buyer after the completion of the sale in mid-October, yet the buyer did not reciprocate. The
buyer has subsequently left feedback for transactions with other eBay users, however.
is a sophisticated user who used the feedback market to artificially boost
his online reputation to appear experienced in the market for a high-valued good. A back-of-the-
envelope calculation illustrates the cost of becoming “experienced”. Consider the extreme case
where all of his feedback was acquired by penny sales—assuming that 70 percent of his listings
resulted in sales,
could have create his feedback profile for approximately $360.
While we cannot know whether
is a good or bad seller, it does seem clear that the
market for feedback has allowed
to create a false sense of experience and,
5 Statistical Study of Market for Feedback
While the case study above is suggestive of the relationship between the market for feedback and
a seller’s other activities, it is useful to examine the market for feedback in greater detail. To this
end, we obtained data for over 6,500 listings in the market for feedback over the period from
June to December 2005. Below, we describe the data collection procedures and a statistical
analysis of key features of the market for feedback.
5.1 Description of the procedures for the study
Using a custom-designed computer script, we gathered data on eBay’s market for feedback on a
weekly basis. The automated script was designed to query the set of terms listed below, and then
retrieve data from completed auctions.
Based on manual searches we conducted in the market
for feedback, we found that the following search terms successfully captured the vast majority of
transactions in this market: “build feedback, eBay reputation, eBay rating, ebook feedback, free
feedback, free joke, free jokes, free riddle, get feedback, gmail feedback, gmail rating, golf shots
hr, increase feedback, new to eBay, ebook money, positive feedback, positive reputation, riddle
no, rocket feedback.” We also used the misspelled search term “postive feedback” since
misspellings in subjects headings are not uncommon.
Each listing was carefully hand-reviewed to verify that it was, in fact, only soliciting feedback
exchange. Any listing also offering a tangible, otherwise valuable or non-feedback good was
excluded. For example, if a seller promised matchbooks, coins, online data storage or low-value
long-distance telephone cards along with positive feedback, the observation was removed from
In total, 6,526 unique listings posted by 526 sellers were retrieved in January and May 2005, and
weekly from June 2005 to December 2005. Seventy-six percent of the listings, 5,127 items,
resulted in a sale. More than 80 percent of the auctions were listed with the Buy-It-Now option,
whereby a seller sets a fixed price and no bidding auction is conducted for the sale.
Table 1 provides summary statistics of the data.
Several features of the table are noteworthy.
First, unlike most other eBay markets, the average opening price is higher than the average
winning bid. Average opening price includes many offers with Buy-It-Now prices set equal to $1
(or £1 when listed in UK pounds), and these listings often failed attract to bidders—not too
surprising given the price points of the available alternatives offered by other sellers. Variation in
the shipping charge for the “virtual” goods sold in the market for feedback is also surprising.
While no postage was involved for any of the items traded in our data, some sellers sought to
attract bidders by setting an extremely low (often 1 cent) Buy-It-Now price and to earn profit on
the shipping and handling charge. Notice that seller profit is miniscule in this market—averaging
about 8 cents. As the case of
suggests, however, sellers enter this market not to
earn profits in the market for feedback itself, but rather to leverage the reputation gained in this
market to obtain price premia for other, presumably larger, transactions.
Turning to the characteristics of participants in this market, the median seller has a feedback
rating of 135 whereas a winning buyer has a much lower “reputation” with a feedback rating of
only 19. Sellers take advantage of the economies of scale inherent to the market for feedback; the
median seller listed 51 items in the market for feedback during the period of our study. Put
differently, if the median seller were to succeed in selling all of his or her listings during the
period of our study, the transactions would raise his or her existing reputation by 38 percent. As
C# HTML5 Viewer: Deployment on ASP.NET MVC
under Views according to config in picture above. RasterEdge.XDoc.PDF.HTML5Editor. dll. Open RasterEdge_MVC3 DemoProject, copy following content to your project: how to cut an image out of a pdf file; paste jpeg into pdf
on the rest of eBay, negative feedback is rare—the median seller enjoys 100 percent positive
feedback. Thus, sellers seeking primarily to build rather than rehabilitate their reputations are
entering this market.
Unlike the usual eBay assortment of electronics, collectibles and cars, feedback is
transaction and is not scarce in any traditional sense—no single user has inherently more to
exchange than another. Moreover, buyers and sellers in this market may transition without
substantial cost from one role to the other. Facing the appropriate incentive, a feedback buyer
can easily become a feedback seller, and vice versa. Users can move nearly costlessly between
buying and selling positions in the market, and because both users gain feedback through the
sale, prices represent cash transfers that smooth out the transaction frictions. While these features
make the data difficult to interpret directly, they also make the market for eBay feedback
empirically interesting. The theoretical justification for our valuation estimates is contained in
We find that an average buyer’s valuation for a point of feedback is at least 61 cents. While this
may seem like a relatively high valuation for a single point of feedback, it is important to note
that a buyer’s willingness to pay for a point of feedback in this market is an expression of the
discounted net present value of the stream of incremental future cash flows from this additional
point of feedback.
