Printer Friendly

Advertising and product quality in posted-offer experiments.



For years the Federal Trade Commission (FTC) has attacked unfair and deceptive advertising. Recently it has turned its attention to errors of omission rather than commission, by challenging agreements among sellers to limit advertising even when it is truthful. One reason for eliminating advertising restrictions is to stimulate competition in the advertised dimension of the good or service. For example, price advertising may result in more price competition. On the other hand, professional associations argue that price competition causes product quality to deteriorate because sellers of high quality services are forced to match the lower prices of sellers of low quality products. And through their codes, professional associations have implicitly or explicitly discouraged some forms of advertising, including price advertising. Restrictions of this nature still exist in some local markets, although they have become the subject of FTC litigation. [1]

The quality deterioration argument of professional associations is plausible only when buyers are unable to observe quality prior to purchase. [2] When buyers cannot distinguish quality, moral hazard invites sellers to lower cost by providing lower quality. Adverse selection can result from price advertising, causing lower cost products to drive out higher cost products as argued by Akerlof [1970]. The opposite point of view is that even when quality cannot be determined prior to purchase, price advertising may enable sellers to signal quality and develop reputations. [3] Moreover, transactions repeatedly involving the same buyers and sellers may enable buyers to form expectations about seller quality based on their experience. That can make it worthwhile for a seller to provide higher quality.

The series of laboratory experiments described here permits the evaluation of market performance when buyers have very limited information. The claim to be investigated is that price advertising prevents quality deterioration and improves market efficiency. Reputation formation through advertising is emphasized instead of other mechanisms, such as warranties, that may prevent quality deterioration. [4]

So that quality can be kept relatively neutral and easily understood by subjects, we represent it as the "grade" of a commodity. The structure underlying the experiments is based on Holt's model of the determination of industry quality, which is summarized in Schmalensee (1979). Demand is induced through the values for which buyers can redeem their purchases. Value is linearly related to grade, so that no difference between average and marginal valuation of quality can arise to cause nonoptimal quality, as it did in Spence (1976). The production cost for each seller is quadratic in the grade selected by the seller. Therefore, the efficient grade level is unique.

The most closely related experimental work on product quality decisions is reported in Lynch, Miller, Plott and Porter (1986). In their experiments, sellers chose one of two quality levels. Prices were determined in an oral double auction. Efficient price and quality outcomes resulted when prospective buyers were given full information about quality. Removing the identities of sellers and allowing only post-purchase quality information sometimes resulted in a "lemons outcome" in which sellers offer a low-quality product, even though the market would be more efficient with the production and sale of a high-quality product. Private information tended to yield less efficient outcomes than full public information.

The posted-offer experiments reported in this paper differ from the oral-double-auction experiments in Lynch, Miller, Plott, and Porter because one of our objectives was to examine the effect of advertising restrictions on market efficiency. In our experiments, price advertising is implemented in a standard posted-offer auction. To implement the case of no price advertising, a buyer selects one seller before learning any seller's price; on learning the price of that selected seller, the buyer can reject it but then will have no transaction for the period. This restriction could not be imposed in a standard oral double auction. In addition, the posted-offer institution resembles many retail markets in which sellers price their goods in advance, and one objective was to determine whether the Lynch et al. results would be observed in posted-offer markets. [5] The grade-per-unit formulation of quality used here also allows us to work with a range of quality levels, as compared with the use of only two quality levels in the Lynch et al. paper. The use of multiple grades makes it possible to have some quality levels that are too high (goldplating) and others that are too low, relative to the surplus-maximizing quality.

The underlying market structure and experimental procedures are discussed in Section II. Results from all experiments are presented in Section III. The final section contains a summary of the results.


