Journal of Retailing 82 (2, 2006) 155–164
From Fatwallet to eBay: An investigation of online deal-forums and sales promotions Ram D. Gopal a,∗ , Bhavik Pathak a,1 , Arvind K. Tripathi b,2 , Fang Yin a,3 a
Department of Operations and Information Management, School of Business, University of Connecticut, Storrs, CT 06269-2041, USA b Management Science Department, Box 353200, University of Washington Business School, Seattle, WA 98195-3200, USA
Abstract Sales promotions are an important part of retail advertising strategy. Traditionally, research on sales promotions has generally assumed that the buyers are end consumers who do not engage in reselling, in large part due to high transaction costs. However, the recent Internet related technologies have dramatically lowered the cost of transferring goods between consumers, leading to relative ease of reselling activity amongst individual consumers. Little is known about the impact of this phenomenon on retailer’s sales promotion strategy. In this research we investigate the reselling activity in online auctions for products that active deal seekers can obtain at deeply discounted prices from retailers. We further investigate the role of deal-forums in the resale process. Data is collected from an online deal-forum (http://www.fatwallet.com/) and eBay to test various reselling-related hypotheses. The results show that there is a significant abnormal increase in the number of newly posted auctions of a product after the deal information of this product is posted on deal-forum website. We also find that there is a significant price incentive for individuals to resell. The implications for sales promotion research and practice are discussed. © 2006 New York University. Published by Elsevier Inc. All rights reserved. Keywords: Online deal-forums; Sales promotions; Advertising strategy; Online auctions; Virtual communities
Introduction Internet technologies have fostered the development of numerous online communities or groups, widely known as virtual communities. These virtual communities facilitate information-sharing among participants with specific interests and maximize the knowledge base of the participating entities. In this study, we examine those virtual communities that focus on sharing various deal information of consumer products that leads to lower purchasing price of an item. We will refer to these virtual communities as deal-forums. The deal information is valuable to all deal-forum participants including both those who have a need for the product and those who are interested in reselling products to make a profit. ∗
Corresponding author. Tel.: +1 860 486 2408; fax: +1 860 486 4839. E-mail addresses:
[email protected] (R.D. Gopal),
[email protected] (B. Pathak),
[email protected] (A.K. Tripathi),
[email protected] (F. Yin). 1 Tel.: +1 860 486 5295; fax: +1 860 486 4839. 2 Tel.: +1 206 221 5369; fax: +1 206 543 3968. 3 Tel.: +1 860 486 6182; fax: +1 860 486 4839.
For the potential resellers, another Internet phenomenon – online auction houses – provides a cost effective way to reach potential buyers without the restriction of space and time. This relatively new phenomenon – reselling activities among consumers facilitated by deal-forums and online auction houses – poses a new challenge for the sales promotion strategies of manufacturers and retailers. Theses resellers may buy discounted items in bulk to maximize their profits by reselling and thus, may defeat the objectives of sales promotions offered by firms. In this research we investigate the reselling activity in online auctions for products that active deal seekers can obtain at deeply discounted prices from retailers. We further examine the role of deal-forums in the resale process. Data are collected to test the hypothesis that there exists an abnormal increase in the supply of a product at online auction houses after the deal information of the same product is available on the deal-forum. The results of this research may have several important implications on firm’s sales promotion strategies. Blattberg and Neslin (1990) discuss the role of sales promotion objectives in sales promotion planning process and highlight important promotion objectives which include: (1)
0022-4359/$ – see front matter © 2006 New York University. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jretai.2006.02.002
156
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
generating store traffic, (2) moving excess inventory, (3) enhancing store image, and (4) creating a price image. In this research we introduce individual resellers as a new dimension for consideration into sales promotion analysis. This could potentially change some of the existing findings in the literature. Sales promotion literature has implicitly assumed that buyers of items under promotion are final users. Based on this, researchers have argued that sales promotion, especially price discounts, has some negative impacts such as causing stockpiling and purchase acceleration (Blattberg et al. 1981), and negatively affecting reference price and brand image (Grewal et al. 1998). However, if buyers of the item under promotion are resellers instead of final users, these negative impacts could be somewhat mitigated. In addition, studies on dynamic pricing have suggested that the Internet has provided a means for mass customization of sales promotions by targeting individual buyers on seller’s website (Kannan and Kopalle 2001). Our findings extend the above argument by suggesting that deal-forums could be used as another way to customize promotions to target specific buyer groups. Next, we discuss some background literature on virtual community and online auctions. Then we develop our hypotheses, the empirical model, and describe the data collection process. After presenting the empirical results, we discuss in detail the theoretical and managerial impacts of our findings.
