Contemporary Research in E-Marketing_3 docx

Since our study involved vehicles with varying model types e.g., DX, LX, EX,mileage, and options, the price ratio is the primary dependent variable of interest.This ratio serves as perce

rate their partner by leaving either a positive, negative, or neutral rating, alongwith a comment of up to 80 characters in length (see Figure 1) As a memberaccumulates feedback, a user rating is calculated with each positive commentearning +1 points, each neutral comment earns +0 points, and each negativecomment earns –1 points (eBay Feedback Forum, 2004) This rating and thepercentage of feedback rated positively are prominently displayed next to theuser’s ID (see Figure 2) Though not required, participation levels at eBay areremarkably high as buyers leave feedback on sellers 52.1% of the time andsellers on buyers 60.6% of the time (Dellarocas, 2003) Once left, a commentcannot be edited and becomes a permanent part of the feedback profile Thus,

Figure 1 eBay feedback form

Figure 2 eBay feedback rating

a negative or even a neutral rating can be detrimental to the user’s ability to sell

in the eBay community in the future

eBay Motors auction listings contain fundamental bidding data such as thewinning bid, ending date and time, number of bids, and so forth The listings alsoallow sellers to provide a formatted Web page that describes the vehicle anddisplays multiple pictures that can be enlarged to show details The descriptionand pictures of the vehicle are very important in overcoming the limitations of anonline automobile marketplace A previous study (Sena, Heath, & Webb, 2004)suggests that the quality of an auction’s description might impact the finalwinning price of the auction

It is important to note that, like traditional eBay auctions, eBay Motors utilizesproxy bidding (users specify their highest price and the system automaticallyincreases the winning bid when necessary), which means that the final bid price

is typically determined by the second highest bidder For example, if the winningbidder listed $10,000 as the highest amount willing to pay and (ultimately) thesecond-highest bidder placed a bid of, say, $9,500, then the winning bid pricewould be $9,600 (the second place amount plus an increment of $100) In otherauction listings, the winning price could be determined by a “buy it now” priceset by the seller which, if selected by a bidder, ends the auction immediately

Research on Internet Auctions

With the success of eBay, a number of studies have examined various measures

of reputation on the likelihood of successful sales occurring, and in particular, onthe final prices for goods sold at online auctions (Sena et al., 2004) Table 1summarizes the results of various studies that have examined the impact offeedback on ratings Such studies have yielded conflicting results as to therelationship between reputation and winning bid prices on eBay For details onprior research, see Dellarocas (2003)

The Internet Auto Market

eBay Motors (2004), a division of the online auction site, introduced used carbuyers and sellers to their bidding process with a category dedicated to cars in

1999 eBay Motors was started as a separate division in April 2000, with sales

of $1.5 billion in cars and parts in its first full year (Wingfield & Lundegaard,2003) In 2002, it sold 300,000 vehicles, while attracting more than 6.1 millionunique visitors in the month of February Total sales for 2002 represented 25%

of eBay’s gross merchandise (Cuneo, 2003a) Sales volume increased to 500,000

per month by 2003 and was expected to reach 1 million per month by 2004.Revenues have been forecast to reach $3 billion for 2005, potentially qualifyingeBay Motors for Fortune 500 status (Verma, 2003) While initial listingsconcentrated more on exotic and high-end vehicles, according to Simon Rothman,originator of eBay Motors and vice president of eBay’s U.S operations, carssuch as the Ford Taurus and Honda Accord top the sales list (Cuneo, 2003b).While eBay Motors has emerged as the leader in Internet car sales, AutoByTel(2004) introduced online car buying to the general public in 1995 While initiallyfocusing on new car sales along with CarsDirect (2004), they have both morerecently entered the used car market AutoTrader (2004) began exclusively as

an online used car dealer as AutoConnect in 1998 It now lists more than 2 millionused vehicles from private owners and dealers Cars.com (2004) also launched

in 1998 by pulling used “vehicle listings from thousands of dealer inventories andclassified ads nationwide.”