Market Structure

The market model that determines earnings will first be introduced in general terms; the parameters used in particular experiments will be specified subsequently. There are four buyers with identical induced preferences, and each may purchase at most one "unit" in a market period. The redemption value of the unit is B dollars times the grade level of the unit. If p is price, Q is market quantity of units, and [alpha] is the grade of the unit, the inverse built up demand function is:

[Mathematical Expression Omitted]

There are three sellers with identical marginal costs:

MC = 2Cq + D[alpha].sup.2

for [alpha [is greater than or equal to] 1 and 1 [is less than or equal to] q [is less than or equal to] 3 where C and D are positive parameters and q is the integer number of units sold. [6]

For each level of [alpha] there is a competitive equilibrium price (CE price). If [alpha] =1, the market supply function has a "step" with three units at 2C+D (the cost of first unit with [alpha] = 1 for each of three sellers) and a second step with three units at 4C + D (the cost of the second unit with [alpha] 1 for each of 3 sellers). Each of four buyers values the unit at B when [alpha]=1, so the market demand has a single step with four units at B. The CE price with [alpha]=1 is 4C + D if the redemption value of a unit exceeds the cost of the second unit, i.e. if B [is greater than] 4C + D. The determination of the CE price for any fixed level of [alpha] is illustrated in Figure 1. The notation above the market demand curve indicates reservation values of buyers B1, B2, B3, and B4; the notation above the supply curve indicates costs of units for sellers S5, S6, and S7. It is apparent from the figure that the competitive price for each level of [alpha] is equal to the cost of each seller's second unit if B[alpha] [is greater than] 4C + [D[alpha].sup.2]. The stability of this equilibrium is discussed in appendix A.

The exchange surpluses can also be illustrated with Figure 1. The surplus for each seller is 2C at the competitive price, and the surplus for each buyer is [B[alpha].-(4C + [D[alpha].sup.2]). With four buyers and three sellers, total surplus is: 10C + [4B[alpha].sup.2]. This surplus is maximized when [alpha]=B/2D, which will always be an integer for the parameter choices we make. The maximum value of the surplus is: 10C + [B.sup.2./D]. Denote the surplus maximizing value of [alpha] by [alpha] (*). The cost of the second unit is 4C + [B.sup.2./4D] when [alpha]=]alpha] (*), and this would be the CE price if all buyers knew in advance that the grade of each unit was [alpha] (*). Efficiency of any market outcome can be measured by the ratio of the total earnings of buyers and sellers to the maximum surplus. The efficiency, expressed as a percentage, is one basis for comparison of results under the different information conditions.

In some market experiments there were six possible quality levels, with B = $2.40, C = $.40, and D = $.40; in others there were twenty-one possible levels with B = $.60, C = $.40, and D = $.025. The characteristics of competitive equilibria with these parameters are shown in Table I for various values of [alpha]. Notice that with six quality levels market efficiency goes from 38.5 percent (at [alpha] = 1) to 84.6 percent (at [alpha] = 2) to 100 percent (at [alpha] = 3) as [alpha] increases from 1 to 2 to 3, and [alpha] (*) = 3 in this case. Similarly, [alpha] (*) = 12 for the twenty-one quality level case. With twenty-one quality levels, the grades 4, 8, 12, 16, 20, and 24 correspond exactly to grades 1, 2, 3, 4, 5 and 6 of the six quality level case. Thus if the number of quality levels appears to have no effect, a simple transformation will allow combining the data from experiments with these different parameter sets.


The experiments were conducted at the University of Virginia with subjects who were recruited from undergraduate economies courses. Subjects were told that they would participate in "an experiment in the economics of market decisions making" and that their earnings would be paid to them in cash privately. The instructions, which are described in appendix B, were read orally to all subjects at the beginning of each session. Subjects were assigned to buyer or seller positions in a nonsystematic manner. Three sellers were seated in the back of the room and were separated from one another by partitions. Four buyers were seated separately in the front of the room. All subjects could see the blackboard in the front of the room.

Each experiment consisted of a sequence of market "periods". Subjects were not told the number of periods that would be conducted, but the subjects' record sheets had rows for many more periods than were actually conducted. A period began with each seller recording the grade and unit price on a decision sheet. All units sold by a seller in a particular period were required to have the same grade and price. A seller could not change the grade from the previous period by more than two in the six-grade experiments, or equivalently, by more than eight in the twenty-one-grade experiments.

