DUTCH AND ENGLISH
AUCTIONS IN RELATION
TO THE TULIP MARKET
Liz DiMascio
Paige WarrenShriner
Mitch Justus
DUTCH AUCTIONS
 High initial starting price —price lowers until acceptable to a
bidder
 Strategically equivalent to first -price sealed bid auctions
 Dutch flower auction takes place every weekday in Aalsmeer,
Netherlands
 Retail flower salesman are participants
 Frequently used in auctions for nondurable non -consumer
goods
T YPICAL FEATURES OF NONDURABLE
GOODS AUCTIONS
 Sequential
 Bidders are purchasing goods for resale
 Identical goods, identical bidders
 Same number of goods each round
 Independent Private Values (IPV): has a dif ferent value for
each bidder (= Revenue – IP Cost)
PRIVATE VALUE LITERATURE
 “Why Do We Use the Dutch Auction to Sell Flowers?
—Information Disclosure in Sequential Auctions” ( Tu)
 Given several simplifying assumptions, Dutch auction with only
winning bid announced produces higher revenue than English auction
in a sequential environment
DUTCH AUCTION WITH WINNER’S BID
REVEALED
 Bidders bid symmetrically according to bid function B(v)
 In second round, winner believes that loser’s value lies in
interval [0,v) where v is winner’s value
 Loser calculates second bid based on first, losing bid
 B(v) for first round = v/2
 First stage winner randomizes over (v/2, 3v/4)
 Loser will not bid if v < v’/4, where v’ is inferred valuation of
first stage winner
 In second stage, winner and loser are brought to relatively
level playing ground (since loser of first round has
informational advantage, while winner randomizes)
PRIVATE VALUE VS. COMMON VALUE
 Common Value Good: has one true underlying value for all
bidders. This value is unknown at the time the bid is placed,
and bidders receive independent signals of it. Values are
influenced by others’ signals.
 Private Value Good: has a dif ferent value for each bidder
 Usually used to describe modern day auctions of nondurable non consumer goods
COMMON VALUE LITERATURE
 “A Theory and Auctions and Competitive Bidding” ( Milgrom,
Weber)
 Consideration of revenue in static auctions (non -sequential)
 Common Value Auctions and the Winner’s Curse (Kagel, Levin)
TULIPS IN THE 1630’S
 1630s: Tulip market in Amsterdam undergoes widespread
speculation leading to a price bubble
 Professional growers and nonprofessional speculators interested
in resale value of tulips (experienced bidders)
 Tulips in the 1630s were common -value goods, not private value
(because of local nature of market, costs were more or less
equivalent, and revenue, a common -value derived from
aggregate demand, was unknown)
 Each bidder receives a “signal” of market demand of certain
tulips
 This market demand is the true resale value of the tulips to each bidder
OUR QUESTION
What conclusions can be drawn from
sequential auctions for common value goods,
such as those that took place for tulip bulbs in
the 1630’s?
 Bidding strategies
 Information asymmetry
 Seller’s revenues
OUR EXPERIMENT
 2 participants, 2 auction rounds, 2 identical goods (with one
being auctioned each round)
 True value of good is unknown at the time each player
submits their bid, and is drawn from a discrete distribution of
5 consecutive integers
 The true value of the goods are unique to each competing pair
 Each player receives a private signal of the value of the good
 Each player’s signal is drawn from the same distribution as the true
value of the good, thus, a player’s signal will be no more than 2 units
away from true value of the good
AUCTION 1: DUTCH AUCTION
 High starting price will be announced, then begin descending
by one
 Each bidder’s profit = True value of good – Bid (Profit = 0 for
loser)
 First bidder among each competing pair to bid wins auction
 Ties will be randomly broken
 Each player will record their signal, bid in each round, whether
they won or lost, and the true value of the good (given after
the second round)
PROBABILIT Y DISTRIBUTION OF SIGNALS,
GIVEN TRUE VALUE OF GOOD = 3
AUCTION 2: ENGLISH AUCTION
 Price will start from 1 , all bidders willing to purchase the good at
a price = 1 will indicate their willingness to bid at this price
 Price will ascend by an increment of 1 until one bidder (in each
competing pair) drops out
 Profit = True Value of Good – Price at which previous bidder
dropped out
 Same relationship between signals and true values of goods
holds
 Each player will record their signal, bid in each round, whether
they won or lost, and the true value of the good (given after the
second round)
OUR MODEL
 2 bidders, 2 rounds
 Symmetric strategies
 Same true value both rounds
 Signals randomly drawn from discrete distribution made to
approximate a normal distribution
 Dutch auction, winner’s bid becomes known
 Experienced bidders (eliminates winner’s curse)
ASSOCIATED PROBABILITIES OF SIGNALS
AND AUCTION RESULTS
PROFIT FROM USING GIVEN STRATEGY
PROFIT FROM USING GIVEN STRATEGY
PROFIT FROM USING GIVEN STRATEGY
PROFIT FROM USING GIVEN STRATEGY
PROFIT FROM USING GIVEN STRATEGY
NEW VALUATIONS
SECOND ROUND STRATEGIES
PLANNED EXTENSIONS
 What about the continuous normal distribution?
 Can’t fully eradicate winner’s curse (unbounded distribution).
 What about the English Auction?
 What is the optimal bidding strategy in the first round?
 How valuable is information about the other bidder’s signal
 Is it necessary to use a model with >2 bidders?
 Revenue
 What is the exact relation between bidders’ auction format and seller’s
revenue?
REFERENCES
 Kagel, John H., and Dan Levin. Common Value Auctions and
the Winner’s Curse. Princeton: Princeton University
Press,
2002. eBook.
 Milgrom, Paul R. and Robert J. Weber (1982); “A Theory of
Auctions and Competitive Bidding,” Econometrica, 50,
1089-1122.
 Tu, Zhiyong. Why Do We Use the Dutch Auction to Sell
Flowers?
– Information Disclosure in Sequential Auctions.
Diss.
The University of Pittsburgh, 2006. Web.