Marketing of credence products: a game theoretic perspective on the underselling of antiques

Tags: resellers, auctioneers, authenticity, certainty, dual channel structure, London, credence goods, authentic antiques, products, auctioneer, signals, sales channel, Journal of Law, Economics, and Organization, offer, Flexibility.on Export Channel Performance, David Gilliland, Nancy T. Gallini, Journal ofMarketing Research 22, Quarterly Journal of Economics, Competitive Price, Jan Heide, London Antiques Trade, Jeffrey L., Nacy A. Lutz, Mark Bergen, Export Market Ventures, Journal of Political Economy, reputable auction houses, channel structure, Uncertainty Sales Channel, James Brown, RAND Journal of Economics, Dual Distribution, Journal of Marketing Research, Shaoming Zou, Journal of Health Care Marketing, Free Competition, selling antiques, auction house, studies focus, Kirmani, National University of Singapore, Department of Marketing, School of Business, auction channel, dual channels, experience goods, dual channel, product quality, reputable dealers, auction, Department of Business Policy, School of Business, Journal of Consumer Research, Credence Goods Darby, Robert G. Eccles, George John, Amna Kirmani, art and antiques, quality screening, School of Business, reputable dealer, Business Link, international dealers, International Journal of Industrial Organization
Content: MARKETING OF CREDENCE PRODUCTS: A GAME THEORETIC PERSPECTIVE ON THE UNDERSELLING OF ANTIQUES January 2002 Khai SheAng Lee*, Wei Shi LIM** & Siew Lien SIM*** RPS #2002-006 (MKTG) * Associate Professor, Department of Marketing, School of Business, BIZ 1 Bldg., National University of Singapore, Business Link, Singapore 117592. E-mail [email protected] ** Associate Professor, Department of Marketing, School of Business, BIZ 1 Bldg., National University of Singapore, Business Link, Singapore 117592 *** Research Assistant, Department of Business Policy, School of Business, BIZ 1 Bldg., National University of Singapore, Business Link, Singapore 117592. Copyright © School of Business, National University of Singapore.
Marketing of Credence Products: A Game Theoretic Perspective on the Underselling of Antiques Abstract This paper seeks to provide a rational explanation as to why authentic antiques are knowingly undersold at prices that are significantly below their actual worth. Using a game theoretic approach, we show that uncertainty and non-verifiability of the quality of credence products (in terms of authenticity) post-purchase result in adverse selection, such that reputable auctioneers are more likely than less reputable resellers to reject items offered to them for auction, even though these items might indeed be authentic antiques. A direct consequence of such adverse selection is the observed plural channel structure comprising auctioneers and dealers of different reputation in the market for antiques, whereby less reputable resellers knowingly undersell authentic antiques that were rejected by the more reputable resellers. Reputable resellers thus perform the role of quality screening, commanding premium prices for the antiques they sell, whereas less reputable resellers fill the role of market clearing, selling antiques at prices that commensurate with buyers' confidence of the product's authenticity. In turn, this implies that a low price does not signal the lack of quality in credence products. While the results suggest that a reseller's reputation might serve as a signal of quality for credence products, it is importantto note that such a signal is imperfect, as the quality of credence goods, by definition, is not verifiable with certainty, and hence, the potential for moral hazard behaviors by sellers is always present.
Marketing of Credence Products: A Game Theoretic Perspective on the Underselling of Antiques INTRODUCTION Credence goods are defined as products or services the quality of which cannot be perfectly ascertained even after purchase and consumption (Darby and Kami, 1973). Antiques, a form of credence products, are objects with antiquity, typically of at least a century old, and range from paintings, sculptures, calligraphy, jewelry, pottery, to even furniture. The international market for antiques is a lucrative multi-billion dollar business. It was reported that an estimated 6,000 major international dealers of art and antiques raked in annual sales of between US$8 billion and US$15 billion (USA Today, 2000). International auctions of art and antiques took in another US$4 billion in sales each year (Hughes, 2000). These sales figures do not even include those of local dealers and auctioneers in their respective countries, which potentially add billions more in sales to the total market for antiques. In general, antiques are sold through two types of channels or resellers: the auction channel, which comprises auction houses, and the dealer channel, which comprises various types of dealers, such as flee-market stall operators, general antique dealers, and specialist traders. Dealers differ from auctioneers in that while dealers generally take title to the items they sell, auctioneers do not. Instead, the latter, for a commission or a fee, serve as agents for sellers in auctioning their antiques. It is not unusual for an antique to pass from one reseller (a dealer or an auctioneer) to another, with a price escalation accompanying each transaction (Cooper, 1985). For example, Cooper (1985) reported a case that traced the transaction history of an enamel swan-shaped pendant, which first appeared in a fleemarket stall in the Bermondsey Market in England. The stallholder had bought the
pendant at an auction in East Anglia as part of the contents of a box of knickknacks. The stallholder subsequently sold the pendant for Ј40 to a dealer, who resold it for Ј80 to another dealer in less than an hour. This dealer, after three weeks of search for another customer, subsequently sold the pendant for Ј150 to a specialist trader in the Bond Street Arcade, who wisely decided to figure out its exact value before reselling it. After he had ascertained that the pendant was likely a valued possession in the Victorian period that could fetch a price in excess of Ј2000, he sold it for Ј850 to a Bruton Street specialist trader. This trader then sold the pendant for Ј1500 to a Jermyn Street specialist trader, who identified it as one from the collection of the Dukes of Rochester. With the pendant boasting of a ducal pedigree, the new owner easily found a buyer for it at over Ј4000, and the swan completed its migration in a deposit box in Fort Lauderdale. Generally, more reputable auctioneers can command higher prices for the items that they auction than less reputable resellers (Cooper, 1985). Given that this is so, for a small fee, a less reputable dealer could apparently sell his ware via reputable auctioneers, instead of selling it directly to buyers, to obtain higher prices. This leads to a curious question as to why a dealer would undersell a piece of supposedly antique item directly to buyers, instead of via a reputable auctioneer? Underselling occurs when an authentic antique item is sold at a price that is substantially below its true worth, or the value that is realizable by a reputable auctioneer in a major international auction. An obvious explanation for this is that the seller who undersold a piece of antique is ignorant of its worth. In the case reported by Cooper (1985), it is probably true that the flee-market stall operator who sold the pendant for Ј40, and the two dealers who resold it for Ј80 and Ј150 were ignorant of the pendant's worth. However, it is puzzling that the specialist trader in Bond Street Arcade should sell the