All experiments began and ended with several full-information periods in which the price and grade selected by each seller were written on the blackboard, in full view of buyers and sellers. Buyers were then selected in a random order and allowed to indicate a desire to purchase one unit from a specific seller according to the procedures of a standard posted-offer auction. A seller could refuse to sell an additional unit at any time, in which case the buyer could indicate a desire to purchase from another seller. The market period ended after all buyers had opportunity to purchase a unit.

The procedural difference under price advertising was that prices, but not the grades, were written on the board. Then, buyers were selected randomly and allowed to make purchases according to the procedures of a posted-offer auction just described. Buyers only learned the grade of a unit they purchased, not the grades of other units sold during the period.

The treatment with no advertising puts the buyers at even greater disadvantage. Sellers choose prices and grades at the beginning of each period, but nothing is posted on the board. After all buyers have been chosen in random order and have selected sellers, each buyer is shown the unit price for the seller selected. Each buyer can then reject the unit on the basis of such price information alone, but a rejection will result in no earnings for the buyer in that period. A buyer who decides to purchase is then told the grade of the unit selected, which means the buyer learns the price and quality of at most one seller. Regardless of whether price was advertised, sellers were always informed of all prices and grades at the end of each period.


Consider experiment 0, where six qualities were used and the efficient grade was 3. The transactions date are plotted in Figure 2. Note that each of the units sold in the first two periods had a grade of 4. The grade levels converge to 3, and the prices converge to the CE level of $ 5.20 for this grade. Even at the efficient grade and CE price, market efficiency will be below 100 percent when all four buyers purchase units from only two of the firms, thereby preventing the third firm from selling its low-cost unit. Market efficiency averaged close to 90 percent over all trials and 94 percent over the last five trials.

Now consider how sellers' incentives might change when buyers have less information. Under no advertising, a buyer will learn a seller's price offer only after all other options are gone, and will still not know quality. Since the outcome of any transaction will be known to the buyer involved but not to other buyers, the seller then has a limited incentive to deliver quality in relation to price. Realizing that, the seller may set a price for given quality above the CE price level. Under price advertising, some buyer inferences about seller quality may be possible, for example, by combining advertised price information with current and subsequent buyer behavior. But scant information is available for reputations. Setting price above the CE level should be less advantageous with price advertising, since buyers can be influenced by price in selecting sellers.

Four experiments were conducted with price advertising, where price information is public but quality information is private for buyers, and four experiments with no advertising, where both price and quality information to buyers is private. Each of these periods began with one to three full information periods and ended with about ten such periods. Thus, quality level, market efficiency or other measures under full or asymmetric information can be compared for the sme subjects. The average quality levels observed in each information condition are shown in Table II over all trials (with the number of trials in parenthesis), and over the last five trials of the treatment. [7] Over all sessions reported in Table II, the average quality under full information was 13.0. From the last five periods, the average grade level under full informaiton was slightly lower, at 12.6

Under both price advertising and no advertising in experiments 1 through VIII, there were lower average quality grades than with full information. Price advertising induced slightly lower average quality than no-price advertising, and in both of these cases quality tended to be considerably below the efficient level, although there was substantial variation across the experiments. Looking at the final five periods under asymmetric information, it is clear that a quality above the efficient level was achieved in one experiment under price advertising and in one under no advertising. (8) In the no-advertising periods, there was no clear relationship between the frequency of buyer rejections and the size of the markup over the CE price for the quality selected, presumably because buyers made the rejection decision on the basis only of current price information and past experience. In experiment IV, for example, when qualities were falling in the last half of the no-advertising treatment it appears that buyers were trying to reduce risk by rejecting any high price, whether caused by high markups on low quality or by low markups on high quality. The sequences of sellers' price and grade decisions are shown in Table III for experiment I, in which the grades fell rapidly to the minimum, apparently because buyers were unwilling to trust sellers at all. In two of the price-advertising experiments the quality for the last five periods remained absolutely at the minimum level, for an extreme lemons outcome. When buyer information was limited, price advertising thus could induce a deterioriation of quality, yielding a result no better than with no advertising.

