Transcript eBay
An Online
Consumer-to-Consumer
Trading Community
Presentation is based on Melnik research:
“Does a Seller’s eCommerce Reputation Really Matter? Evidence from eBay Auctions” (with
James Alm). Journal of Industrial Economics, September 2002.
“Reputation, Information Signals, and Willingness to Pay for Heterogeneous Goods in Online
Auctions”, (with James Alm). Southern Economic Journal, October 2005.
1
eBay: A True Success Story
From a simple website in 1995 to being
synonymous with online auctions!
1.9 billion listings in 2005
4.552 billion in revenues
71.8 million active users
96.2 million accounts listed with PayPal*
But what about the economics.....
* All information is taken from QIV05 eBay Financial Results report
2
Asymmetry of Information
Akerlof, 1970
Asymmetry of Information on eBay
Buyer’s problem
Uncertainty about delivery of the item (general compliance with the
terms of transaction)
Uncertainty about the accuracy in the description of the item
Seller’s problem
Payment/return
Past Reputation as a Signal of Current and Future Behavior
Theoretical support
Klein and Leffler, 1981; Shapiro, 1983; Allen, 1984; Houser and
Wooders, 2000
Experimental support
Miller and Plott, 1985; DeJong, Forsythe, and Lundholm, 1985;
Camerer and Weigelt, 1988; Holt and Sherman, 1990
3
Reputational Mechanism on eBay
SIMPLE * MEASURABLE * DIFFICULT TO MANIPULATE
Structure of the mechanism
Quantitative
Positive, negative, neutral rating choices only
Difficult to manipulate through collusive behavior
Rating left by unique registered eBay users
Feedback score = unique positives – unique negatives
Informative
Overall eBay experience of the seller
Past complain history
Does the reputational measure help overcome
asymmetries of information?
Is it valued by members of the community?
Is it valued by competing communities?
4
Choice of Data
2002: Homogeneous good study
US $5 1999 Gold Coin in Mint Condition
Possibility of encountering a fraudulent seller
2005: Heterogeneous good study
US Morgan Dollars in Almost Uncirculated
Condition
Accuracy in the description of item-specific
characteristics
Possibility of encountering a fraudulent seller
5
Modeling Reputation
P = f (seller’s reputation, X)
X – a set of auction specific variables
Transaction costs (shipping, insurance)
Time exposure, closing (duration, closing
time/date, day of the week)
Supply characteristics (number of available
items)
Payment methods
6
Empirical formulation
Censored observations and the use of Tobit
model
Fixed price auctions and no-bid auctions
Pi X i i
*
Pi Pi if Pi Pi
0
Pi Pi
0
*
if Pi Pi
b
b
*
Pi Pi otherwise
*
• 105 price distributions: Huber-White estimation of
robust standard errors
7
Estimation Results
1% increase at
the mean
increase from 0
to mean
increase from 0
to 1
rating
negative (1 point)
rating
negative
rating
negative
Certified
Non-certified
Non-certified, no scans
stat insig
$-3.450 (1.1%)
stat insig
$-95.055 (29.0%)
stat insig
$-27.477 (8.4%)
$ 0.029 (0.05%)
stat insig.
$21.848 (37.6%)
stat insig.
$2.009 (3.5%)
stat insig.
$0.05 (0.084%)
-$0.89 (1.5%)
$36.69 (63.2%)
-$15.76 (27%)
$3.37 (5.8%)
-$5.25 (9.05%)
Mean prices: Certified: $327.50; Non-certified: $58.08
1. Seller’s reputation impacts buyer’s willingness to pay
2. In heterogeneous goods: A reduction in available
information increases the premium to positive reputation
and the penalty to negative reputation.
3. Negative feedback effect increases with the value of the
item
4. Substantial penalty is imposed on new sellers in noncertified coins auctions
8
Some Previous Findings
- Lucking-Reiley et al. (1999): 1%
increase in rating -> 0.03% increase in
willingness to pay
- Houser and Wooders (2002): 10%
increase in rating -> 0.17% increase in
willingness to pay
- Melnik and Alm (2002): Doubling in
rating -> 0.55% increase in willingness
to pay
9
Conclusions
Non-transferable across communities reputational
mechanism in online consumer to consumer
communities acts as a club good
• Valued by buyers and sellers
• Enables a community to overcome
asymmetries of information problem
• Establishes a barrier to entry for a competing
community
10