pendant for Ј850 to another specialist trader, even when he had ascertained the pendant to be worth more than Ј2000, and might have it auctioned via a reputable auctioneer for a price in excess of Ј2000, or even Ј4000. Why then would a dealer who was informed about the authenticity and the true worth of a piece of antique be willing to undersell it? A possible explanation is that either the seller is less than rational, or the item being undersold by an informed seller is a fake! Although this may be true in some situations, there are several limitations to this argument. As Cooper (1985) has illustrated, the pendant that was undersold by the Bond Street specialist trader, who was an informed seller, was indeed authentic. Moreover, the underselling of authentic antiques by informed dealers is not unusual (Cooper, 1985; Brown, 1992). Indeed, as we shall show, were it true that items undersold by informed dealers are fakes, then like the market for lemons (Akerlof, 1970), the dual channels for the marketing of antiques will collapse, such that only reputable auctioneers will continue to operate, while less reputable dealers and auctioneers will cease to exist. That resellers with varying degrees of reputation continue to operate in the market for antiques therefore raises serious doubts about the argument that items undersold by informed sellers are fakes. This paper examines the phenomenon of antiques being undersold by less reputable resellers, and the co-existence of reputable auctioneers and less reputable dealers and auctioneers in the market for antiques. In particular, we seek an alternative and rational explanation as to why an informed dealer (or any seller) would undersell a piece of authentic antique directly to a buyer, when the former might auction it through a reputable auctioneer to obtain a substantially higher price. Indeed, this puzzle was first put forth to us by the chief executive officer of a major international auction house, whichmotivated the studyin this paper.
Following a game theoretic approach, we examine a dealer's choice in selling his products directly or via an auction house, when quality in terms of authenticity cannot be verified with certainty, and derive the Nash equilibrium in terms of sales channel and price outcomes. Compared to prior studies in marketing channels, we propose a different explanation, one that is based on uncertainty and non-verifiability of quality, for the existence of a dual channel structure that comprises of an auction and a dealer channels in the marketing of a class of credence products. Implications with regards to signals of quality for and the marketing of credence products and services, as opposed to experience goods, are discussed. Literature Review Two streams of research that are more directly related to our study are those on the marketing of credence goods and dual channels of distribution. Research in Credence Goods Darby and Kami (1973) defined credence goods as products or services the quality of which cannot be perfectly ascertained even after purchase and consumption. This is unlike search and experience goods, which possess quality characteristics that may be perfectly evaluated with certainty before and after purchase, respectively (Nelson, 1970). An example of a search good is a dress, which may be tried on before purchase to accurately evaluate its fit, and an example of an experience good is canned tunas, for which the most effective way to determine its quality would be by actually tasting it after purchase (Nelson, 1970). Compared to search and experience goods, the quality of credence goods cannot be determined so simply. Examples of credence goods, in addition to antiques
and objects of art, include health foods and tonics, and services such as geomancy, car-repairs (Emons, 1997), and medical diagnoses (Lynch and Schuler, 1990). For instance, the quality of tonics (such as ginseng), in terms of their purported benefit of improving one's health after consumption, may not be easily evaluated even after consumption, as it is difficult to ascertain if one's health has really improved from its consumption. Even if one did feel better, it is difficult to attribute one's improvement in well being to the ginseng consumption alone. Likewise, for a supposedly antique item, the characteristic that is difficult to ascertain is the item's authenticity, which is the main factor that gives rise to its value. For example, a pair of gold-coated enamel dishes attributable to the imperial collection of the Emperor Qianlong could fetch as much as HK$250,000 at a major auction, whereas a pair of near perfect reproduction of these dishes would cost only HK$20,000 (Hoh, 1998). Given the difficulty in evaluating the quality of credence goods, extant literature on credence goods has mostly focused on the problem of fraud, especially in the provision of expert services, where sellers are asymmetrically informed against consumers (Pitchik and Schotter, 1987; Wolinsky, 1993; Emons, 1997 and 2001). These studies examine the problem of moral hazard on the part of sellers, who may act opportunistically by misrepresenting and/or over-prescribing the type of services required. In contrast to these studies, our model considers the case in which agents are honest, and hence, the issue of moral hazard does not arise. Instead, we investigate a problem of adverse selection, in which informed sellers self-select to undersell high quality credence products. Akerlof (1970) also examined a problem of adverse selection in the market for used cars, which are experience goods. The author showed that only lemons would be offered at the average market price, which would lead to a market collapse. Unlike
used cars, the quality of antiques (in terms of authenticity) cannot be verified with certainty even after purchase. For such credence products, we show that underselling does not signal that the item offered for sale lacks quality (or authenticity). Otherwise, if it was true that underselling signals the lack of quality, then the dual channel structure in the market of antiques will collapse such that only reputable auctioneers will continue to operate, while less reputable resellers will cease to exist. We further show that the dual channel structure will also collapse such that only the dealer channel will continue to operate, while the auction channel will cease to exist, in the special case that if indeed authenticity can be verified with certainty. Hence, under this special case, we obtain results for credence products that are consistent with those of Akerlof s (1970) for experience goods. Extensive research has also been conducted on signals of product quality. Particularly helpful is a recent review of the literature on the signaling of unobservable product quality by Kirmani and Rao (2000). From the authors' review, it is evident that the majority of the studies on signals of product quality focus on experience goods (e.g., Gerstner, 1985; Tellis and Wernerfelt, 1987; Boulding and Kirmani, 1993; Chu and Chu, 1994). Little research attention, if any at all, was devoted to the investigation and identification of signals of quality for credence goods. It is therefore unclear if the results obtained from the studies on experience goods would apply similarly to credence goods, since, unlike experience goods, the quality of credence goods is not readily revealed even after purchase and consumption. In fact, Kirmani and Rao (2000) suggested that signaling may not be possible for credence goods, because "postpurchase inspection does not unambiguously reveal quality, [and therefore] consumers are unlikely to be able to exact retribution on the offending seller" (p72-73). While we do not explicitly