Background Virtual community began as a spontaneous, noncommercial, and social event (Rheingold 1993) to provide the community members the opportunity to share their experiences, opinions and knowledge with others (Bickart and Schindler 2001), and has been studied extensively in the field of sociology and psychology. There is also an increasing discussion on how the economic potential of virtual communities can be exploited (e.g., Rosenoer et al. 1999). Hagel and Armstrong (1997) suggest that the greatest profit-making potential on the Internet centers on the development of virtual communities, which provide consumers with the ability to develop relationships, exchange information on specific topics, and buy and sell products. Balasubramanian and Mahajan (2001) suggest that a member of virtual community can derive monetary gain when, for example, purchasing products at a lower price using deal information, but most of the time can only derive utility from the satisfaction of contributing to virtual communities and gaining respect from other members. Interestingly, very little is known about opportunities for members of virtual communities to derive monetary gains through interaction with other consumers. This research attempts to fill this gap. One possible domain that is relevant in searching for ways for virtual community participants to realize economic gains is to examine the internal linkages between deal-forums and
online auctions. Online auction sites such as eBay.com have grown into a huge market place where millions of people conduct transactions worth of millions of dollars everyday. Hundreds of thousands of people engage in product selling via online auctions on a fulltime basis (Adler 2002). In addition to individual consumers, small businesses are also starting to jump on the online auction bandwagon (Carlton 2000). Interestingly, only a few large wholesalers and retailers, online or bricks-and-mortar, have moved to actively participate in online auctions (Adler 2002), plausibly to avoid creating conflicts between online auction channel and existing channels. The fundamental incentives to engage in reselling activity stem from the differential pricing strategies that sellers practice widely. Such price discrepancies can arise from the practice of price discrimination across customer or geographic market segments, or from periodic promotional activities. Significant reselling activity occurs in secondary markets of used goods, and this phenomenon has been widely studied. Arunkundram and Sundararajan (1998) explore the economics involved in an Internet-enabled electronic secondary markets for used goods and find that the secondary market has an adverse effect on the new goods seller who has a high market share and for products with high technological obsolescence rates. However, they also report that such electronic secondary markets have a positive impact on the consumer surplus in almost all situations. Lee and Whang (2002) investigate an electronic secondary market where reselling occurs between retailers, as opposed to individual consumers. They present a model with one manufacturer and two retailers, where the retailers can engage in reselling of excess inventory. They present an analysis of the impact of the secondary market on the supply chain, and present strategies for the manufacturer to enhance sales by effectively leveraging the secondary market. In contrast to the existing work, our research considers (a) reselling activity for unused items, (b) reselling undertaken primarily by individual consumers, and (c) when reseller has no pre-arrangement with or consent from the original retailer. Our central contention is that the emerging Internet-based technologies provide opportunities for individuals to procure items at a deep discount, and subsequently resell at online auctions at sufficiently higher prices. These technologies significantly lower the arbitrage costs for individuals to engage in reselling; online auctions lower the transaction costs, and deal-forums lower the search cost to procure information about product availability at deeply discounted prices.
Research hypotheses The Internet has had an enormous impact on marketing research and practice. One such impact is the increasing ease with which firms can dynamically change the prices of their products in the virtual value chain, either by updating the posted prices or by offering customized coupons targeting individual shoppers. The impacts of such dynamic pricing
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
practices on consumer behavior such as price search, trust, learning, and reference prices have been studied (Kannan and Kopalle 2001). Since dynamic pricing and promotions are becoming more prevalent in marketing practice in comparison to the traditional mass marketing approach, this phenomenon raises issues regarding the reach and impacts of deals offered by firms. Some of the promotions are not easily accessible to all consumers. When a seller offers a deal, the likelihood of a consumer taking advantage of the deal depends on (a) whether the deal information actually reaches that consumer and (b) whether that consumer has the desire and capability to conduct the purchase. With respect to a specific deal on a specific product, the following three groups of consumers are pertinent: (1) aware of the deal information but do not need the product; (2) need the product but not aware of the deal information; and (3) need the product and are aware of the deal information but cannot use the deal due to the restrictions on the redemption of the discount/coupon, or geographic and temporal constraints. The third category of consumers could be the result of many different situations such as: a deal is only offered in certain locality; or a deal involves lengthy rebate process; or a deal is only offered during a specific period of time. From the utility maximization perspective (Balasubramanian and Mahajan 2001), there exists incentive for the consumers in (1) to utilize the information they obtain to purchase the product and resell it to the consumers in (2) and (3) at such a price that both parties are better off by the transaction. Typically sellers in online auction market obtain their items from various sources such as manufacturers, wholesalers, retailers, other consumers, etc. Items offered in online auctions could be either used or unused. The supply of an item fluctuates on a daily basis and is likely to correlate with the supply of similar items. For example, the supply of a certain model of HP laser printer is likely to correlate with the supply of other HP laser printers since there exist common supply-affecting factors that