To compare prices requires conditioning on quality. The CE price is used for any given quality level as a reference point. In Table IV the average difference between actual sale price and competitive equilibrium (CE) price for the actual quality of the unit is reported for experiments I through VIII. Although the greatest difference occurred with the price advertising treatment in experiment VII, the average price of $1.26 is lower with price advertising than the average price of $1.61 with no advertising. Sellers tend to set lower prices for given quality when they advertise price. Market efficiencies by information treatments are also reported in Table IV. Efficiencies are clearly lower in the private information treatments than in the full information case, where efficiency averaged 83.6 percent. Efficiency averages were no higher (at 46.4 percent) under price advertising than under no advertising (at 52.8 percent), although the difference is small compared with the ranges of outcomes in the two treatments.

The no-advertising treatment could work to the relative disadvantage of buyers, since it forces buyers to complete their transactions on a yes-or-no basis at a time when they have no alternative supplier. To examine the relative gains for buyers, we apportioned the total surplus between buyers and sellers, and the buyers' share is reported in Table IV. Based on surplus calculations with our parameters we should expect 76.9 percent of the surplus to go to buyers at the competitive equilibrium. Under full information the buyers quite consistently received more than half the total surplus, with an overall average being 63.8 percent of the total surplus. In the asymmetric information treatments the buyers did much worse. (9) They tended to suffer more during price advertising periods, where they actually lost money, than under no advertising, where they averaged only 10.4 percent of the total surplus earned. (10) The difference in buyers' surplus between price advertising and no advertising again is small relative to the range of outcomes, but it indicates that buyers tended to do no better under price advertising.


Price advertising has been examined here in experimental posted-offer markets to determine its effect on product quality under limited information conditions. Each of our experiments had a number of full information periods in which sellers advertised price and quality (truthfully). Fully information periods yielded a relatively high efficiency (84 percent) and an average quality of 13, slightly above the efficient level of 12.

In eight of the experiments buyers functioned with much less information. Tables II and IV reports results of four experiments where there was price advertising and four others where there was no advertising. Average quality over the four price-advertising experiments was only 7.1, and over the four no-price-advertising experiments it was 9.2. Two out of the four price-advertising experiments ended with five periods in which all transactions were at the lowest possile quality level of 4. For given quality, price tended to be slightly lower under price advertising. Market efficiencies under asymmetric information, with or without price advertising, were only slightly more than half as high as they were under full information. The buyers suffered; in the market periods with price advertising they lost money on average, and under no-advertising their earnings were small, as Table IV shows. When buyers knew only the quality of units from their own transactions, reputations apparently were not strong enough to allow efficiency. The experimental design exaggerates the effect of this lack of buyer information by having one buyer deal with only one seller in each period. Nevertheless, the effect would be even greater if there were more buyers and sellers, in that one buyer would then learn from a smaller portion of the market each period.

Thus price advertising appears unable to assure efficient market outcomes when buyers know qualities only from their own transactions. The lack of ex post information for reputations allows sellers to take advantage of buyers, and can cause buyers to accept only the lowest possible quality level, which yields a "lemons" outcome. (11) Yet under each asymmetric information treatment there was one experiment in which the quality ended up being inefficiently high. Although the no-advertising alternative provides buyers with even less information to use in selecting a seller, qualities and efficiencies tended to be no worse under this condition. Lack of buyer information about quality thus handicaps market performance, whether or not price advertising is permitted, in an experimental setting where full information produces an efficient outcome reliably. In these experiments, limited quality information for buyers is a potential problem that cannot be solved by price advertising alone.