examine signals of quality for credence goods, several implications arise from our results on price as a signal of authenticity of antiques, and of the quality of credence products in general. Research in Dual Channels Although there is an extensive literature on marketing channels (e.g., Pondy, 1967; Stem et al., 1973; Frazier and Rody, 1991; Anderson et al. 1987; John and Weitz, 1988; Anderson and Weitz, 1992; Heide, 1994; Lusch and Brown, 1996; Heide and John, 1992; Cavusgil and Zou, 1994; Bello and Gilliland, 1997), the majority of these studies focus on a firm's decision to make or buy (or to internalize or outsource). There are few studies that examine dual channels (Dutta et al., 1995) or plural distribution forms (Bradach and Eccles, 1989), of which the focus is on a firm's decision to make and buy, or to internalize and outsource its distribution function, simultaneously. For example, Carlton (1979) examined partial integration and suggested that it arose because firms integrated and used the market to satisfy the high and low probability demands respectively. Gallini and Lutz's (1992) study of franchisors' store-mix decision suggested that information asymmetry between franchisors and potential franchisees lead to a franchisor's decision to have its own stores as well as to franchise others, as stores owned by new franchisors signaled business prospects to potential franchisees. Dutta et al. (1995) examined a dual channel that comprised a direct sales force and a rep channel, and proposed that "adding a direct sales force to augment the rep channel served as a safeguard against lock-in problems with the reps" and "provides a manufacturer with insight into downstream marketing activities" when performance is ambiguous (pl94). Farrel and Gallini (1988) examined why
firms make-and-license, and proposed that a monopolist intentionally licensed other firms to utilize its proprietary technology to produce and compete directly with it because second sourcing or invited competition served as a safeguard for a buyer's specific investment. Extending the authors' study, Dutta and John (1995) empirically tested the theoretical propositions presented by Farrel and Gallini (1988). Unlike these studies on dual channels, our study involves a dual channel structure that comprises two types of resellers: auctioneers and dealers. In particular, we are interested in the coexistence of reputable auctioneers and less reputable dealers, in relation to the phenomenon of the underselling of authentic antiques by less reputable dealers. We propose a different explanation from those advanced by prior studies for the dual channel structure in a market for credence products. MODEL DEVELOPMENT Stylized Facts (I) Reputable Auctioneers Command Higher Prices In general, reputable international auctioneers such as Christie's and Sotheby's (which together account for 95% of sales in the international auction market for art and antiques (The New York Times, 2001)), command higher prices for the items they auction than do less reputable dealers and auctioneers (Cooper, 1985). It is not uncommon for dealers to substantially mark down product prices based on the prices of comparable pieces published in the auction catalogues of reputable auctioneers. Furthermore, the prices quoted by dealers, which have already been marked down, are further negotiable, and it is well known in the trade that the eventual transacted prices are often much lower than those listed in the auction catalogues (see Brown, 1992, for an example). It has also been reported that even in flea markets, authentic antiques
can at times be bought at fire-sale prices (Kwang, 2000). In fact, Sotheby's recently reported a record sale in an online auction of an original copy of the Declaration of Independence that was auctioned for US$8.14 million, but which was initially sold for a mere four dollars by an unknown seller (Sotheby's Annual Report 2000)! (2) Uncertainty in the Verifiability of Authenticity The authentication of a supposedly antique item is based on experts' opinions and/or scientific testing performed on a sample of the item. Scientific testing is not only costly to conduct, but the margin of error in scientific testing is also high (Hoh, 1998). Even if an item can be scientifically attributed to a particular period or age, one cannot be absolutely certain that it is authentic. This is because there is still the possibility that the item is an antique reproduction made during that period, or a modem reproduction made using materials recovered from that period (Cadogan, 1987). This is especially true of antique artwork, the value of which is highly dependent on it being an artist's original production. As one reporter pointed out, science can only say that "this is a work of art which was created with materials you would expect to find in that age, ... but it will never know by whom... a machine cannot identify the hand of the artist; the work might come from a pupil" (Adam, 1995). For these reasons, authentication by scientific testing is not necessarily more accurate than that based on experts' opinions. Yet, authentication based on experts' opinions is highly subjective, and it is well known in the trade that conflict of opinions among experts is not unusual. This is evident from the numerous reported disputes among experts (like are art historians, auction house experts or museum consultants) over the authenticity of various art and antique objects (see for examples, Battles, 2000; Joseph, 2000; Mendick, 2000; Sian,
2000a and b). It was reported that even items that were identified as fakes by some experts remained to be on public display in museums, which is suggestive of the inconclusiveness of opinions among experts (Watson, 2000). Even items retrieved from scientific archeological excavations may not be certainly authentic, since the integrity of the excavation is only as good as that of the archaeologists undertaking the project. For example, it was reported that a prominent archaeologist was discovered to have planted a purported archaeological find in a major scientific excavation that he led (The Nikkei Weekly, 2000)! As a result of the difficulty in authenticating supposedly antique items, it is therefore of no surprise that the trade has turned to various signals of authenticity, such as price and provenance. However, these signals are not infallible (Kirmani and Rao, 2000). For example, provenance - the history of ownership of an antique item - can be faked, as it is very difficult or even impossible to verify the authenticity of the provenance for a supposedly antique item simply because many a time, the galleries listed on the provenance are already defunct and/or the owners have passed on (Athineos, 1996). Given the difficulties in the authentication of antiques, it is apparent that experts in the trade can only express an opinion on the likelihood that a supposedly antique item is authentic. model specification As our objective is to investigate whether an informed seller might rationally undersell a piece of authentic antique, we present a game of uncertainty as shown in Figure 1, with the following features. 10