decide, among other factors, the availability of both the former and the latter in online auctions, such as the overall demand for laser printers. In addition, it is also reasonable to assume that sellers of this specific printer are responding to the activities of sellers of similar printers in the market. All these contribute to the correlation among the fluctuations in the supply of similar items. If many participants take advantage of a deal that is offered on a deal-forum and start reselling via online auctions, one expects to observe an abnormal increase in the supply of the item offered in the deal. Here, we define abnormal increase in supply as the increase of the supply that cannot be explained by the fluctuations in the supply of similar items. In addition, if this abnormal increase is due to the reselling activity, one expects that the abnormal increase in unused items instead of used items. The magnitude of the abnormal increase in supply depends on, among other factors, depth of discount, popularity of the product, and the number of individuals participating
157
in reselling activities; which in turn depends on the number of individuals who are aware of the deal information. Deal-forums formed by savvy shoppers to exchange various money-saving tips provide a cost-efficient way to disseminate deal information to a large audience in a short period of time. Shoppers around the country post information and their experiences about deals, discounts and rebates, sometimes even pricing-errors, on these websites. Sometimes the deal information is posted even before the retailers officially announce it. Of course, consumers can obtain deal information from various other channels such as word of mouth, newspaper and store fliers. However, these channels do not provide the reach and speed offered by the deal-forums. Therefore, the posting of a deal on an online deal-forum is a signal that the deal information has reached a large audience. Reselling activity should follow subsequently if the deal is offered at a price level that provides sufficient incentives to resell. This leads to the following hypothesis: H1. Abnormal supply increase of an unused product on the online auction site is preceded by the posting of deal information of this product on the deal-forum. Alternate factors, besides reselling, can potentially cause abnormal increase in the supply at online auction channel as well. It is conceivable that the retailer itself could be engaged in both the deal-forums and subsequently in the online auction channel. However, this possibility could be ruled out by only considering online auctions conducted by individual consumers, and not the retailers. Although quite a few retailers have presence on eBay in the form of “eBay Stores” (e.g., Sears, Dell, and HP), their selling activities are minimal and are not a significant portion of their overall sales. And we have avoided these sellers in our data collection. Another possible confounding factor may be the obsolescence of the product under consideration, rather than the specific sales promotion offered by the retailer. This is particularly relevant for computer and consumer electronic products, which have high obsolescence rates. Obsolescence may lead owners of older models, who wish to upgrade to newer model, to dispose of the products at online auctions. One would expect such auctions to be for used items. Consideration of only unused items helps rule out this explanation for the abnormal increase in the number of auctions. Another concern regarding obsolescence is that the deal information of a product serves as a signal of pending obsolescence; therefore triggering hastened selling by current sellers through online auctions. However, the direction of the impact of this concern on the online auction activities is not clear since these sellers also have the incentive to hold back selling activity to avoid price competition. The products used in this study have only been released to market for an average of 6.6 months at the time of our data collection, which somewhat relieves the possible complication resulting from the issue of obsolescence. As discussed earlier, the viability of the reselling activity depends critically on economic incentives to engage in the
158
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
activity. From the perspective of the resellers/auctioneers, it is desirable that there exists a profit margin between the average auction price, which could be considered as the expected revenue from reselling, and the price at which the resellers/auctioneers obtained the product based on the deal information from the deal-forum. Hence, the following hypothesis:
product as the response to the overall supply quantity of similar products. Specifically, we assume that, without the event of information of a deal being disseminated on deal-forum, the new supply of a specific product i at time t, NSi,t , can be predicted from the new supply of all the products in the same general category as i, NSci ,t
H2. The average auction price of a product is higher than the discounted price that is posted on the deal-forum.
where εi,t is a random shock while αi and βi are productspecific parameters to be estimated. Here, the general category refers to those used by eBay, such as the general category of “Dell UltraSharp 1800FP 18 Flat Panel LCD” is “Flat panel/LCD monitor”. To avoid endogeneity, NSci ,t should exclude NSi,t . Estimates αˆ i and βˆ i can be obtained from ordinary least square (OLS) estimation using data from a period before the event. This regression model implies that sellers of a product will decide the supply of this product in reaction to the supply of the general product category that the product belongs to, NSci ,t , in addition to reacting to those factors that are idiosyncratic to this product, εi,t . As long as all the idiosyncratic factors are accounted for by the random shock item, the relationship between NSi,t and NSci ,t should hold. However, since the deal information is a new idiosyncratic factor that is not accounted for by the random shock item, the relationship that is captured by the regression model is expected to change after the event occurs. During the event window the abnormal new supply is defined as:
According to the basic economic principle of supply and demand, if the supply of a product increases abnormally, this should result in lower average price of the product on the auction sites, compared to the prices of auctions that end prior to the posting of the deal and the abnormal supply increase. Hence the following: H3. After the deal information of a product is posted on the deal-forum, average auction prices are lower than before.