Selling cars on the Internet also has its drawbacks Online sellers have to contendwith frugal buyers searching for a bargain, possibly leading to a lower sale price.While this lower revenue may be offset by reduced costs for dealers, along with

Table 1 Prior research on impact of eBay feedback on winning bid price (Adapted from Sena et al., 2004)

Negative Feedback Effect on Winning Bid Price

Houser & Wooders (2000) – computer chips

Kalyanam & McIntyre (2001) – PDAs

Lucking-Reiley, Bryan, Prasa, &

Reeves (2000) – coins Melnik & Alm (2002) – gold coins Standifird (2001) – PDAs

Livingston (2002) – golf clubs

Net score increases price Cabral & Hortacsu (2003) – coins, Beanie Babies, and laptop

computers Dewan & Hsu (2001) – stamps McDonald & Slawson (2002) – dolls Sena, Heath, & Webb (2004) – designer watches and DVDs

quicker sales for both dealers and private owners, according to estimates by theGoldman Sachs Group, only “about 30% of auto listings on eBay close with awinning bid” (Wingfield & Lundegaard, 2003).

Chip Perry, president of AutoTrader, notes that his company research showsthat online used car sales “are inherently limited by the fact that consumers arereluctant to make purchases sight unseen” (Cuneo, 2003a) His site has recentlyentered the auction car sales market as direct competitor to eBay, and offers a

“conditional bidding” process where the winning bidder is not obligated to buyuntil the car’s condition has been verified by an inspector eBay also makes aspecial effort to “build trust, confidence and support to both buyers and sellers,”

by insisting on ethical behavior Feedback about both the seller and buyer arereadily available, and a strict set of “rules” govern transactions For instance,

“eBay will throw out a seller who regularly receives negative feedback”(Piszczalski, 2003) Most vehicles on eBay come with protections such aspurchase insurance at no extra cost Cars that are never delivered or misrepre-sented are insured for up to $20,000 These extra efforts by online marketersseem to have had an influence on the car-buying public While many shoppersstill choose to buy locally, three-quarters of all car sales on eBay involve out-of-state transactions (Wingfield & Lundegaard, 2003)

For the buyer, online vehicle sales seems to be a shopper’s mecca At any givenmoment, a shopper may find 20,000 cars listed just on eBay Motors (Fahey,2003) Hundreds of choices for a given car model, such as Honda Accord, may

be available at any given time With multiple search options, buyers have theultimate flexibility in comparison shopping They also have a wealth of informa-tion about the vehicle immediately available, and may contact the seller forfurther details for clarification

Still, as with used car buying in general, some shoppers are happy and some arenot Reports of misrepresentation and fraud occur for online sales as well as forthe stereotypical used car lot Some dealers who have tried online sales have alsobeen disappointed and Internet car sales have not yet had a serious impact ontraditional sales Although a few dealers are changing their way of business,moving from the traditional car lot to exclusive online sales, only 0.6% of the 43million used cars sold annually are sold on eBay Motors (Wingfield & Lundegaard,2003)

Research Questions and Methodology

From February through August of 2004, 126 observations were collected fromcompleted eBay Motors auctions Our data include only auctions offering Honda

Accords made between 1992 and 2003 with winning bid prices between $4,000and $20,000 Data were only collected on completed auctions in which the

“reserve price” (minimum seller is willing to accept) was met and in which theautomobile is described as being in good condition Autos that had been damaged,salvaged, or customized were not considered

Using the data (model type, year, mileage, options, etc.) from each auction listing,

“blue book” values were collected for each vehicle using the Kelley Blue Book

(2004) Web site (kbb.com) If the necessary data were not included in the listing(model type, options, etc.), the observation was not included in the data set (see

Figure 4 for an example of a Kelley Blue Book retail price listing).