i. Authentic Antique Item: Given that the objective of our study is to examine the phenomenon of underselling of authentic antiques, we consider the case in which the item offered by the dealer (D) is an authentic antique. ii. Knowledgeable Players: As our aim is to seek a rational explanation for underselling, we consider the case when all players (dealer, collector and auctioneer) are knowledgeable in assessing the authenticity of an antique item. However, although all players are knowledgeable, they cannot determine with certainty the authenticity of the item presented to them (stylized fact). iii. Asymmetric Information: In order to investigate the phenomenon of underselling by an informed dealer, we consider the case in which information is asymmetric, that is, the dealer (D) possesses more information about the authenticity of the item that he offers, and hence, believes with greater certainty (which we normalize to one) that the item is authentic, compared to the auctioneer (A) and the collector (Q. However, the seller D cannot prove or credibly communicate to A or C that the item is indeed authentic, since authenticity cannot be verified independently with certainty. Hence, even though A and C are knowledgeable, they cannot ascertain the authenticity of the item presented to them, but can only assess the likelihood that it is authentic. The players' assessments of the authenticity of the item need not agree among themselves. iv. Honest Players: To seek an alternative explanation for underselling, other than one that is based on moral hazard and misrepresentation, we therefore consider the case where all the players are honest. [Insert Figure 1 here] With reference to Figure 1, in Stage 1, seller D decides whether to sell an item directly to C or indirectly via A (Ag{H,L}), who may be more (H) or less (L) 11
reputable than D. For example, more reputable auction houses (H) are those such as Christies' and Sotheby's, whereas less reputable ones (L) may be lesser-known regional or local auction houses such as Gilding's Auctioneers, Priddy's Auction Galleries, and Raffles Fine Art and Auctioneers. The reputation of an auctioneer is assumed to be common knowledge among all players in the game. In Stage 2, when Player /, where i e {C,H,L}, is offered an item, i evaluates the probability 6i that the item is authentic. Based on his assessment 0n Player i decides whether to accept or to reject the item. If i is an auctioneer H (or L) and decides to accept the item, then it will auction the item on behalf of D at a price v# (or vЈ). If i is a collector C and decides to accept the item, then he will purchase it at a price Pc. If all players reject the item, there is no trade and each player receives a zero payoff. Let v be the worth of an authentic antique item, which we define as the price that is realizable by a reputable auctioneer. We use v^ to denote the realizable value of an authentic antique item in an auction, which depends on the auctioneer's reputation. Hence, vH = v, and since a reputable auctioneer can generallycommand a higher price for the item than a less reputable auctioneer, we write vL =kvexpected payoff is given by n = $c (v - Pc) +(1 - 0C){-PC), where the loss to C is the price paid Pc if the item 12
bought turns out to be a fake, and the gain by C is the surplus (v- Pc) if the item is an authentic antique. On the other hand, if i-A, then A's expected payoff in successfully auctioning an item is given by nA =&A(aAvA) + (l-0A)[-(A)], where the gain to A is the fee he receives (aAvA) if the item is authentic. However, if the item turns out to be a fake, then the loss to A is some value {L)). The notation aA = (asA +abA) is the commission received by A from seller D (asA) plus that received from the eventual buyer of the item (abA). Without loss of generality, we normalize abA and the cost of the item to D to zero. Hence, we write aA =asA and, if/ accepts the item offered, the payoff to D is simply Pt(Auctioneer's and Collector's Decision Criteria (a) Dealer Sale: IfD Offers the Item to C In Stage 2, C agrees to purchase the item at a price Pc if and only if his expected payoff is positive, that is, Oc(v-Pc) + (\-0c)(-Pc) > 0, which can be
rewritten as 6C > --P . Let --P-- = 6*c > 0. Hence, for trade to occur, it is necessary
v
v
that Condition (1) is satisfied.
#c> -- =<Ј>0.
(1)
v
This requires that either, (i) C is sufficiently confident that the item offered by
D is authentic, that is, 6C is sufficiently large, or (ii) 6*c is sufficiently small, which
happens when the transacted price Pc is sufficiently small compared to the item's
p worth v. Note that since v > Pc > 0, the expression -- is therefore always less than v p or equals to 1, that is, 1 > -- > 0. v
(b) Auction Sale: If D Offers the Item to A On the other hand, ifD offers the item to A, A accepts it for auction at price vA if and only ifA's expected payoff is positive, that is, 6A(aAvA) + (1 - 0A)[-0(A)] > 0,
which reduces to Condition (2), where
is rewritten as 6\ .
eA> ^A) =e\
(2)
· ccAvA+(j)(A)
For a reputable auctioneer (v4 = H), 6*H tends to 1 if the cost of damage to its
reputation ^(H) due to the acceptance of a non-authentic item for auction is very
14
large. Hence, Condition (2) with A= H is more difficult to satisfy than Condition (1). This implies thatunless a reputable auctioneer is verycertain (0H close to 1) that the item is authentic, it will not accept the item from the seller D for auction. However, at lower prices, collectors and/or less reputable auctioneers will accept the item even if they are less certain of the authenticity ofthe item, since 0C < 6H and 6L Pc. Note that, when v = Pc, a collector will accept
the dealer's offer only if 0C = 1. Note also that, since {H) » (L), we must have 0*H > 0[ so that the curve 0*H (v) is always above 0*L (v). [Insert Figure 2 here]
In Figure 2, the curves 0*c=-- and 0*H =--^-^-- intersect uniquely at
v
aHv + (/)(H)
(v ,0 ) (Lemma 1(a)). On the other hand, the curves <9*=P-- and v ML) eL = --; 7777 do not intersect (Lemma 1(b)). These results are stated in Lemma aLkv + 0*H,Pc};
(ii) A={(^,v):6>>6';,and v>Pc};
15
(iii) C = {(0,v):0 > 0*c, and v> Pc}; and (iv) I={(Ј,v):(6?,v)eAnC}.
Lemma 1:
(a) The curves 0*c - -- and 0*H = ----- intersect uniquely at (v*,0*), where
v
aHv + 0(H)
v = PCJY(KH)-- and, 0_. = -Pfc = ^<^t-->{^H)-aH^P-^c . ^Furth, ermore, v. i.s a,lways
v
0(H)
greater than Pc for all values (H)>aHPc.
(b) The curves 0*c =-- and 0* = -- intersect uniquely at (v*,0*), where
v
aLkv+ (f)(L)
n. Pc v=
^^^
and 0 = --r =
----. Furthermore, v approaches
(L)-aLkPc
v
(L)
infinity when (L) approaches aLkPc. For (L) less than aLkPc, there is no
intersection point (v* ,0*) in the first quadrant of the Cartesian plane.