Methodology To link the posting of discount/coupon information on the deal-forum to the possible increase in the supply on the auction site, we designed an empirical model in a way that bears close resemblance to an event study. Event study methodology has been adopted by numerous accounting and finance studies to examine the relation between a certain type of management events and the abnormal stock price/volume changes that are possibly caused by these events (McWilliams and Siegal 1997). Other applications of event study methodology include research in law and economics, although the price of the stock remains the focal variable of interest (MacKinlay 1997). Occasionally, event studies have been conducted on nonstock prices such as the price of real estate property (Colwell et al. 2000). We construct an empirical model to test whether the supply of a product on an auction site increases abnormally after the event of releasing deal information of the same product on a deal-forum. As defined in the previous section, the abnormality is relative to the normal fluctuation in the daily supply, which is correlated to the fluctuation in the supply of similar products. To describe the normal fluctuation in the supply of a product, we use a supply reaction function that is similar to the price reaction function used to model manufacturer reaction to competitors’ price changes (Kopalle et al. 1999). Industrial organization and marketing literature has studied the interaction among firms in terms of their responses to each other’s competitive price promotion activities. The approaches used in these studies are to model the activities of firms as responses to activities of every other firm (see e.g., Kadiyali et al. 1996; Kopalle et al. 1999; Leeflang and Wittink 1996). Following this approach, in the current context, we model the supply quantity decision of sellers of a
NSi,t = αi + βi NSci ,t + εi,t
(1)
ANSi,τ = NSi,τ − (αˆ i + βˆ i NSci ,τ )
(2)
where τ refers to individual date within the event window. The variance of ANSi,τ can be calculated as: 2 1 (NSci ,τ − NSci ) 2 Var(ANSi,τ ) = si 1 + (3) + 2 Ti Ti (NSc ,t − NSc ) t=1
i
i
where si2
is the variance of the OLS estimation, Ti is the total number of days in the estimation period, and NSci is the mean of the general category new supply over the estimation period (Greene 2000, p. 307). Accumulating the abnormal new supply over the event window, we get a measure of cumulative abnormal new supply (CANS) for each product: CANSi =
K
ANSi,τ
(4)
τ=1
Var(CANSi ) =
K
Var(ASi,τ )
(5)
τ=1
where K is the length of the event window. Following the standard practice of event study CANSi is assumed to be identical and independently distributed. This assumption implies that the causes of the abnormal supply are not related to each other among different products. This is a
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
reasonable assumption since the occurrences of these abnormal supplies spread over a period of time and across different product categories. Finally, the average CANSi across n products (ACANS) is: n
ACANS =
1 CANSi n
(6)
i=1
n
Var(ACANS) =
1 Var(CANSi ) n2
(7)
i=1
The null hypothesis H0 : ACANS = 0 can be tested using: ACANS Z= √ t(n − 2) Var(ACANS)
(8)
To show that the abnormal supply increase only occurs after but not before the event date, we plan to estimate (1) and calculate (8) using both real event date, when the deal information is posted, and hypothetical event date, from 1 to 5 days before the posting date, to see whether there is abnormal supply increase before the deal information is posted.
Data A deal-forum (http://www.fatwallet.com/) is used to identify the event of announcement of a specific deal. This is a popular website for deal information of consumer products (Pressler 2003). We first identify the product for which deal information is posted. We do not include the generalized discounts such as “30% off everything in store” but focus on deals on single products with clearly defined specifications. The related product is matched to the auctions on eBay. The number of newly posted auction items for the same product on a specific day is used as the measure of the new supply of this product on that day. The keyword “new” is used in the search to identify unused items. The new supply in the general category is measured similarly. Due to the limited availability of auction history data on eBay, the estimation of
159
the normal supply uses data up to 20–30 days before the deal is posted. Datum was collected during a period of a month in August 2003. As a result, 15 product deals were identified on Fatwallet and are matched with the auctions on eBay. These products are mainly computer-related or personal electronics and are listed in Table 1. We also collected the release dates of these products from http://www.zdnet.com/ when available. The price information including the deal price as well as the actual auction prices are also collected for these products to test hypotheses H2 and H3. The event window could be set equal to the period between the day of the initial posting of the deal and the day after which the deal expires. Alternatively, one can extend the event window to expiration date plus a certain number of days to compensate for the possible shipping time, lagging effect, or both. One implicit assumption of the above methodology is that the deal is unexpected. Notice that the beginning date of the event window could be earlier than the date that the deal is officially announced if, occasionally, the deal information has been leaked to the deal-forums prior to the official announcement of the deal (Lieber 2002). On the other hand, there is another implicit assumption implied by the event study methodology, which is that there are no confounding effects during the event window. The justification of this assumption necessitates using shorter event windows (McWilliams and Siegal 1997). We test the robustness of the results by employing different length of event windows ranging from 1 to 5 days.