Since our study involved vehicles with varying model types (e.g., DX, LX, EX),mileage, and options, the price ratio is the primary dependent variable of interest.This ratio serves as percentage of retail value that an auction listing achieved.For example, if an auction’s winning bid price was $7,000 and the automobile’s

retail value (as determined by using the Kelley Blue Book price) was $10,000,

the price ratio would be 70%

Based on the variables shown in Table 2, some interesting research questionsemerge Many of these research questions help to explore the role of risk in eBayMotors auctions In some eBay markets, more expensive items could sell for alower percentage of retail value For example, Sena et al (2004) found that theretail value of DVDs was negatively correlated with the price ratio (percent ofretail value) However, in the case of automobiles, given a fixed model type(Honda Accords), more expensive (or newer) models may be considered lessrisky and thus may realize a higher price ratio

Table 2 Description of variables

Winning Bid Price – Includes only completed auctions where bid price exceeds “reserve price” (the

minimum price specified by the seller)

Blue Book Value – Retail value of automobile as listed by Kelley Blue Book (kbb.com)

Price Ratio – The ratio of (Winning Bid Price/Kelley Blue Book Value)

Year – Model year of the automobile

Seller’s Feedback Rating – Number of completed auctions in which seller was rated as positive (serves

as an estimate of seller experience)

Seller’s Percent Positive – Number of positive feedback ratings divided by the total number of

feedback ratings (positive, negative, and neutral)

Buyer’s Feedback Rating - Number of completed auctions in which buyer was rated as positive (serves

as an estimate of buyer experience)

Number of Pictures – Number of unique images that users can access within the auction listing

(commonly presented as “thumbnail” photos that can be enlarged to show detail)

Dealer – Whether the listing indicates that the seller is an automobile dealership or an individual seller Bids – Number of bids placed during the auction (a “1” bid auction may indicate a “buy it now” auction;

eBay uses “proxy bidding” in which bids are automatically submitted by the system when a bid exceeds the current price but is below a prior bidder’s maximum price)

To examine these factors from different perspectives, we have focused on ninespecific research questions as shown in Table 3 Questions 1–3 focus on therelationship between price ratios and winning bid prices, retail values, and the age(model year) of the autos Research question 4 focuses on the impact ofautomobile dealerships on bid prices Beyond the perception that dealers may beless likely to commit fraud (perhaps because users have a name, address, etc.),they may also have the ability to offer services, warranties, and so forth, that mayentice buyers to offer higher bids.

Research question 5 explores the relationship between the number of bids at anauction and the winning bid price while research question 6 examines whetherlistings that include more pictures realize higher prices This research questionbuilds on a finding from Sena et al (2004) that higher quality descriptions (fordesigner watches and DVDs) resulted in higher bid prices (see Figure 3 for anexample of a listing with 28 thumbnail photos)

Prior research has indicated, with some exceptions, that seller feedback lates positively with winning bid prices As described in Table 2, two sellerreputation variables were collected from eBay listings: seller percent positiveand seller feedback rating The feedback rating serves as a measure of theseller’s experience, as estimated by the seller’s number of previous feedbackresponses These variables are generally the only measures of seller reputationthat eBay buyers observe as they are displayed on the main auction listing.Research questions 7 and 8 examine whether seller feedback ratings have animpact on winning bid prices It is important to reiterate that our sample includesonly completed auctions, excluding auctions where bid prices did not exceed theseller’s reserve value Thus, it is possible that seller feedback plays an important

corre-Table 3 Research questions

Research Question 1: Do autos with higher winning bid prices sell for a higher percentage of retail

value?

Research Question 2: Do more expensive autos (those with higher blue book values) sell for a higher

percentage of retail value?

Research Question 3: Do autos with more recent model years sell for a higher percentage of retail

value?

Research Question 4: Do autos listed by dealerships sell for a higher percentage of retail value (as

compared with those listed by individual sellers)?

Research Question 5a: Do auctions with more bids sell for a higher percentage of retail value?

Research Question 5b: Do auctions with one bid (i.e., “buy it now” auctions) sell for a higher

percentage of retail value?

Research Question 6: Do auction listings that contain a greater number of pictures sell for a higher

percentage of retail value?

Research Question 7: Do autos listed by sellers with higher feedback scores (i.e., more experienced

eBay users) sell for a higher percentage of retail value?

Research Question 8: Do autos listed by sellers with higher percent positive feedback sell for a higher

percentage of retail value?

Research Question 9: Do autos purchased by winning buyers with higher feedback scores (i.e., more

experienced eBay users) sell for a lower percentage of retail value?

Figure 3 Example of “thumbnail” photos in eBay Motors listing

Figure 4 Example of Kelley Blue Book listing

role beyond what our study captures For instance, the seller feedback (or lackthereof) may result in fewer or lower bids that fail to meet the seller’s minimumacceptable price.