As Figure 2 shows, for Pc e (j-^--1-, -----), both C and A are subsets of A. kaL aH Lemma 1 establishes that the sets A, A, C and I are non-empty, and hence, justifies the existence of pooling and separating strategies for each player / (i e {H, L, C}), that is, Lemma 1 specifies the conditions on 0 as to when Player i will accept Ј)'s offer. In particular, C will accept any offer made by D if (0c, v) e C, while H and L will accept any offer made by D if (0H, v) e A and (0L, v) e A, respectively. These results are stated in Lemma 2.
16
Lemma 2: (a) C will acceptD's offer if (0c,v) e C. (b) //will accept Ј>'s offer if (0H, v) e A. 0 (c) L will accept Ј>'s offer if (0L, v) e A. Underselling by an Informed Dealer With reference to Figure 2, if all players H, L and C are sufficiently confident that the item offered to themis authentic (that is, 0. is sufficiently high such that 0t > 0*), then all of them will accept the item. In this case, sellerD will offer the item to the player who provides him with the highest payoff. Hence, D will offer the item to H, and the latter will accept the offer if (0,v) e (0H,v) e{A-l} and (\-aH)v>(\-aL)vL, or if (0H,v)eI and (\-aH)v>Max{Pc,(\-aL)vL} (Proposition 1(a)). In these cases, Us payoff will be the auction value net of the commission paid to the auction house H, that is, PH = (1 - aH )v. On the other hand, if H is not sufficiently confident that the item is authentic (that is, 0H < 0*H), then it will reject D's offer of the item for auction. The seller D will then offer the item to either the less reputable auctioneer L or the collector C, whoever is sufficiently confident that the item offered is authentic and can provide D a better payoff (Propositions 1(b) and 1(c)). These results are stated in Proposition 1, which identifies the equilibrium in terms ofZ)'s optimal choice of sales channel. Proposition 1: The Nash equilibrium in terms of Z)'s optimal choice of sales channel is as follows: 17
(a) D offers the item to H, and H accepts the offer, if (0H,v) e{A-l} and (l-a^v^Cl-aJv^orif (0tf,v)eland (1 - aH )v > Max{Pc, (1 - aL )vL }. D's payoff is PH = (1 - aH) v. (b) D offers the item to C, and C accepts the offer, if (0c,v)el and Pc>Max{(l-aH)v,(\-aL)vL},orif(0c,v)e {C - 1} and Pc > (1 -aЈ)vL. D's payoff is Pc. (c) D offers the item to L, and Z, accepts the offer, if (0L,v) e {A-A-C}. D's payoff is Pz =(l-aL)vL. Proposition 1 provides an explanation as to why an informed dealer D who knows that an item is an authentic antique, might sell it (to a collector C or via a less reputable auctioneer L) at a price (Pc or v^) that could be well below the item's worth v (that is, Pc :0 < Pc < v and vi'.O Proposition 2: (a) A more reputable auction house (H) is more likely to reject an item offered to it than C and L, even though the item is indeed authentic. (b) A less reputable auction house (L) almost always accepts the dealer's offer, if its reputation cost (L) is small. Hence, unless a reputable auctioneer is confident that an item is authentic with a very high probability (Proposition 1(a)), it will reject the item for auction even though the item may truly be an authentic antique. Corollary 1 therefore follows. Corollary 1: The rejection for auction of a supposedly antique item by a reputable auction house, or the experts representing it, does not imply that the item is not an authentic antique. When a reputable auctioneer rejects a supposedly antique item for auction, the dealer's next best alternative will be to offer the item to either a collector or a less reputable auctioneer, depending on who offers him the higher payoff since a dealer always prefers a trade to no trade at all. If the price for the item Pc is sufficiently low compared to its worth v, then a collectorwill be willing to take the risk and accept the item even though the probability that it is authentic 0C is low (Proposition 1(b)). Similarly, less reputable auctioneers will also accept items with lower probability of authenticity 0L (Proposition 1(c)), as their buyers are willing to accept such items provided that the auction prices vL are sufficiently low compared to their potential 19
worth v. As less reputable auctioneers will accept items with lower probability of authenticity, they therefore serve as a sales outlet for items that dealers fail to sell directly or via reputable auctioneers at higher prices (Proposition 2(b)). Corollary 2 therefore describes the best strategy for an informed dealer in selling an authentic antique. Corollary 2: A dealer, who knows that an item is truly authentic, will offer the item for sale via a reputable auction house and the latter will accept the item if and only if it believes that the probability that the item is authentic is sufficiently high. Otherwise, it will reject the item offered, even though the item is authentic. The dealer's next best alternative is to sell the item directly to a collector or via a less reputable auction house at a price less than its true worth. We therefore observe the phenomenon that even though a dealer knows with certainty that the item is truly authentic, he would still sell the item to a collector or via a less reputable auction house at lower prices Pc or vL , respectively, which may even be substantially lower than the worth of the item (v). Propositions 1 and 2 therefore imply that, less reputable dealers and auctioneers can coexist alongside reputable auctioneers. It follows that, a dual channel structure that comprises the auction and the dealer channels is sustainable even though authentic antiques are knowingly undersold by less reputable resellers. Corollary3 states this result. Corollary 3: A dual channel structure that comprises the auction and dealer channels is sustainable even though authentic antiques are knowingly undersold by less 20
reputable auctioneers and dealers at prices that could be substantially lower than those realized by reputable auctioneers, thatis, Pc :0 < Pc < v and vL: 0Max{Pc,(l-aH)v}), then the dominant strategy for any seller is to sell the item through less reputable auctioneers, as the latter are the least demanding in accepting items for auction. This result is stated in Proposition 3, which implies that if it were true that less reputable resellers can command prices that are no worse than those achievable by reputable auctioneers, then the dual channel structure that is observed in the market will collapse into a single sales channel comprising only less reputable resellers, while reputable auctioneers will cease to operate. Proposition 3: Suppose less reputable auctioneers are able to realize values that are no worse that the prices that dealers can obtain in selling direct or via reputable auction houses. At the equilibrium, only a single sales channel occurs in that D's dominant strategy is to offer the item only to L, andL accepts the offer, (a) if (0L,v)e {A-I}and (l-cxL)vL >(\-aH)v. D'spayoffis PL =(\-aL)vL; (b) if (0L,v)e I and (l-aL)vL >Max{Pc,(l-aH)v}. D'spayoffis PL =(l-aL)vL; 21