Results and discussion The average CANS observed for the 15 products are shown in Table 2, along with results of the test for the significance using different length of event windows. The results in Table 2 show that the ACANS related to the posting of deal information on the deal-forum are positive and significant at the end of the event window ranging from 1 to 5 days. The magnitude of the ACANS ranges from seven auctions on the event day
Table 1 List of products Product
Posting date
eBay category
Release date
Epson CX5200 All in one printer Samsung ML 1710 Laser printer Yamaha YST MS50 Speakers Toshiba e335 Logitech Quickcam Pro 4000 Netgear MR814 Linksys WRT54G Samsung 151V Canon G3 Canon S400 Canon S50 Dell 1800FP Dell 1901Fp Fuji S5000
8/09/03 8/10/03 8/12/03 8/13/03 8/16/03 8/17/03 8/18/03 8/19/03 8/20/03 8/20/03 8/20/03 8/20/03 8/20/03 8/20/03
Printers > Multifunction all in one > Other Laser printers Video and multimedia > Multimedia > Speakers Handheld Units WebCams Wireless networking > Access points, Routers Wireless networking > Access points, Routers LCD/Flat panel > 15 Digital cameras > Compact Zoon > 4.1 to 5.0 MP Digital cameras > Compact Zoon > 4.1 to 5.0 MP Digital cameras > Compact Zoon > 4.1 to 5.0 MP LCD/Flat Panel > 18 LCD/Flat panel > 19 and Larger Digital cameras > Point and Shoot > 3.1 to 4.0 MP
10/15/2002 04/28/03 – 10/15/02 08/14/02 11/18/02 12/01/02 – 10/15/02 03/01/03 04/01/03 – 06/24/03 08/15/03
160
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
Table 2 Average cumulated abnormal new supply (ACANS) Window length (days)
ACANS (number of abnormal auctions)
t-statistic
1 2 3 4 5
7.07 13.28 21.27 30.97 33.35
6.85*** 9.08*** 12.53*** 16.40*** 16.27***
***
Two-tailed, degree of freedom = 13, p < 0.001.
to 33 auctions at the end of the 5-day event period. All these positive abnormal new supplies are significant at p = 0.001 level, providing strong support for H1. Moreover, the ACANS increases at a relatively constant rate as the length of window increases, which is expected since this is an accumulative measure. Furthermore, the same statistical analysis has been repeated using hypothetical event dates ranging from 1 to 5 days prior to the posting date of deal information. The results are listed in Table 3. These results show that the abnormal supply increase is consistently insignificant before the posting date. The abnormal supply increase after the posting date simply echoes the results in Table 2. To test H2, Z-tests are conducted on the mean price of auctions listed within 5 days after the event date against the deal price. It is expected that the mean price of auctions posted after the event date should be high enough for the potential reseller to make a profit, given the deal price. The results of this test are shown in Table 4. For 11 out of 14 products, the mean auction prices are significantly higher than the deal prices while only one product shows a mean auction price significantly lower than the deal price. For the remaining two products, the mean auction price is higher than the deal price
but the difference is not statistically significant. The margin ranges from 3 percent to 222 percent. Therefore, hypothesis H2 is partially supported. Finally, two sample t-tests are conducted on the mean price of auctions of each of 15 products listed before and after the event date to test H3. As many auctions as available before the event date and auctions within 5 days after the event date are used for these tests. The results are shown in the following Table 5. In general, the direction of the change in price level is inconclusive from these 15 products. The mean price drops significantly after the event date for seven products, increases significantly for two products, and does not change significantly for the remaining six products. Overall, the hypothesis H3 is not supported. In summary, the results of various statistical analyses show that there is a significant abnormal increase in the number of newly listed auctions of unused products after the deal information is posted on the deal-forum website. It appears that the deal-forums are effective in providing useful information to potential resellers. The mean price of auctions listed after the event date tends to be significantly higher than the deal price available to the deal-forum members. This difference in prices provides sufficient economic incentive to engage in reselling. After the posting of the deal information and the increase of the abnormal new supply, the mean price of auctions is not necessarily higher or lower than the mean price before the abnormal new supply. Our empirical evidence supports that there exists a temporal precedence between the postings of deal information on the deal-forum and the subsequent auction activity, and that there exists sufficient economic incentive for individuals to engage in reselling. There could be several reasons why hypothesis H3 is not supported. First, even though the abnormal increase in the number of new auctions is statistically significant, it might
Table 3 Average cumulated abnormal new supply before the event
Shaded cells: before event date (all t-statistics insignificant).
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
161
Table 4 Z-test of mean price of auctions listed after event date against the deal price Product
Epson CX5200 All in one printer Samsung ML 1710 Laser printer Yamaha YST MS50 Speakers Toshiba e335 Logitech Quickcam Pro 4000 Netgear MR814 Linksys WRT54G Samsung 151V Canon G3 Canon S400 Canon S50 Dell 1800FP Dell 1901Fp Fuji S5000 ***
Auctions listed after event date Mean
n
$ 87.25 $ 96.56 $ 35.82 $ 160.23 $ 64.71 $ 40.79 $ 85.90 $ 228.61 $ 514.38 $ 421.24 $ 459.62 $ 400.27 $ 508.38 $ 496.60
45 19 11 59 9 16 71 2 61 57 58 36 130 3
Deal price
Z
Margin (%)
$ 50.00 $ 30.00 $ 24.00 $ 199.00 $ 48.00 $ 25.00 $ 57.00 $ 200.00 $ 454.00 $ 372.00 $ 413.00 $ 396.00 $ 494.00 $ 415.00
10.744*** 40.009*** 3.407*** −17.687*** 7.663*** 8.240*** 37.137*** 7.355*** 7.607*** 14.086*** 6.985*** 1.357 8.975*** 1.464
75 222 49 −19 35 63 51 14 13 13 11 1 3 20
Two-tailed, p < 0.001.