This calculation of buyer valuations was undertaken assuming that only a single price prevails in
the market for feedback.
However, as Figure 2 illustrates, the dispersion of winning prices in
the market is remarkable; prices in completed auctions ranged from well below eBay’s seller fees
to nearly 10 times marginal cost. While the distribution is concentrated near 1 cent, there is also
considerable mass around $1.
The feedback profiles of participants in this market are also highly diverse. Seller feedback
levels range from 0 to 6,732, while winning buyers have feedback ratings ranging from 0 to
9,231. Previous empirical studies suggest that the
of a point of feedback to a
buyer should depend on the existing reputation of that buyer. That is, buyers who already have
established reputations are likely to have a lower valuation for an additional point of feedback
than buyers with lower reputation. Since the feedback rating of the winning bidder is contained
in our dataset, it is possible to stratify the data by winners’ feedback ratings. We would expect
that the bound on buyer valuations would decrease with the feedback rating of the buyer.
In Figure 3, we divide winning buyers by feedback ratings to obtain valuation bounds for buyers
in each decile. As the figure shows, buyer valuations are mainly decreasing. For buyers in the
lowest decile, with feedback ratings of 4 or less, calculations suggest that they value a point of
feedback at, at least, 71 cents. In contrast, for buyers in the highest decile, those with feedback
ratings of 439 or higher, an additional point of feedback is worth much less—approximately 42
cents. Furthermore, the steep decline in the valuations of buyers in the highest two deciles
suggests that there is indeed “diminishing returns” to the value of an additional point of
At first glance, these values may seem implausibly high. Yet, when one considers the potential
payoff of improved reputation in markets
the market for feedback, the average valuation
is not unreasonable. Consider the following example: Assume for a moment that an increase
from zero to 20 feedback points results in a 5 percent increase in auction revenue in an average
eBay product category such as golf clubs.
A new user (with zero feedback points) wants to sell
a driver and expects to sell it for $250. If he purchases 20 positive points in the feedback market
for approximately $10.80, he can earn 5 percent more on the golf club sale, $262.50. That is, his
investment in reputation is more than offset by the extra auction revenue.
Is the market for feedback profitable to sellers?
As we saw in Figure 2, there is a considerable range in transactions prices in the market for
feedback. This also suggests that sellers experience a range of profit outcomes. One of the
unique features of this dataset is that seller costs (excluding hassle costs associated with posting
an item) are completely transparent to the researcher. As a consequence, we can determine the
profitability of each listing in our dataset. Figure 4 graphically displays the distribution of profit
and loss outcomes in this market. As the figure shows, the distribution of profits is bimodal. The
higher of the two modes, which comprises 1,100 transactions making a loss of 29 cents, arises
when sellers offer an item in the market for feedback at a Buy-It-Now price of 1 cent with free
shipping and handling. Since the insertion fee on the US eBay site is 25 cents and the Buy-It-
Now option costs an additional 5 cents at this price point, a seller’s costs for this transaction
amount to 30 cents; thus leading to a net loss of 29 cents. The other mode occurs at a profit level
of approximately 64 cents. This profit level, which occurs in 263 transactions, arises when sellers
successfully offer an item with a Buy-It-Now price of 99 cents and free shipping and handling.
Figure 5 illustrates the overall profit or loss obtained by each seller in our dataset. Most sellers
incur losses—the modal seller loses 30 cents per listing. Moreover, even for sellers setting an
item price equal to nearly one dollar, the feedback market is still relatively unprofitable, since
these items fail to sell approximately 45 percent of the time. Thus, about half of the 99-cent
listings lead to a loss of 30 cents while the other half (when the item sells) earn a profit of about
65 cents. The net expected profit from a 99-cent listing is only 22 cents. It should perhaps not be
all that surprising that the market for feedback is not a profitable one for sellers. As the case of
illustrates, motives other than direct profits from feedback sales are often
paramount. Moreover, since there is an effectively limitless supply of the “good” in this market,
there is little reason to expect sellers to earn profits.
6 Managerial Implications
In markets with millions of nearly-anonymous agents buying and selling a plethora of goods,
trust is critical. Moreover, the higher the value of the items being bought and sold, the more vital
is trust to the successful engagement of buyers and sellers. As trade in high-value item becomes
increasingly profitable on the Internet, online merchants and auctioneers face enormous
challenges in overcoming the trust problem and creating attractive trading environments. Our
work suggests that one current “state of the art” solution, the trust system employed by auction
giant eBay, is vulnerable to being undermined in precisely those areas targeted for future growth.