(c) if (0L,v)e {C-I}and (l-aL)vL >PC. D'spayoffis PL =(l-aL)vL. If the authenticity of a supposedly antique item can be verified with certainty (that is, 0i e {0, 1}), then all knowledgeable buyers (whether they are collectors, dealers, or auctioneers) will reject fakes and accept only authentic antiques. Hence, since the item offered by the.dealer is an authentic antique, both the reputable and less reputable auctioneers, as well as the collector, will accept the item. The dealer will therefore offer the item to the one who provides him the highest payoff. In this case, the dealer will always offer the item to the collector C, since C is willing to pay the price that equals the worth of the item (v) if the item is authentic with certainty (that is, 0cv = v when 0C = 1). By selling the item directly to C, D saves on the sales commission chargeable by the auctioneers, and hence, maximizes his payoffs. Proposition 4 states this result, which implies that, since a dealer will always sell directly to their collector clients, the auction channel will cease to exist if it were true that authenticity can be verified with certainty. Proposition 4: If authenticity is verifiable with certainty, that is, 0 e {0,1}, then, (a) If 0 = 0, all parties H, L, and C will reject the item. (b) If 0 = 1, all parties H, L, and C will accept the item. However, D will offer the item to C since 0cv> Max{(I- ccH )v,(1 - aL )vL } , when 0C = 1. Propositions 3 and 4 therefore imply Corollary 4, which describes the channel structure consequences if it weretrue that less reputable resellers can command prices that are no worse than reputable auctioneers, or if it were true that authenticity can be verified with certainty. 22
Corollary 4: (a) If stylized fact (1) is not true, that is, if less reputable resellers can command prices that are no worse than those achievable by reputable auctioneers, then only a single sales channel consisting of less reputable resellers will exist. (b) If stylized fact (2) is not true, that is, authenticity can be verified with certainty, then authentic antiques will be traded through a single sales channel only, whereby dealers sell directly to collectors, while the auction channel will cease to exist. Moreover, if authenticity can be verified with certainty and can be credibly communicated, then any seller is assured of a high price (v) for an authentic antique item by selling directly to a collector. Hence, any items traded by informed sellers at prices that are below v are fakes. It also follows that any supposedly antique items that are undersold by less reputable resellers are fakes, which implies that, being rational, collectors would buy only from reputable auctioneers and less reputable resellers (dealers and auctioneers) will cease to exist. However, if authenticity cannot be verified with certainty, then it follows from Proposition 1 that prices do not signal authenticity, since there is no cost to imitators in imitating a high price strategy. These results are stated in Corollary 5. Corollary 5: (a) A low price (below v) signals that an item is a fake (0 = 0) if and onlyif it is true that authenticity can be verified with certainty. Furthermore, if price signals 23
authenticity, then a single sales channel comprising of reputable auctioneers will exist, (b) A low price does not signal that an item is a fake if authenticity cannot be verified with certainty. Hence, from Corollaries 4 and 5, it can be inferred that any theoretical arguments against stylized facts (1) and (2) cannot be supported by the observations of a dual channel structure that comprises dealers and auctioneers with varying degrees of reputation, which continue to operate in the market. Furthermore, given that stylized facts (1) and (2) are true, it follows from Corollary 5 that a low price does not signal that a supposedly antique item is a fake. DISCUSSION & IMPLICATIONS Signaling Implications In this study on credence products, we show that uncertainty and non- verifiability of quality results in a dual channel structure that comprises the auction and dealer channels, even though underselling is practiced by less reputable resellers. In addition, we also show that a low price does not signal the lack of quality for a credence product. These results differ from those for experience goods, for which Akerlof (1970) showed that the market would collapse when quality is uncertain, but is verifiable post purchase, and that low price signaled the lack of quality. However, our model is in coherence with that of Akerlof s (1970) when we assume as a special case in Corollaries 4 and 5 that verifiability of authenticity is absolute. It cannot be simply assumed that the results obtained from prior studies on experience goods can be readily generalized to credence goods. 24
Our analysis might seem to suggest that a reseller's reputation could serve as a signal of quality for credence products. However, it is worthwhile to note that such a signal is not perfect. This is because the quality of credence goods, by definition, is not verifiable with certainty, and hence, the potential for moral hazard behaviors by sellers is always present. Although our analysis is in the context of antiques, the implications about price and reputation as signals of quality are applicable to a class of products (such as contemporary art objects and collectibles, health foods, and perhaps even gemstones) and services (such as automobile and equipment repairs, medical diagnoses, and consulting services), which possess quality characteristics that are not verifiable with certainty or are prohibitively costly to verify. Arbitrage Implications The dual channel market structure with resellers of varying degrees of reputation caters to buyers with different levels of product knowledge and risk aversion. Less knowledgeable and more risk averse buyers are likely to buy via reputable auctioneers at prices that reflect the true worth of the products. Reputable auctioneers perform the role of screening for quality to ensure a greater likelihood of high quality, and command a price premium for this role. However, it must be explicitly recognized that the screening performed by the reputable auctioneers is not perfect, given the nature of credence products. The reputation of an auction house may therefore be viewed as serving as a partial insurance against the risk of buying fakes. On the hand, more knowledgeable and less risk averse buyers are likely to be better off buying through less reputable resellers at lower prices. Less reputable resellers therefore perform the market-clearing role, whereby products are sold at lower prices that are commensurate with buyers' confidence of the products' quality. 25