not be large enough compared to the overall size of the auction market to create a significant impact on the overall price level of the auction market. The phenomenon of reselling through online auctions is still nascent and the impact it generates might not have reached a critical point to significantly change the price level. Second, the level of the impact is arguably related to the depth of the discount and the popularity of the product. If a discount is very deep and the product is popular, more people may buy and resell, therefore resulting in more auctions and lower prices. To see whether this is the case, we examined the number of views and replies that the original posting of the deal information received on the deal-forum website. These factors are indicators of the interest generated in the deal-forums by a particular sales promotion, and serve as a useful proxy for the popularity of a deal. A casual observation reveals that those postings that are associated with a significantly lower after-event mean auc-
tion prices (two products) received more number of views and replies than the rest. Although more rigorous analysis is needed, our preliminary analysis suggests that the price change is likely related to the popularity of the deal. Yet another reason for lack of support to H3 is related to the skillfulness of auction sellers. If the number of auctions is relatively low and there are significant number of adept sellers who can sell high, then that defeats our H3 as well. There are some theoretical and data-related limitations to this study. It could be argued that the abnormal increase in the number of auctions is simply due to the deal itself instead of the posting of the deal in the virtual community. To counter this argument, it is necessary to use a control group of products for which a deal is available but the deal information is not posted on the deal-forum. If there is no significant abnormal increase in the number of auctions listed after the deal is out, the hypothesis H1 can be more strongly supported.
Table 5 Two-sample t-tests for mean price of auctions Product
Epson CX5200 All in one printer Samsung ML 1710 Laser printer Yamaha YST MS50 Speakers Toshiba e335 Logitech Quickcam Pro 4000 Netgear MR814 Linksys WRT54G Samsung 151V Canon G3 Canon S400 Canon S50 Dell 1800FP Dell 1901FP Fuji S5000 * ** ***
p < 0.5 (t-statistic for one-tail). p < 0.01 (t-statistic for one-tail). p < 0.001 (t-statistic for one-tail).
Listed before event date
Listed after event date (within 5 days)
Mean
n
Mean
n
$ 76.83 $ 120.55 $ 64.75 $ 167.39 $ 58.46 $ 45.40 $ 116.10 $ 219.25 $ 534.81 $ 417.76 $ 462.57 $ 411.17 $ 526.23 $ 489.50
51 42 2 230 68 107 35 10 227 241 216 227 260 4
$ 87.25 $ 96.56 $ 35.82 $ 160.23 $ 64.71 $ 40.79 $ 85.90 $ 228.61 $ 514.38 $ 421.24 $ 459.62 $ 400.27 $ 508.38 $ 496.60
45 19 11 59 9 16 71 2 61 57 58 36 130 3
t-statistic
−1.928* 6.158*** 0.532 2.737** −2.669* 1.913* 5.460*** −1.311 2.195* −0.868 0.401 3.268*** 7.499*** −0.115
162
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
Table 6 Summary of survey Questions
Responses
Frequency of visit
Once or twice a year 3 (4%)
Once or twice a month 5 (7%)
Once or twice a week 10 (14%)
Everyday 54 (75%)
Deal-forum helps to find deal
Strongly disagree 3 (4%)
Disagree 0 (0%)
Neutral 3 (4%)
Agree 18 (25%)
Ever purchased the deal found in forum
No
Yes
2 (3%)
70 (97%)
Never considered it
Considered but found not worthwhile 10 (14%)
Considered and may try later
Yes
16 (22%)
24 (33%)
Ever resold items on eBay
22 (31%)
Strongly agree 48 (67%)
Sample size = 72.
However, it is difficult to assemble such a control group of products given the fact that the reach and the penetration of the Internet are so that it is virtually impossible for a valuable deal to circulate exclusively outside of the Internet. We do observe that the level of CANS is somewhat positively related to the level of interest in the deal, which can be measured by the number of views and replies received by the original posting. However, we need more data points to be statistically confident about the positive relationship.
Additional validation To further validate the possible relationship between online deal-forums and auctions, we conducted a smallscale survey on two deal-forums: fatwallet.com and anandtech.com. The survey asked the forum members to provide their opinions on the usefulness of deal-forums in identifying deals, their purchasing activities, and their intention to resell. We post the solicitation on two deal-forums and all the responses were electronically collected from volunteers in a period of 45 days. Among the 72 respondents, 70 have purchased deal items found on the forum and 24 (33 percent) had resold these items on eBay. Another 16 (22 percent) considered reselling even though they had not done so. Therefore, although it is impossible to track whether the sellers on eBay are the members of deal-forums, this survey does provide further support for the relation between dealforums and auction activities. The survey questions and the responses are detailed in Table 6.
