe a multitude of non-traditional pricing mechanisms on the net? The typical popular answer is that these mechanisms were not practical to implement "on land" at the consumer level because the cost of implementation is higher than the expected saving from efficiencies. However, the online environment permits what might be considered an ultimately stock-market model for daily products. The argument is that these mechanisms flatten industry inefficiencies and move the economy closer to the textbook model of perfect competition, which assumes abundant information, many buyers and sellers, zero transaction costs and no barriers to entry.

However, even theoretically, the gains from efficiencies do not always shared "fairly" among all parties and since it is not always possible for one side to dictate the trading mechanism, it might be expected that not all of these new rules will survive in the market. Secondly, even if theoretically the gains are obvious it is not clear if the mechanisms will survive the bounded rationality status of both consumers and sellers.

The session presented three studies on various Iternet market institutions. The common theme of this session was the growing interest in Internet trading rules on the one hand and the development of a descriptive (experimentally based) variant of game theory that combines psychological and economic elements, often referred to as behavioral game theory. In recent years, more and more social scientists have combined in their work the theoretical framework of traditional game theory to model the environment (including the incentive structure) in which decisions are made, and the experimental methodology to study actual consumer behavior under specific conditions. The combination of rational-choice framework and psychological models seems to fit very well within the context of consumers’ extensive decision making behavior. Consequently, the objective of the session was to demonstrate how this methodological approach could be applied to investigate and explain an interesting observation about Internet trading rules.

 

"AN EXPERIMENTAL STUDY OF SEVERAL INTERNET PRICING MECHANISMS"

Teck H. Ho, University of Pennsylvania

Steve Hoch, University of Pennsylvania

The Internet has opened up a whole host of opportunities for sellers and buyers to experiment with different pricing mechanisms. It has spurred the development of numerous new pricing models where not only sellers post prices (e.g., amazon.com) but also buyers, both statically (e.g., priceline.com) and interactively (ebay.com). Enthusiasts of e-commerce argue that these new pricing mechanisms can dramatically increase efficiency by appropriately matching buyers’ willingness-to-pay with sellers’ willingness-to-sell.

Some industry observers claim that the future of business-to-business and business-to-customer e-commerce will be dominated by auctions while others believe that there is room for several pricing formats to coexist. There is however little empirical evidence to either support or refute these claims. The goal of this study is to evaluate these claims by examining 3 broad classes of pricing mechanisms along several economic performance criteria. These criteria include the total social surplus generated by a pricing model and the way it divides this social surplus between the buyers and sellers.

We conducted a total of 12 experimental markets in which 4 different pricing mechanisms were used to determine the prices of a product. Subjects were motivated by monetary incentives and were paid based on their performance in the markets. Our results show that seller and buyer posted pricing models have a lower total surplus than open English auctions and the buyer posted pricing models have the most equitable way of dividing the total surplus between the buyers and sellers. Interestingly, the ability to make price commitment allows a price-setting party (either buyer or seller) to gain a greater portion of the surplus. Sellers posted pricing models tend to lead to higher prices but lower sales volumes than the other pricing models. Finally, we show that subjects’ behaviors are often inconsistent with theoretical predictions from Industrial Organization.

 

"NAME YOUR OWN PRICE: THEORETICAL AND EXPERIMENTAL INVESTIGATION OF BIDDING FOR A SHOPPING BASKET"

Wilfred Amaldoss, Purdue University

Sanjay Jain, Purdue University

With the advent of the Internet, numerous innovative ways to price products have become feasible. One such model is that of Priceline.com. The company flipped traditional pricing methods on its head by allowing consumers to place bids for airline tickets. In order to purchase an airline ticket a customer could log in to the Priceline website and place a bid. If some airline is willing to sell the ticket at or below the bid price then the transaction goes through, otherwise the bid is rejected.

This pricing model gained substantial popularity among consumers as evidenced in the rapid growth of the company. In fact, many compeitors including Microsoft’s Expedia were lured toward similar pricing formats (e.g., Flight Price Matcher). Recently, Priceline.com tried to extend this successful pricing model to sell groceries. However, this extension was not a success. One of the several potential reasons for the failure of this project is the inconvenience associated with bidding for each grocery item separately. Consider a consumer who purchases several items in a typical shopping visit. In order to purchase these items from Priceline.com, the consumer will have to place numerous bids, one for each item. As most grocery items are relatively low priced it may not be worthwhile for many consumers to invest time in such a cumbersome bidding process. Recognizing this problem, the company was considering allowing consumers to bid for a basket of goods instead of one item at a time. However, the company’s financial troubles precluded implementing this idea. Nevertheless, allowing consumers to bid for a shopping basket raises several interesting theoretical and managerial issues. In this paper, we examine these issues both theoretically and experimentally.