The underselling of credence products by less reputable resellers provides opportunities for arbitrage by more reputable and knowledgeable resellers. For antiques, more reputable and knowledgeable resellers could acquire authentic pieces at low prices (from less reputable ones), then resell these items to clients, who are willing to pay higher prices just for the endorsement by the former of the authenticity of these items. For this same reason, more reputable and knowledgeable resellers could also resell the items, which they had acquired at low prices, via even more reputable resellers to obtain higher prices. The underselling phenomenon, which results in arbitrage opportunities, therefore leads to the observation made by Cooper (1985) that it is not unusual that a piece of antique passes from one reseller to another, with an escalation in price with each transaction until the item achieves its true value as exemplified by the author's reported case of the swan pendant. CONCLUSION In this paper, we propose that the difficulty in verifying quality with certainty leads to the observed phenomenon of underselling and of the dual channel structure that comprises resellers of varying degrees of reputation in the market for antiques. Given that quality cannot be verified with certainty, buyers and resellers therefore form subjective beliefs of the probability that an item is of high quality or authentic (0t). However, we do not investigate the factors that affect how buyers and resellers form their subjective beliefs, which is probably best examined empirically in future research. Future research to gain a more in depth understanding of the buying behavior for credence products is important, as current literature on this is scarce. Such an understanding would also allow managers to devise more effective risk 26
reduction strategies to overcome or reduce buyers' risk perceptions in purchasing credence products/services. Given that our aim is to identify a rational explanation for the underselling of authentic antiques, we developed and presented a game of uncertainty in our analysis, in which we considered the case where the item offered by the dealer is indeed authentic and assumed that all the players are honest and knowledgeable. This allows us to eliminate moral hazard and misrepresentation, and lack of knowledge, as alternative explanations for the observed underselling phenomenon. Future research could extend our study to examine the case when fake items might be traded through the various channels, and how buyers' and sellers' type (in terms of honesty and knowledge) and interactions among the various types, affect the price and distribution outcomes in even more sophisticated games of incomplete information. REFERENCES Adam, Georgina. 1995. Art - Sales: Guessing at ages. The Daily Telegraph March 20: 20. Akerlof, George. 1970. The Market for 'Lemons': Quality and the market mechanism. Quarterly Journal ofEconomics 84 (August) 488-500. Anderson, Erin and Barton Weitz. 1992. The Use of Pledges to Build and Sustain Commitment in Distribution Channel. Journal of Marketing Research 29 (February) 18-34. Anderson, Erin, Leonard Lodish and Barton Weitz. 1987. Resource Allocation Behavior in conventional Channels. Journal of Marketing Research 24 (Febrary) 85-97. 27
Athineos, Doris. 1996. Phony provenances shake the art world. Forbes 158 (4) 168- 170. Battles, Jan. 2000. Art world divided on Dublin's Bacon 'fakes'. Sunday Times (London) March 19. Bello, Daniel and David Gilliland. 1997. The Effect of Output Controls, Process Controls, and Flexibility.on Export Channel Performance. Journal ofMarketing 61 (January) 22-38. Boulding, William and Amna Kirmani. 1993. A Consumer-Side Experimental Examination of Signalling Theory. Journal of Consumer Research 20 (1) 111- 123. Bradach, Jeffrey L. and Robert G. Eccles. 1989. Price, Authority, and Trust: From Ideal Types to Plural Forms. Annual Review ofSociology 15 97-118. Brown, Christie. 1992. China Dolls. Forbes 149 (9) 178-180. Cadogan, Gerald. 1987. Diversions: bridge across the cultural divide. Financial Times (London) September 5: 14. Carlton, D. W. 1979. Vertical Integration in Competitive Markets Under Uncertainty. Journal ofIndustrial Economics 27 (3) 189-209. Cavusgil, S. Tamer and Shaoming Zou. 1994. marketing strategy-Performance Relationship: An Investigation of the Empirical Link in Export Market Ventures. Journal ofMarketing 58 (January) 1-21. Chu, Wujin and Woosik Chu. 1994. Signaling Quality by Selling Through a Reputable Retailer: An Example of Renting the Reputation of Another Agent. Marketing Science 13 (Spring) 177-189. Cooper, J. 1985. Dealing with Dealers: The Ins and Outs of the London Antiques Trade. Thames and Hudson, Great Britain. 28
Darby, Michael R. and Edi Kami. 1973. Free Competition and the Optimal Amount of Fraud. Journal ofLaw and Economics 16 67-88. Dutta, Shantanu and George John. 1995. Combining Lab Experiments and Industry Data in Transaction cost analysis: The Case of Competition as a Safeguard. Journal ofLaw, Economics, and Organization 11 (1) 87-111. Dutta, Shantanu, Mark Bergen, Jan Heide and George John. 1995. Understanding Dual Distribution: The Case of Reps and House Accounts. Journal of Law, Economics, and Organization 11 (April) 189-204. Emons, Winand. 1997. Credence Goods and Fraudulent Experts. RAND Journal of Economics 28 (1) 107-119. Emons, Winand. 2001. Credence goods monopolists. International Journal of Industrial Organization 19 375-389. Farrel, Joseph and Nancy T. Gallini. 1988. Second-sourcing as a commitment: Monopoly incentives to attract competition. Quarterly Journal of Economics 103 (4) 673-694. Frazier, Gary and Raymond Rody. 1991. The Use of Influence Strategies in Interfirm Relationships in Industrial Product Channels. Journal ofMarketing 55 (January) 52-69. Gallini, Nancy T. and Nacy A. Lutz. 1992. Dual Distribution and Royalty Fees in Franchising. Journal ofLaw, Economics, and Organization 8 (3) 471-501. Gerstner, Eitan. 1985. Do Higher Prices Signal Higher Quality? Journal ofMarketing Research 22 (May) 209-215. Heide, Jan and George John. 1992. Do Norms Matter in Marketing Relationship. Journal ofMarketing 56 (April) 32-44. 29