Implications The results have significant implications for sales promotion research and practice. Since 1980s, substantial numbers of studies have been conducted to study the cause and effects of sales promotions. Sellers offer sales promotions for various reasons including shifting inventory cost from retailers to consumers (Blattberg et al. 1981), dealing with demand uncertainty (Lazear 1986), and price discriminating among
different segments of consumers (Narasimhan 1984), among others. Price promotions affect sales through mechanisms such as brand switching, repeat purchasing, purchase acceleration, and category expansion (Blattberg and Neslin 1990). Research on sales promotions has generally assumed that those who buy from the promotions are the end consumers, and as a result, various consumer behavior theories have been applied to understand why consumers respond to sales promotions. However, this study shows that some of these buyers may actually be resellers. Although the percentage of such resellers might still be low at this point, we have every reason to expect a growth that is as fast as that of the online auctions. Loosely speaking, the deal-forum creates a new segment of consumers who do not buy a product for consumption but to resell for a profit, utilizing the information and liquidity provided by the Internet. As this new phenomenon begins to gain traction, it could have a profound impact on the practice of offering sales promotion. For example, marketing literature has long recognized the existence of reference prices of consumers. Research also shows that consumers’ reference prices are likely affected by past price promotions (Kalwani et al. 1990; Levy et al. 2004). Therefore, too frequent price promotions might lower consumer reference prices and hurt brand images (Dodds et al. 1991). On the other hand, if many consumers purchase the promoted items from resellers, the negative impacts might be somewhat mitigated since the resellers will mark up the prices. Sales promotion literature also recognizes the dynamic impact of promotions on base line sales. Excessive price promotions might cause the base line sales to decrease and consumers price elasticity increase (Kopalle et al. 1999). Again this negative impact could be partially avoided by customizing the promotions to target resellers. Our analysis reveals that the potential reselling activities of unused items by consumers could have both positive and negative impacts on retailer’s sales promotion strategies, depending on the original motivation of the sales promotion. Shifting inventory cost from firms to consumers has been argued to be one of the motivation of sales promotions (Blattberg et al. 1981). From this point of view, the fast and timely dissemination of information by deal-forums and the
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
following reselling activities could have a positive impact on sales promotion strategies in a variety of market scenarios. For example, if a seller announces a deal for inventory clearance, deal-forums by spreading this information help in faster inventory clearance. Although the profit of the retailer could be affected by the reselling activities, this can be compensated by lower inventory cost. Further, deal-forums can also help retailers in shifting the promotion strategies, from costly mass advertising to limited or more focused target advertising. Frequent and large-scale price discount might lower consumer’s reference price level of the promoted products and could hurt the brand image. Recently, studies on dynamic Internet pricing suggest that firms can benefit from offering coupons dynamically on their websites to mass customize promotions (Kannan and Kopalle 2001). Along these lines, the reselling phenomenon identified in this paper can also be leveraged by firms to design “stealth” promotions that are not widely advertised and are only available to those who actively search for such information. Such promotions, besides enabling faster liquidation, may enable more flexibility in price-setting than if explicit and well-advertised promotions were offered. Retailers can thus liquidate inventory more effectively with little or minimal impact on the overall long-term pricing strategy. Online deal-forums may also have significant negative impact on retailer’s sales promotion strategies. For example, if a seller announces a deal to expand the customer base in one market (geographic area) but keeps the prices unchanged in other market, deal-forums by spreading this deal information provide an information premium to its members which may defeat the seller’s objective of expanding its market base and getting new consumers. Further, the consumers after obtaining the product at a low price may engage in arbitrage that could adversely affect the sales after the deal is expired. Thus, we observe that in this scenario, deal-forums have an adverse effect on seller’s sales promotion strategy. We recommend that in such scenarios, firms may benefit by devising strategies to discourage reselling by various means such as limited quantity and warranty. Some retailers are increasing their footprint in online auctions to shortchange the incentives for individuals to resell. For example, both Dell and Sears now operate as sellers on eBay. However, the auctions conducted by these retailers have been mainly for refurbished items. Also the auction volume continues to be low and these retailers are yet to establish a strong and high profile presence in the online auction houses. Reselling could also hurt retailer’s initiatives in crossselling and target marketing. A retailer’s attempt at using a discounted product as the “loss leader” to attract consumers and create retailer brand effect may become less successful. Target marketing may be impeded by the resellers who make it more challenging to reach the right segment of consumers through promotional deals. Reselling can also impact the practice of price discrimination. According to Tirole (2000), price discrimination is sustainable and profitable only when
163
the cost of transferring goods between consumers in different markets is high. As the cost of engaging in arbitrage is lowered due to the Internet technologies, it may result in diminished benefits from price discriminating across customer segments. As this phenomenon becomes more prevalent, it poses a major challenge for retailers attempting to design effective sales promotion strategies. For example, consumer response is one of the important factors determining the optimal frequency, timing, and duration of promotion (Silva-Risso et al. 1999). If the existence of significant number of resellers changes the dynamics of consumer response, it should be incorporated into the decision-making process. One possible direction for further study is to look into how firms can incorporate the reselling activities through online auctions into their promotion planning to design the promotion so as to deter possible reselling activities.