Finally, research question 9 examines the role of feedback ratings for the buyerrather than the seller The seller feedback is an estimate of buyer experiencewith eBay From our anecdotal observations, it appears that many buyerspurchase multiple vehicles on eBay Motors, presumably with the intention ofreselling This variable, if significant, would likely be negatively related to priceratio, as one would expect more experienced users to recognize better deals (andthus realize lower price ratios)

Statistical Analyses and Findings

Descriptive Statistics

To begin our analysis, we examine the descriptive results of our data set Asshown in Table 4, the mean winning bid price for the 126 automobiles in oursample was $8,765, while the mean retail value of these automobiles was

$12,092, resulting in a mean price ratio of just over 72% The authors collecteddata for listings with model years ranging from 1992 to 2003 with a mean year

of 1998.75

Compared with other eBay marketplaces, buyers and sellers seem to have fewerfeedback ratings Buyers in our sample have an average of 17.83 feedbackratings while sellers have an average of 177.49 Like other eBay markets,feedback tends to be heavily positive with sellers in our sample having a meanpositive feedback percentage of 97.43% It is important to note that eBaycombines all feedback into one rating regardless of whether the user was a buyer

or seller and whether the item was sold on eBay Motors or another eBay listing.Thus, feedback scores and percent positive ratings can occasionally be mislead-ing (e.g., a rating based on Beanie Baby purchases rather than auto sales).Given the limitations of using eBay Motors for such an important purchase, theauction listing plays an important role in marketing the auto and conveying theimportant information that potential buyers require Thus, it is not surprising thatsellers provide numerous digital images in most listings In our sample, the meannumber of pictures provided was 18.59 Automobiles offered by dealerships may

be considered less risky by some eBay users In our sample, 71% of the sellerswere deemed to be automobile dealerships based on the item description Thenumber of bids on automobile auction may vary depending on the starting

Table 5 Correlation between price and age of auto and price ratio

(minimum) bid price, the reserve price, and whether the seller offers a “buy itnow” option In our sample, auctions had a mean of 19.84 bids, with 13 auctionsending after just one bid

Research Questions 1-3

As shown in Table 5, the correlation between price ratio and the winning bid price

is very strong while the correlations between price ratio and blue book value andyear are positive but insignificant in our sample However, as shown in Table 6,

a test of mean differences at selected values show that there may still be somerelationship between these variables This suggests that perhaps the relation-ships are not linear For example, in the case of model year, perhaps buyers arewilling to pay a higher percentage of retail for a recent (and presumably moretrouble-free and less risky) car, but the relationship fails to hold once cars reach

a certain age

1 Excludes six observations with zero feedback ratings

Table 4 Descriptive statistics (n=126)

Variable

Winning Bid Price $4,050 $18,900 $8,765.50 3421.31

Blue Book Value $5,775 $21,175 $12,091.83 3946.19

Seller’s Feedback Rating 0 8856 177.49 802.01

Seller’s Percent Positive 1 80% 100% 97.43% 4.00

Table 6 Mean differences among price and age of auto and price ratio

Winning Bid Price >= $9,000 (n=52)

Winning Bid Price < $9,000 (n=74)

76.3%

69.1%***

Blue Book Value >= $14,000 (n=40)

Blue Book Value < $14,000 (n=86)

*** significant at p<=.01; ** significant at p<=.05; * significant at p<=.10

Variable Correlation With Price Ratio

Dealership (n=89) Individual Seller (n=37)

in private sellers and are willing to pay a higher price under certain stances

circum-Research Question 5

As shown in Table 8, there was zero correlation between the number of bids andthe percentage of retail value earned in our sample The data seem to indicatethat perhaps auctions with a single bid (indicating the likelihood of a “buy it now”purchase) result in higher price ratios However, given the small sample size, thisdifference in means is not statistically significant, leaving this as an item forfuture study