Indeed, the impact of the market for feedback stretches beyond eBay. As new businesses enter
the online space, either as start-up e-retailers or virtual versions of established brick-and-mortar
stores, managers face the challenge of creating an environment of trust that attract and maintain a
stable customer base. While the enormous success of eBay might tempt the new managers to
emulate eBay’s feedback system, our work identifies a significant weakness and suggests that
solutions to the trust problem should be sought elsewhere. The need for new approaches to
online reputation systems is especially critical for firms seeking growth in emerging online
markets—e-retailers and auction platforms in these markets may face additional pressure as users
race to catch up to developed markets. And so, we ask: What can a manager do to build an
environment of trust online, without being vulnerable to markets for feedback that may
undermine the system?
While eBay’s existing reputation mechanism has seemed to work well for smaller items where
the benefit of investing in a “false reputation” is relatively modest, this is not the case for high-
value items. Moreover, as eBay attempts to expand internationally, particularly in developing
countries such as China, the need for reliable mechanisms to distinguish good sellers from bad
will be all the more important—especially since financial systems and credit card use are far less
developed in these new markets. The presence of markets for feedback suggests a potential
“Achilles heel” for eBay and other online auction sites seeking these growth opportunities.
As we have shown, there are several important problems in eBay’s existing reputation system
that are being exploited in the market for feedback. First, since reputation is not weighted by the
value of the transactions giving rise to the overall feedback score, there is no way for a buyer to
distinguish between a seller whose reputation derives from legitimate transactions and one whose
reputation derives from what are arguably only notional transactions. Second, eBay only retains
a detailed archive of the transactions comprising the reputation of a seller for 90 days. Thus, it is
possible for a seller to affect his or her perceived reputation by taking advantage of this short
time horizon of transparency. Third, since feedback is bilateral, eBay dilutes the incentives for
buyers to give negative feedback, even when a seller’s performance is not especially good.
Sellers can, and often do, retaliate against buyers leaving negative feedback by reciprocating the
negative review. Since the reputation of an individual on eBay is a composite of transactions
made as a buyer and as a seller, buyers who also sell items on eBay (or who expect to sell items
in the future) may be reluctant to risk their reputations by leaving negative feedback.
How can eBay address these problems? First, eBay may wish to offer transaction-weighted
reputational statistics based on the dollar value of the trade rather than the current practice where
the sale of a car and the sale of a digital photo of Bigfoot have the same reputational effect.
Second, given the dramatically falling cost of storage and computing, there would seem to be
little technological reason for eBay to limit the time horizon of its detailed archive of
transactions. More broadly, greater transparency in providing information about the past history
of a seller should improve the ability of buyers to distinguish between good and bad sellers, and
thereby avoid the trust problem. Third, there seems to be little reason to pool reputation earned as
a buyer and reputation earned as a seller. As we mentioned above, this pooling creates a
disincentive for honest reporting and helps to undermine the informational value of the system.
EBay could easily create separate reputational accounts for a given user, segregating reputation
Why doesn’t eBay and others then implement these solutions? For established platforms, such as
eBay, a central concern in any reform of an existing reputational system is that it will damage the
loyalty of its existing user base. For instance, how would existing feedback ratings be treated
under a system with transaction-weighted reputation? In principle, there could be adverse
litigation consequences from eBay from sellers who felt that their businesses were harmed by
such a change and who had relied substantially on eBay’s existing rules in determining their
business strategy. Such a change might also provide an opening for eBay’s formidable
competitors—Amazon and Yahoo—to grab market share at eBay’s expense. Thus, to some
extent, eBay appears to be “locked in” to its existing reputational system by virtue of its own past
success. One wonders, however, whether the pernicious effects of the market for feedback will
not ultimately undermine eBay’s competitive advantage for the future.
The managerial challenges in solving the trust problem differ for existing platforms and new
platforms. For existing platforms, the transition from current reputational systems to more robust
systems that can operate effectively for high-value items and emerging markets is a central
business consideration. For new platforms, the weaknesses in existing systems highlighted here
offer a unique opportunity to overcome the built-in first-mover advantages of the established
players and gain a competitive advantage by innovating new solutions to the trust problem.
The situation in trust markets appears to us to be analogous to that in online search in the 1990s.
At that time, the state of the art solution was to use data encoded in metatags to provide search
results. New players, such as Google, recognized the vulnerability of existing search engines to
manipulation of data contained in these metatags and were able to overcome first-mover
advantages enjoyed by established players, such as AltaVista. We see the same vulnerability to
manipulation in existing reputational systems in online markets and, perhaps, an opportunity for
another Google to leapfrog the competition with solutions that scale to high-value items and
Documents you may be interested
Documents you may be interested