If buyers are able to observe or somehow infer the level of [alpha] prior to purchase, then the surplus maximizing CE price would be the CE price for [alpha] = [alpha]*. To see how this might be achieved, suppose the market is intially at another CE, say at a CE for an integer level of [alpha], denoted [[alpha].sub.0], that is below [alpha]*. Then a one-unit increase in [alpha] will raise the value of the unit determined in equation (1) by $B. It follows from equation (2) that a one-unit increase in [alph] will also raise the cost of any unit by an amount [D([[alpha].sub.0] + 1).sub.2] - [D[alpha].sup.2.sub.0], an increase D([[alpha].sub.0] + 1). Thus a single seller, who could have sold the second unit at the CE price for the initial level [alpha], could inrease the grade of the unit to [[alpha].sub.0] + 1 and raise the price by an amount B. The increase in price, B, exceeds the increase in cost for the unit if B [is greater than] D(2[[alpha].sub.0] + 1), or equivalently, if [alpha] [is less than] B/2D - 1/2 = [alpha]* - 1/2. This inequality is satisfied because [[alpha].sub.0], being integer valued, must be less than or equal to [alpha]* -- 1. It follows that a seller can profitably raise the quality level if it is below [alpha]* and if the qualify increase can be observed or inferred by buyers pior to purchasing. An analogous argument can be used to show that a unilateral reduction in is profitable for the seller of a second unit if the initial level of [alpha] exceeds [alpha]*. Under the asymmetric information treatments, buyers could not observe the grade prior to purchase, so the full information CE should be interpreted as providing a benchmark for efficiency calculations, not a prediction.

Although unilateral quality deviations are unprofitable at the CE for [alpha] = [alpha]*, for the CE to be a Nash equilibrium it must also be true that unilateral price deviations are unprofitable. The CE is a Nash equilibrium if we assume that rationing maximizes surplus at identical prices. In Figure 1, for example, this implies that all three sellers will sell their low cost unit if they choose the CE price; it rules out the case in which two sellers each sell two units. Given this rationing rule, a price cut from the CE level will not increase a firm's sales and a price increase will result in zero sales, so unilateral price deviations are unprofitable. Thus sellers have no market power. (12) The rationing rule actually used in the experiments was random. With this rule, a seller has an incentive to lower price or raise quality slightly to ensure the sale of a first unit.



The instructions describe a standard posted-offer auction in which sellers first select a grade for their product. The "Market Organization" section of the instructions, which differs from that of a standard posted-offer auction, is reproduced below. Passage 1 and passage 2 in square brackets are altered to implement the various information treatments, as indicated. A complete copy of the instructions is available from the authors on request.

Market Organization

The market for this commodity is organized as follows. The market will be conducted in a series of trading periods. After we open the market for each trading period, each seller decides on a sale price per unit and a grade per unit, which he or she will write on one of the cards provided. The grade selected by a seller must be no more than two levels above or below the level selected by that seller in the previous period. Sellers will be given up to three minutes to submit their prices and grades. [Passage 1] Buyer's bids will be made as follows: a buyer will be chosen randomly and will state the seller from whom he or she wishes to buy. The seller may then accept or reject the buyer's bid. If the first seller selected by the buyer is not willing to sell a unit, the buyer is free to choose a second seller, and so on. After the first buyer has made a purchase, another buyer is selected at random, and this buyer will make his or her desired purchase. [Passage 2] This process continues until all buyers have had a chance to purchase. Then all buyers and sellers calculate their own earnings for that period. This completes the trading period. We will reopen the market for a new trading period by having sellers submit new prices and grades. The whole process is then repeated.

1. Full Information Treatment:

Passage 1: [The cards will be collected, and the prices and grades will be written on the blackboard. Buyers will then be free to make bids to buy units and to specify the seller with whom they wish to deal.]

Passage 2: (nothing inserted).

2. Price Advertising Treatment:

Passage 1: [The cards will be collected, and the prices will be written on the blackboard. Buyers will then be free to make bids to buy units and to specify the seller with whom they wish to deal, but at the time bids are made, buyers will not know the exact grades that sellers have selected.]