Research Question 6

Table 9 reveals, in our opinion, the most interesting finding of this study Whilethe correlation between the number of pictures and price ratio is somewhat weak(with a p-value of 07), one would expect that this relationship would probablynot follow a linear pattern That is, if an auction includes very few pictures, thismay increase the perceived risk and result in a lower price However, at somepoint, additional pictures probably do not return the same marginal benefit In oursample, there was a large number of listing that included 12 pictures (perhapsfrom a template offered by eBay) These listings and those that included fewerpictures earned on average nearly 6% less of retail value compared with listingsthat include 13 or more pictures It is very likely in the near future that multimediapresentations with video or panoramic images will become common on eBayMotors and other sites demonstrating used vehicles

Research Questions 7-9

In our initial analysis, as shown in Table 10, it is somewhat surprising thatfeedback does not play a substantial role in determining price ratios None of thecorrelations between feedback variables and price ratio were statisticallysignificant While Table 11 shows some moderate differences in mean price ratio

Table 9 Relationship between number of pictures and price ratio

Variable Correlation With Price Ratio

Number of Pictures >=13 (n=66) Number of Pictures <=12 (n=60)

74.9%

69.0%*

*** significant at p<=.01; ** significant at p<=.05; * significant at p<=.10

Table 8 Relationship between number of bids and price ratio

Variable Correlation With Price Ratio

among selected subsections of the data, these are also not statistically cant Of course, feedback may still play an important role in a buyer’s decision

signifi-to bid on a particular vehicle or on vehicles that fail signifi-to result in a sufficient bidprice to meet the seller’s minimum (reserve) price However, our data set fails

to show substantial relationships between eBay’s feedback and winning bidprices (as compared with the respective retail value)

In an attempt to further explore the role of seller feedback (in particular, thepercent positive variable), we selected a subset of the data using only observation

in which the seller had been rated at least 25 times In sellers with very fewfeedback details, the percent positive is likely not as meaningful to prospectivebuyers The results presented in Table 12 show that perhaps when buyers

Table 11 Mean differences among price and age of auto and price ratio

Table 12 Relationship between percent positive and price ratio: Limited to auctions listed by sellers with a minimum of 25 feedback ratings

Table 10 Correlation between price and age of auto and price ratio

Variable Correlation With Price Ratio

Seller’s Percent Positive >= 98% (n=75)

Seller’s Percent Positive < 98% (n=45)

Seller’s Percent Positive >= 98% (n=43)

Seller’s Percent Positive < 98% (n=31)

74.7%

69.1%**

*** significant at p<=.01; ** significant at p<=.05; * significant at p<=.10

observe an adequate number of feedback ratings, then the percent positive ratingdoes play a role in the amount they are willing to bid for an automobile Althoughthe correlation is still not statistically significant in this subset, a comparison ofmeans among subgroups with greater than 98% positive feedback ratings versusthose with less than 98% positive feedback shows a statistically significantdifference in price ratio.

Conclusion

The principal findings of our study may be of interest to both practitioners andscholars of various disciplines Our results provide an empirical basis for futurestudies and reveal several research questions that can be probed in greater detailusing additional methodologies and more extensive data sets

While numerous studies have analyzed eBay exchanges, this study has thepotential to be among the most significant because of the importance ofautomobiles in our economy While our data set was limited, it captured over $1.1million worth of transactions The results provide a starting point for academicassessment of this exciting and important market

The results of this analysis promote an understanding of the factors that impactbid prices of automobiles in Internet auctions For example, our findings revealthat the price of the automobile and the inclusion of numerous pictures may play

an important role in predicting the percent of retail value that an auction listingwill achieve

The study also adds to the growing body of literature focused on the impact ofInternet-based reputation systems While the relationships between feedbackratings and price ratios in our data set were not statistically significant (contrary

to some past studies), more studies are needed to further explore this ship Our analysis does point out that the relationships between the variables inour study may not follow a linear pattern Thus, there is an opportunity forresearchers to conduct more robust statistical analyses on data sets of thisnature

relation-Although the market for automobiles on the Internet, particularly on eBayMotors, has exploded in the past year, the marketplace is still in its infancy.Consumer habits are likely to adjust over time as sellers learn to use the mediummore effectively and buyers become more comfortable with the marketplace.Similarly, advancements in technology and new business ventures will undoubt-edly continue to play a role in these exchanges This study provides a cursoryanalysis of the eBay Motors marketplace as it currently exists for the data we