One of the key issues from a company’s perspective is the profit implication of allowing consumers to place bids for a shopping basket. On the surface, it appears that allowing joint bidding could encourage consumers to place lower total bids than they would have if they were bidding separately. This is consistent with pricing that we observe for product bundles. Typically, consumers pay less if they buy a product bundle. Similarly, if a consumer buys multiple units of the same product s/he typically would expect a quantity discount. Therefore, we might expect that producer profits can potentially decline if it allows for joint bids. A related issue is how allowing joint bids affects consumer surplus. Since the option to place joint bid for a basket offers more flexibility to the consumer, we would expect that consumer surplus to be higher.

In order to examine these issues we develop an analytical model. In our model, consumers can either bid for a basket of goods or can choose to bid for each item separately. Surprisingly, we find that in many situations producers’ profits are higher when they offer consumers the option to bid jointly. Interestingly, this increase is not at the expense of the consumerBin fact, consumer surplus is also higher when joint bidding is allowed. We test these theoretical results in an experimental setting. Our experimental results are consistent with the model predictions.

 

"THE AUTOMATIC PRICE CHANGE PROTOCOL: THEORETICAL AND EXPERIMENTAL INVESTIGATION"

Rami Zwick, Hong Kong University

Eythan Weg, Hong Kong University of Science and Technology

Benjamin Hak-Fung Chiao, Hong Kong University of Science and Technology

Several of the better-known consumers pricing alternatives on the Internet (e.g., auctions and bargaining) are currently prevailing in the offline b2b environment. It is assumed that applying the same methodology at the consumer level is not practical unless advances in information technology can make the implementation itself feasible. The Internet is providing a test-bed environment for experimenting with various old and new pricing mechanisms at the consumer level.

The study is motivated by a pricing rule implemented, for example, by OutletZoo.com. The site is using the Automatic Price Drop (APD) protocol resembles the basement pricing procedures at various large department stores. Prices fall by a seller-determined percentage at regular intervals. The Automatic Price Drop mechanism adds a new dimension for buyers, "should I grab the goods at today’s prices or wait until the next price drop and hope something’s left?" The procedure is becoming more sophisticated online, allowing the implementations of various variants. The procedure allows sellers of excess inventory to reach a global market of buyers interested in wholesale or retail deals, and allows them to quickly deplete their excess inventory at the best prices the market can bear.

In this paper we examine a generalized Automatic Price Change protocol. In particular, we study theoretically and experimentally the following questions:

1. What is the optimal price path for a fixed number of intervals, and does this price path indeed yields in practice the highest surplus for the seller? For example, if the price reduction must be done in four periods, what is the optimal discount path?

2. What is the optimal number of intervals for a fixed total price reduction and does this number of intervals indeed yields in practice the highest surplus for the seller? For example, given a total reduction of 90%, what is the optimal number of reduction periods?

3. Should the seller makes the exact price path (both in terms of number and magnitude of reduction) known in advance to the buyers? is this knowledge affect buyers’ timing decision?

4. How do buyers decide in practice when to enter the market?

To answer the descriptive questions we developed a simulated web site, adopting various variants of the Automatic Price Change protocol. In particular we manipulated the following variables:

1. The supply and the total number of potential buyers is either known with certainty or uncertainly known as a probability distribution.

2. The exact pricing schedule is either known or only the number of intervals or the total reduction is known.

3. All buyers are of the same type (i.e., have the same evaluation of the product), or two buyers types of high and low valuation exists in the market in known proportion.

4. Buyers either know or do not know the number of still available units at each period.

Maybe not surprisingly, we find that most of the theoretical prescribed path (for risk neutral buyers) does not yield the highest surplus for the seller. Further, risk attitude does not seem to explain buyers’ deviations from perfect coordination. We next present a behavioural model that account for most of the behavioural regularities observed in our study.

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