Passage 2: [Then sellers are informed of the numbers of units that they actually sold, and each buyer is informed of the grade of the unit that he or she purchased. Buyers are not told the grades of the units purchased by others. Sellers will be informed of one another's offerings at the end of each period.]

3. No Advertising Treatment:

Passage 1: [The cards will be collected. Buyers will then be free to make bids to buy units and to specify the seller with whom they wish to deal, but buyers will not know either the prices or grades of units at the time bids are made.]

Passage 2: [Then each buyer is shown the price charged by the seller, and each buyer may decide whether or not to accept the unit at this price. If the unit is rejected, there will be a zero profit on that unit for both buyer and seller. Then sellers are informed of the number of units that they actually sold, and each buyer is informed of the grade of the unit he or she purchased. Buyers are not told the grades or prices of units purchased by others. Sellers will be informed of one another's offerings at the end of each period.]

(1) For example, in 1984 the FTC charged that the Rhode Island Board of Accountancy had restained competition by prohibiting advertising and other forms of soliciting business. The Board of Accountancy consented in the case, and now restricts advertising only if it is false or misleading. Other cases have been pursued under an active FTC program opposing advertising restrictions, especially against state boards. One recently litigated case involves the Massachussetts Board of Optometry.

(2) Even if restrictions on price advertising raise quality, it does not necessarily follow that such restrictions will increase overall market efficiency. Chan and Leland [1982] present a theoretical analysis of the question of whether advertising can improve welfare. They find improvement in welfare is likely when advertising is permitted, but improvement cannot be assured. Rogerson [1986] considers a model in which price-advertising generally increases welfare when consumers' preferences for quality attributes are heterogeneous.

(3) For analysis of this type of situation see Cooper and Ross (1954), Riordan [1986], Shapiro [1982], Smallwood and Conlisk [1979], and von Ungern-Sternberg and von Weizsacker [1985]. Miller and Plott [1985] examined signaling of product quality in experimental markets.

(4) Warranties are less likely to be used effectively when monitoring and settlement costs are high.

(5) Knowing they cannot adjust their offers, sellers in a posted-offer market are motivated to behave strategically as bidders do in a first-price, sealed-bid auction (Plott and Smith [1978]) and, under full information conditions, posted-offer markets have tended to yield higher prices and lower efficiencies than double auctions (Ketcham, Smith and williams [1984]).

(6) The total cost function which yields equation (2), with a fixed cost of zero, is TC = 0 if q = 0,TC = Cq + [Cq.sup.2] + D[alpha].sub.2.q] if q [is greater than] 0. Note that the marginal costs in equation (2) are for discrete changes in q, i.e., for the "cost of first unit," "cost of second unit," etc. Thus MC in equation (2) is not the derivative of TC. For example, consider total cost with [alpha] = 1: the total cost for q = 0 is zero, the total cost for q = 1 is 2C + D, and the total cost for q = 2 is 6C + 2D. These total cost nubmers yield marginal costs of 2C + D for the first unit and 4C + D for the second unit, which are consistent with equation (2).

(7) Experiments I through IV were conducted with six quality choices and experiments V through VIII were conducted with twenty-one quality levels. Since no significant difference was observed, qualities from the six quality condition have been multipled by four; that makes them numerically comparable to qualities in the twenty-one quality condition. A data appendix is available from the authors on request.

(8) Since the asymmetric information treatment periods were always followed by a series of full information periods, the last five periods of the asymmetric information treatment were not near the end of any experiment.

(9) At each experimental session, the same subjects participated in at least two information treatments, one of which was always full information. To examine the persistence of performance by particular buyer and seller groups, we estimated the correlation in buyers' share of surplus between the asymmetric information condition and the full information condition. There are only eight observations (reported in Table IV from experiments I through VIII) but they yield a very high nonparametric Spearman rank correlation coefficient of 0.952, which is significant at the 0.01 level. The strategies or skills of particular buyer and seller groups appear to have an effect on outcomes regardless of infomation condition.