Heide, Jan. 1994. Interorgamzational Governance in Marketing Channels. Journal of Marketing 58 (January) 71-85. Hoh, Erling. 1998. Immaculate deceptions. Far Eastern Economic Review 161 (12) 36-38. Hughes, Robert. 2000. The auction house scandal. Time 155 (March 6) 56-58. John, George and Barton Weitz. 1988. Forward Integration Into Distribution: An Empirical Test of Transaction Cost Analysis. Journal of Law, Economics, and Organization 4 (Fall) 121-139. Joseph, Joe. 2000. You may not know art, but neither do the experts. The Times (London) January 26. Kirmani, Amna and Akshay R. Rao. 2000. No Pain, No Gain: A Critical Review of the Literature on Signaling Unobservable Product Quality. Journal ofMarketing 64 (April) 66-79. Kwang, Mary. 2000. Chinese peasants hawk fakes for fast cash. The Straits Times (Singapore) July 5:21. Lusch, Robert and James Brown. 1996. Interdependency, Contracting, and Relational Behavior in Marketing Channels. Journal ofMarketing 60 (October) 19-38. Lynch, James and Drue Schuler. 1990. Consumer Evaluation of the Quality of Hospital Services from an Economics of Information Perspective. Journal of Health Care Marketing 10 (2) 16-22. Mendick, Robert. 2000. Tell-Tale Sign that Pounds 40M Rubens Could be a Copy; Expert Reopens Debate on National Gallery Masterpiece. The Independent (London) May 21: 9. Nelson, Phillip. 1970. Information and Consumer Behavior. Journal of political economy 78 (March/April) 311-329. 30
Pitchik, Carolyn and Andrew Schotter. 1987. Honesty- in a Model of Strategic Information Transmission. American Economic Review 11 (5) 1032-1036. Pondy, Lou. 1967. Organizational Conflict: Concepts and Models. Administrative Science Quarterly 12 (September) 296-320. Sian, E. Jay. 2000a. Four more paintings under suspicion. The Straits Times (Singapore) October 4: 5 (Life). Sian, E. Jay. 2000b. Works are genuine. The Straits Times (Singapore) October 5: 3 (Life). Stem, Louis, Brian Stemthal and Samuel Craig. 1973. Managing Conflict in Distribution Channels: A Laboratory Study. Journal of Marketing Research 10 (May) 169-179. Tellis, Gerard J. and Birger Wemerfelt. 1987. Competitive Price and Quality Under Asymmetric Information. Marketing Science 6 (Summer) 240-253. The New York Times. 2001. Indictment names two ex-chairmen of auction houses. May 3: 1 (Business/Financial Desk). The Nikkei Weekly. 2000. Archaeologist's deceit unearthed. November 20. USA Today. 2000. Art discovers the Internet. January 10: 1A. Watson, Peter. 2000. Fake antiquities litter top museums. Sunday Times (London) December 10: Home news. Wolinsky, Asher. 1993. Competition in a market for informed experts' services. RAND Journal ofEconomics 24 (3) 380-398. 31
Stage 1
Figure 1: A Game of Uncertainty Sales Channel for Antiques
Stage 2 0t (**A)
{nL,pL)
Rejects Item (0,0) 9, Oc,Pc)
32
Figure 2: Graph of Indifference Curves 0(v) (0(H)>aHPc,0(L)Appendix
Proof of Lemma Ua): First, we find v such that -- = --------- . Solving, we have v aHv+0(H)
v* = Qz±--I-- which is always no lower than Pr, since ---- >1, i.e.,
0(H)-aHPc
0(H)-aHPc
the cost of reputation 0(H)>aHPc. Also, note that this critical value v* is
increasing in 0(H) and aH . Inparticular, when oris 0, v* is exactly equal to Pc.
PML)
Proof of Lemma 1(b): For v. =
to approach infinity, we must have
0(L)-aLkPc
0(L) very close to aLkPc.
Proof of Lemma 2(a): By our definition of C, the conditions that 0 > 0*c and v > Pc are satisfied for all elements of (0c,v) e C.
Proof of Lemma 2(b): For all elements of (0H, v) e A, 0 > 0*H and v > Pc.
Proof of Lemma 2(c): For all elements of (0L, v) e A, 0 > 0L and v > Pc.
Proof of Proposition 1: If (0H,v) e {A - I}, it is easy to check from Lemmas 2, 3 and 4 that only H and L will acceptD's offer. Hence, at the equilibrium,D offers the item to H if and only if (1 -aH)v>(l-aL)vL. On the other hand, if (0H,v) e I, then all of H, L, and C will accept D's offer. Hence, D offers the item to H if and only if (l-aH)v>Max{Pc,(l-aL)vL). If (0c,v) e I, all ofH, L, and C will accept D's offer. Hence, D offers the item to C if and only if Pc >Max{(\ - aH )v, (1 - aL )vL}. If (0C, v) e {C - I}, then only Cand L will accept D's offer. Hence, D offers the item to C if and only if Pc > (1 - aL )vL . Only Lwill accept the item for (0L,v)e {A-A-C}. Hence, the dealer offers it to L and gets a payoff of (1 - aL)vL.
Proofof Proposition 2: As 0(H) increases, 0*H tends to 1, which implies that the set A decreases and approaches an empty set. As 0(L) decreases and approaches 0, 0\ tends to 0. Proof ofProposition 3: Using the same arguments for Proposition 1(a), if (0L,v) e {A - I}, only H and L will accept D's offer. Hence, at the equilibrium, D offers the item to L if and only if (1 - aL )vL >(l-aH)v. Using the same arguments for Proposition 1(a), if (0L,v) e I, all ofH, L, and C will accept D's offer. Hence, D offers the item to L if and only if (\-aL)vL >Max{Pc,(\-aH)v}. If (0L,v) e {C - 1} only Cand Lwill accept D's offer. Hence, D offers the item to L if and only if (1 - aL)vL > Pc .
34
Proof of Proposition 4: For A (A e {H,L}) and C to accept an item, it is necessary that 0A>O and 0C >0, respectively. Hence, H, L, and C will reject the item if 0 = 0. Conversely, if 0 = 1, then//, Z, and Cwill accept theitem. D therefore offers the item to whoever (auction house or collector) offers him the higher payoff. Proposition 4 thus follows. 35

File: marketing-of-credence-products-a-game-theoretic-perspective-on.pdf
Published: Wed Oct 11 14:17:54 2017
Pages: 38
File size: 1.77 Mb


The Entrepreneurial City, 6 pages, 0.39 Mb

Art and the scalpel, 20 pages, 0.54 Mb

Elementary linear algebra, 7 pages, 0.44 Mb

How to think with your gut, 6 pages, 0.05 Mb
Copyright © 2018 doc.uments.com