If there is no snipping (auction is closed only when no bids are submitted for a certain duration), the auction theory predicts that buyers should slowly increase their bids until the highest bid exceeds their WTPs
The open English auction mechanism induces truth revealing of WTPs (price points where bidders drop from the auction)
Revenue = Second highest WTP
Predicted Versus Actual Bids and Reserve Price ? ? Reserve Price ? 99% Bid as % of WTP (except winners) Actual Behaviors Predicted By Auction Theory
Prices at Each Round Round 5 Round 4 Round 3 610.0 Round 2 732.0 Round 1 3 rd highest WTP Price of 2 nd item Price of 1 st item Round 6 580.0 631.0 671.0 790.0 839.0 350.0 850.0 850.0 662.0 642.0 798.0 760.0 630.0 613.0 674.0 650.0
Predicted Versus Actual Bids and Reserve Price 500 Reserve Price 95.5% 99% Bid as % of WTP (except winners) Actual Behaviors Predicted By Auction Theory
Expected Seller Revenue (one seller)
Let buyer WTPs (independently drawn from an uniform distribution) are denoted by WTP 1 , WTP 2 , …, WTP n-1 , WTP n
Arrange them in an ascending order and denote ranked bids by WTP (1) , WTP (2) , …, WTP (n-1) , WTP (n) . That is, we have WTP (1) < WTP (2) < …< WTP (n-1) < WTP (n)
If the reserve price is smaller than the second highest WTP, reserve price has no revenue effect on the bidding process
If it is between the highest and the second highest WTPs, it will force the highest bidder to go beyond the second highest WTP (the point where the second highest bidder drops out).
Predicted Versus Actual Bids and Reserve Price 289 500 Reserve Price 95.5% 99% Bid as % of WTP (except winners) Actual Behaviors Predicted By Auction Theory
Revelation Principle : All buyers (except the winner) end up revealing their WTPs for the item to the sellers. Put differently, sellers can “discover” distribution of buyers’ WTPs via the auction mechanism. This is not the case for other pricing mechanisms (e.g., seller posted price and priceline model)
Application : Use open English auction to discover the distribution of WTPs and then use seller posted price model to maximize share of surplus
Efficiency Principle : The auction mechanism guarantees that the item always goes to the buyer with the highest WTP in a market. Again, this is not always the case for other pricing mechanisms
(inefficiency = 0% (predicted) versus 1.6% ( actual ) )
Observation: Auction model is likely to dominate the other price models in terms of total social surplus generated.
Can be stressful for some buyers, especially for end consumers (this can be however solved by proxy bidding). Thus, it is less likely to work for low purchase-involvement product categories
There are more losers than winners (and losers are less likely to revisit the website)
Strategic manipulation (e.g., shilling)
Next Class: Design of Auction Format
Dutch auction (D) (e.g, www.landsend.com )
First-price sealed-bid auction (F) (e.g., US Treasury bills, supply contracts; government procurement (10% GDP))
Second-price sealed-bid (Vickrey) auction (S) (e.g., www.sandafayre.com ; NZ’s communication spectrum rights)
English auction (E) (e.g., www.ebay.com ; www.yahoo.com )
Ebay’s business model Seller Buyer Seller & Buyer Price Setting Party Price Over Time Static Dynamic ebay.com http://www.ebay.com/
Ebay’s Current Status in 2004
Listings - New listings totaled a record 348.0 million in Q3-04, 48% higher than the 234.6 million listings reported in Q3-03.
Confirmed Registered Users - Cumulative confirmed registered users at the end of Q3-04 totaled a record 125.0 million , representing a 46% increase over the 85.5 million users reported at the end of Q3-03.
Active Users - Active users, the number of users on the eBay platform who bid, bought, listed or sold an item over the trailing 12 months, increased to 51.7 million , a 38% increase over the 37.4 million active users reported in the same period a year ago.
Ebay’s Original Revenue Model
M = Total number of items listed per year
Fee i = Insertion fee of item i
Com i = Final value fee of item i
I i = 1 if item i is sold; 0 otherwise
M depends on number of people in the community
Com i x I i increases with price or seller surplus
Ebay’s Original Revenue Model Biggest strength of the revenue model: Revenues increase linearly with M but costs do not increase with M . (scalability property)