References Adler, Jerry (2002). “The eBay Way Of Life,” Newsweek, 139 (24), 50–59. Arunkundram, Ravi and Arun Sundararajan (1998). “An Economic Analysis Of Electronic Secondary Markets: Installed Base, Technology, Durability And Firm Profitability,” Decision Support Systems, 24 (1), 3–16. Balasubramanian, Sridar and Vijay Mahajan (2001). “The Economic Leverage Of The Virtual Community,” International Journal of Electronic Commerce, 5 (3), 103–138. Bickart, Barbara and Robert M. Schindler (2001). “Internet Forums As Influential Sources Of Consumer Information,” Journal of Interactive Marketing, 15 (3), 31. Blattberg, Robert C., Gary D. Eppen and Joshua Lieberman (1981). “A Theoretical And Empirical Evaluation Of Price Deals For Consumer Nondurables,” Journal of Marketing, 45 (1), 116–129. Blattberg, Robert C. and Scott A. Neslin (1990). Sales Promotion: Concepts, Methods, and Strategies. Upper Saddle River, NJ: Prentice-Hall. Carlton, Jim (March 16, 2000). “EBay Diversified To Meet Needs Of Small Firms”, in Wall Street Journal (Eastern ed.). New York, NY, B.12. Colwell, Peter F., Carolyn A. Dehring and Nicholas A. Lash (2000). “The Effect Of Group Homes On Neighborhood Property Values,” Land Economics, 76 (4), 615–637. Dodds, William B., Kent B. Monroe and Dhruv Grewal (1991). “Effects of Price, Brand, and Store Information on Buyers’ Product Evaluations,” JMR, Journal of Marketing Research, 28 (3), 307–319. Greene, W.H. (2000). Econometric Analysis 4th ed. Upper Saddle River, NJ: Prentice Hall. Grewal, Dhruv, R. Krishnan, Julie Baker and Norm Borin (1998). “The effect of store name, brand name and price discounts on consumers’ evaluations and purchase intentions,” Journal of Retailing, 74 (3), 331–352. Hagel, John and Arthur G. Armstrong (1997). Net Gain: Expanding Markets through Virtual Communities. Boston, MA: Harvard Business School Press. Kadiyali, V., N.J. Vilcassim and P.K. Chintagunta (1996). “Empirical analysis of competitive product line pricing decisions: Lead, follow, or move together?,” Journal of Business, 69 (4), 459–487. Kalwani, Manohar U., Chi Kin Yim, Heikki J. Rinne and Yoshi Sugita (1990). “A Price Expectations Model of Customer Brand Choice,” JMR, Journal of Marketing Research, 27 (3), 251–262. Kannan, P.K. and P.K. Kopalle (2001). “Dynamic Pricing on the Internet: Importance and Implications for Consumer Behavior,” International Journal of Electronic Commerce, 5 (3), 63–83.
164
R.D. Gopal et al. / Journal of Retailing 82 (2, 2006) 155–164
Kopalle, Praveen K., Carl F. Mela and Lawrence Marsh (1999). “The Dynamic Effect of Discounting on Sales: Empirical Analysis and Normative Pricing Implications,” Marketing Science, 18 (3), 317– 332. Lazear, Edward P. (1986). “Retail Pricing And Clearance Sales,” American Economic Review, 76 (1), 14–32. Lee, Hau and Seungjin Whang (2002). “The Impact Of The Secondary Market On The Supply Chain,” Management Science, 48 (6), 719– 731. Leeflang, Peter S.H. and Dick R. Wittink (1996). “Competitive Reaction Versus Consumer Response: Do Managers Overreact?,” International Journal of Research in Marketing, 13 (2), 103–119. Levy, M., D. Grewal, P.K. Kopalle and J.D. Hess (2004). “Emerging Trends in Retail Pricing Practice: Implications for Research,” Journal of Retailing, 80 (3), XIII–XXI. Lieber, Ron (November 21, 2002). “Savvy Shoppers Get Sneak Peeks At Holiday Sales—Web Sites Post Secret Details On After-Thanksgiving Bargains, But Big Retailers Fight Back”, in Wall Street Journal (Eastern ed.). New York, NY, D.1.
MacKinlay, A. Craig (1997). “Event Studies In Economics And Finance,” Journal of Economic Literature, 35, 13–39. McWilliams, Abagail and Donald Siegel (1997). “Event Studies In Management Research: Theoretical And Empirical Issues,” Academy of Management Journal, 40 (3), 626–657. Narasimhan, Chakravarthi (1984). “A Price Discrimination Theory Of Coupon,” Marketing Science, 3 (2), 128–146. Pressler, Margaret Webb (April 6, 2003). For A Free Hassle, Claim Your Rebate, in Washington Post. Washington, D.C. (Financial section), H06. Rheingold, Howard (1993). The Virtual Community: Homesteading On The Electronic Frontier. Reading, MA: Addison-Wesley. Rosenoer, J., D. Armstrong and J.A.R. Gates (1999). The Clickable Corporation. New York, NY: Free Press. Silva-Risso, J.M., R.E. Bucklin and D.G. Morrison (1999). “A Decision Support System for Planning Manufacturers’ Sales Promotion Calendars,” Marketing Science, 18 (3), 274–300. Tirole, Jean (2000). The Theory Of Industrial Organization. Boston, MA: MIT press.