(10) Although buyers' earnings were declining during parts of these experiments, we did not get the impression that incentive control was lost. In retrospect, however, it would have been better to adjust the parameters to raise buyers' share of earnings.

(11) We conjecture that market efficiency would be enhanced with an ex post public information condition in which seller price and quality information would be made public to buyers at the close of each period. We did conduct two pilot experiments with this ex post public information, one with experienced subjects. With inexperienced subjects, results were not very different from those reported here under private information, but it was obvious that learning was still going on when the experiment ended. Experienced subjects reached quite efficient outcomes.

(12) Holt [1988] discusses the measurement of market power in laboratory experiments.


Akerlof, George A. "The Market for 'Lemons': Quality-Uncertainty and the Market Mechanism." Quarterly Journal of Economics, August 1970, 488-500.

Chan, Yuk-Shee and Hayne Leland. "Prices and Qualities in Markets with Costly Information." Review of Economic Studies, October 1982, 499-516.

Cooper, Russell and Thomas W. Ross. "Prices, Product Qualities and Asymmetric Information: The Competitive Case." Review of Economic Studies, April 1984, 197-207.

Holt, Charles A. "The Exercise of Market Power in Laboratory Experiments." Working paper presented at the FTC conference on Empirical Approaches to Market Power, August 1988, forthcoming in The Journal of Law and Economics.

Ketcham, Jon, Vernon L. Smith, and Arlington W. Williams. "A Comparison of Posted-Offer and Double Auction Pricing Institutions." Review of Economic Studies, October 1984, 595-614.

Lynch, Michael, Ross M. Miller, Charles R. Plott, and Russell Porter. "Product Quality, Consumer Information and 'Lemons' in Experimental Markets," in Empirical Approaches to Consumer Protection in Economics, edited by Pauline M. Ippolito and David T. Scheffman. Washington, D.C.: Federal Trade Commission, 1986, 251-306.

Miller, Ross M. and Charles R. Plott. "Product Quality Signaling in Experimental Markets." Econometrica, July 1985, 837-72.

Plott, Charles R. and Vernon L. Smith. "An Experimental Examination of Two Exchange Institutions." Review of Economic Studies, February 1978, 133-53.

Riordan, Michael H. "Monopolistic Competition with Experience Goods." Quarterly Journal of Economics, May 1986, 265-79.

Rogerson, William P. "Price Advertising and the Deterioration of Product Quality." Presented at the Econometric Society Meetings, Durham, N.C., June 1986.

Schmalensee, Richard. "Market Structure, Durability, and Quality: A Selective Survey." Economic Inquiry, April 1979, 177-96.

Shapiro, Carl. "Consumer Information, Product Quality and Seller Reputation." Bell Journal of Economics, Spring 1982, 20-35.

Smallwood, Dennis E. and John Conlisk. "Product Quality in Markets Where Consumers Are Imperfectly Informed." Quarterly Journal of Economics, February 1979, 1-23.

Spence, Michael. "Product Differentiation and Welfare." American Economic Review, May 1976, 407-14.

von Ungern-Sternberg, Thomas and Carl Christian von Weizsacker. "The Supply of Quality on a Market for 'Experienced Goods'." Journal of Industrial Economic, June 1985, 531-41.

Professors of Economics, University of Virginia. This research was supported in part by the National Science Foundation under grants SES 82-19720 and SES 87-20105. We would like to thank Amy Beherent, Eric Engen, Philip Jefferson, and Kevin Terhaar for research assistance.
COPYRIGHT 1990 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1990 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Holt, Charles; Sherman, Roger
Publication:Economic Inquiry
Date:Jan 1, 1990
Previous Article:Does exchange rate appreciation 'deindustrialize' the open economy? A critique of U.S. evidence.
Next Article:Worker attitudes and the cost of production: hypothesis tests in an equilibrium model.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters