MANAGERIAL ECONOMICS

LECTURE 9: AUCTIONS

Rudolf Winter-Ebmer
Summer 2021
 Aims of this lecture

 Explain how auctions work

 Describe the importance of auctions
 How to use auctions in strategic decisions

RWE Managerial Econ 9 Winter term 2020 1 / 38

 A short history of auctions

It is an old form of market transaction:

 Ancient Greece, Babylon: Mineral rights, women

 Roman authorities:
 Collected debt through auctions of goods (slaves)
 Candle auctions:
 Auction is limited until a candle burned out
 Procurement auctions
 Government sells rights: To build motorways, schools, ...
 Auction use expanded rapidly with the development of the Internet and
e-commerce in the 1990s.
 Radio spectrum: Auctions of bandwidth for mobile telephones

RWE Managerial Econ 9 Winter term 2020 2 / 38

 Types of Auctions I

English Auction (ascending-bid):

 Initial bid set at seller’s reservation (minimum) price

 Auctioneer bids up using a fixed step size
 Bidders indicate whether they accept the offer or not
 The last bidder buys the good at her bidding price
 Information about willingness to pay is revealed
 Sometimes there are time limits
 Sniping: When bidders time their bids to the last possible moment (software)
 Other ascending auctions: Clock or Japanese auctions
 Examples: Art auctions, livestock, online-auctions

RWE Managerial Econ 9 Winter term 2020 3 / 38

 Types of Auctions II

Dutch Auction (descending-bid):

 Starts with a high initial price

 The price is lowered over time
 First bidder who accepts buys the good
 Used for Dutch flower auctions
 Very fast format
 Convenient when it is important to auction goods quickly

RWE Managerial Econ 9 Winter term 2020 4 / 38

 Types of Auctions III

Sealed-bid auctions:

 One bid per person in a sealed envelope

 First-price sealed bid auction: bidder with the highest bid receives the good at
her stated price
 Reverse sealed-bid auctions
 bidder with lowest cost wins
 public procurement: construction (highway)
 Mineral rights auctions, oil leases

RWE Managerial Econ 9 Winter term 2020 5 / 38

 Types of Auctions IV

Second-price sealed-bid auction:

 “Vickrey” auction
 William Vickrey, Prize in Economic Sciences in Memory of Alfred Nobel 1996
 The highest bidder receives the good, but she has to pay only the price of the
second-highest bidder

RWE Managerial Econ 9 Winter term 2020 6 / 38

 Why use an auction to sell anything?

 Rules of the auction can easily be changed

 Goods are limited (often unique, but not necessarily)
 A way to elicit willingness to pay from consumers
 Online auctions are easily organized

RWE Managerial Econ 9 Winter term 2020 7 / 38

 How do auctions work?

Assumptions:

 Bidders are symmetric

 Bidders who have identical reservation prices and who observe the same signal
will submit the same bids.
 Bidders select their bid from the same distribution of possible bids
 But they do not necessarily choose the same bid
 Bidders are risk-neutral (i.e., expectations matter)
 Bids are based on signals from an independent distribution of valuations
 My valuation of the object does not influence your valuation

RWE Managerial Econ 9 Winter term 2020 8 / 38

 Taxonomy of auctions

 Private-value auctions:
 Bidders have different, private valuations of the good being sold
 Reservation prices are a function of information and utility
 E.g., art, paintings, collectors’ items
 Common-value auctions:
 Auctions in which all bidders value the good similarly
 Each bidder forms expectations about the true value based on (incomplete)
information about the good being sold
 E.g., rights to extract oil from an oil field

RWE Managerial Econ 9 Winter term 2020 9 / 38

 Mechanism and Revenues

Revenue equivalence theorem

The type of auction does not affect the expected total surplus and hence does
not affect the expected revenues for the seller.

Notation:

 b: bid (an amount of money a customer is willing to pay for a good)

 p: price paid by winning bidder
 P [W ]: probability of winning the auction
 Expected profit: P [W ] × (b − p)
 Expected profit is the same for the four types of auctions.

RWE Managerial Econ 9 Winter term 2020 10 / 38

 Optimal bidding strategies

 Bidders should always be willing to bid up to their reservation prices.

 If the current bid is higher than or equal to your reservation price, don’t bid.
 If the current bid is lower than your reservation price, increase the bid.

RWE Managerial Econ 9 Winter term 2020 11 / 38

 Bidding strategies

 English auction:
 Bidders learn about others’ reservation prices
 Always bid up to the reservation price (in fact, bid one cent above the last bid of
the others)
 Good is sold at the reservation price of the second-highest bidder (plus 1c or
the minimum additional step size)
 Second-price sealed bid auction:
 Dominant strategy to bid exactly the reservation price
 Vickrey auction is a “revealing mechanism”:
• The design of the auction (“the game”) provides bidders with an incentive to reveal
their true willingness-to-pay
• The study of such mechanisms, “mechanism design”, has become an important
branch of economics.

RWE Managerial Econ 9 Winter term 2020 12 / 38

 Dutch and sealed-bid auction

The bidder’s problem: Should you write your true WTP in the envelope in a
first-price sealed-bid auction? Dutch and first-price sealed bid auction are
strategically similar

 There is no dominant strategy; you have to consider what the others are
bidding
 In a Dutch auction: you should not bid your reservation price, because if you
buy the good, there is no surplus.
 If the distribution of valuations is known, bidders can anticipate bidding.
 Nash strategy: assume others are behaving rationally; then you should
maximize your expected profit

RWE Managerial Econ 9 Winter term 2020 13 / 38

 Dutch and sealed-bid auction II

Nash-strategy:

 Estimate the reservation price of the second-highest bidder and bid it.
 How to do it?
 Number of bidders n is important: the more bidders, the closer you should bid
to your own reservation price
 Example: If distribution of bids is uniformly distributed between L (lowest
value) and v (the bidder’s reservation price):
 Bid: v − [(v − L)/n]

RWE Managerial Econ 9 Winter term 2020 14 / 38

 Example

Consider a descending auction or a first-price sealed-bid auction:

 Assume: v = 3
 Assume: L = 0
 n = 2: Your bid = 3 − [(3 − 0)/2] = 1.50
 n = 3: Your bid = 3 − [(3 − 0)/3] = 2.00

RWE Managerial Econ 9 Winter term 2020 15 / 38

 Efficiency vs. Expected revenues

 The seller aims at getting the highest possible revenues

 Another aim is to sell the item to the bidder with the highest valuations
 Consider procurement auctions where the government is the seller:
 E.g., license to construct a highway
 Aim: to guarantee lowest construction cost
 Bidder with lowest cost estimate should get the deal.
 With symmetry, all four auction formats are efficient

RWE Managerial Econ 9 Winter term 2020 16 / 38

 Optimal Strategy for the Seller

Auctions can be interpreted as being similar to third-degree price discrimination

Example:

 Four units to sell, M C = 0

 Six consumers with different WTP: 90, 60, 50, 40, 20, 15
 Expected revenues from an ascending auction vs. a fixed price of 40?

RWE Managerial Econ 9 Winter term 2020 17 / 38

 Calculation of expected revenues
Consumers Reservation price Consumer surplus
Price = 40 Auction
1 90 50 29.99
2 60 20 9.99
3 50 10 9.99
4 40 0 19.99
5 20 – –
6 15 – –

Total consumer surplus 80 69.96

Total seller surplus 160 170.04
Total surplus 240 240

Notes: The auction is an ascending auction where customer 1 bids 60.01 to win the object, et cetera. Maximum surplus with these valuations is 240.
(Why?) With an auction, sellers can obtain a greater share of the overall surplus. Tables 13.1 and 13.2 in Allen et al., Managerial Economics (8th ed.),
p511.
RWE Managerial Econ 9 Winter term 2020 18 / 38
 Seller’s expected revenue and Bidder’s expected marginal
revenue

Notes: Bidder i values a good at b̄, this is the maximum she is willing to pay for it. Her bid and all other bids are from some distribution F (b). The
probability that a bid is the winning bid is given by 1 − F (b). Each point on the curve b is the probability that this is the winning bid. Denote b∗ the
actual winning bid (in this case, the reservation price of the second highest bidder). The seller’s expected revenue is b∗ [1 − F (b∗ )], this is the
shaded area. This is also the expected marginal revenue of the winning bid. Figure 13.1 in Allen et al., Managerial Economics (8th ed.), p511.

RWE Managerial Econ 9 Winter term 2020 19 / 38

 Auctions to obtain more information about buyers

Consider a Repurchase Tender Offer (RTO), where a company wants to buy back
shares from current shareholders:

 The RTO typically offers a premium over the current price at the stock
exchange as owners are not required to sell
 The company may offer a price and wait how many shares will come back
 It could use a Dutch auction: firm announces a price range and current
shareholders announce how many stocks they would sell for different prices
 These announcement are aggregated to the market supply

RWE Managerial Econ 9 Winter term 2020 20 / 38

 Citizens First’s RTO

 Company wants to buy back 391,000 shares which were values at $14 per
share
 Company assumes share is worth $20
 Modified Dutch auction TO (MDATO): Company will pay between $15 and $17
per share

RWE Managerial Econ 9 Winter term 2020 21 / 38

 Shareholder supply schedule

Price Strong ΠStrong Medium ΠMedium Weak ΠWeak

15 400,000 2,000,000 310,000 1,550,000 280,000 1,400,000
16 415,000 1,660,000 400,000 1,600,000 315,000 1,260,000
17 600,000 1,800,000 415,000 1,245,000 400,000 1,200,000

Probability 0.4 0.3 0.3

Notes: Assumed distribution of current shareholders’ willingness to tender. Probability states the chances of such a state of the world. Π give the
profits if the true value of the share is $20. Table 13.3 in Allen et al., Managerial Economics (8th ed.), p517.

RWE Managerial Econ 9 Winter term 2020 22 / 38

 Fixed-price RTO

To set an optimal fixed price, Citizens First will have to assume how the willingness
to sell is distributed among its current shareholders. (See previous slide for
assumed distributions.)
The firm might set a price based on the expected value

 EV (15) = 2.0 × 0.4 + 1.55 × 0.3 + 1.4 × 0.3 = $1.685

 EV (16) = 1.66 × 0.4 + 1.6 × 0.3 + 1.26 × 0.3 = $1.522
 EV (17) = 1.8 × 0.4 + 1.245 × 0.3 + 1.2 × 0.3 = $1.453

The firm might therefore choose a price of 15.

RWE Managerial Econ 9 Winter term 2020 23 / 38

 MDATO
The price is set only after the auction, based on the response.

 The firm can exploit the response from the current shareholders: if the
response is strong, it cab set the price at $15, if the response is medium, the
price is set at $16, and the response is weak, the price will be set at $15.
 EV = 2.0 × 0.4 + 1.6 × 0.3 + 1.4 × 0.3 = $1.700
 The firm can now set a higher price in one of the cases, obtaining a greater
profit.
 How does it work?
 Importance of incentives to reveal the correct supply schedule (“truth-telling
mechanism”)
 Contract is needed: shareholders cannot sell more shares at the specified price
(otherwise they would lie)
RWE Managerial Econ 9 Winter term 2020 24 / 38
 Auctions of airtickets
priceline:

 Auction for airline seats

 Airline seats are perishable goods, marginal costs are low, empty seats bring
no cash
 Discounts endanger pricing structure of regular seats
 Buyers name the price they are willing to pay: if an airline accepts, the ticket
is sold

RWE Managerial Econ 9 Winter term 2020 25 / 38

 Risk aversion

 In second-price auctions risk-aversion does not matter. (Why?)

 What about a first-price auction or Dutch auction?
 Higher bids increase probability of winning
 If risk-averse bidders want to avoid losing, they will increase their bid
 If bidders can be assumed to be risk-averse, then sellers should chose
first-price auctions!

RWE Managerial Econ 9 Winter term 2020 26 / 38

 Asymmetries

 In second-price auctions it does not matter whether bidders are asymmetric.

(Why?)
 What about a first-price auction or Dutch auction?
 Bidders with higher valuations will underbid more (“shade”) than bidders with
lower valuations
 Bidders with lower valuations may win, this is inefficient
 If bidders are (expected) asymmetric, then sellers should choose ascending
auctions
 But: It is easier for bidders to collude in ascending auctions

RWE Managerial Econ 9 Winter term 2020 27 / 38

 Number of bidders

More bidders will increase profit for the seller:

 Expected bid (sale) = reservation price of the second-highest bidder.

 Assume reservation price of bidders is uniform between 0 and 100
 Expected revenue for seller:
 E[revenue] = (N − 1)/(N + 1) × (highest reservation price)

RWE Managerial Econ 9 Winter term 2020 28 / 38

 E[revenue] and number of bidders

Notes: The E[revenue] increases with the number of bidders as the probability of higher bids increases. Figure 13.2 in Allen et al., Managerial
Economics (8th ed.), p519.

RWE Managerial Econ 9 Winter term 2020 29 / 38

 Is it always good to win the auction?

The Winner’s Curse

When the true value of the good is unknown, the bidder with the highest bid might
bid more than the good is worth.

 Assume That the true value of the good is not known for sure, but common to
all (e.g., the profits from an oil field)
 Sealed-bid first price auction: The most optimistic bidder wins
 Similar in procurement auctions (reverse auction)

RWE Managerial Econ 9 Winter term 2020 30 / 38

 The Winner’s Curse

Notes: Unbiased estimates of the good’s true value are given by distribution B. The mean of this distribution is indicated by the dashed line. The
distribution of bids will be to the left of B, e.g., the distribution given by A. Note that some of the bidders, whose estimates of the true value are in the
right tail of A, will bid more than the mean of B. Figure 13.2 in Allen et al., Managerial Economics (8th ed.), p519.

RWE Managerial Econ 9 Winter term 2020 31 / 38

 How to avoid the winner’s curse

 A bidder with relatively less information than others should bid lower.
 A bidder with relatively little confidence in his or her estimate of the value of
the good should bid lower.
 The more bidders there are in the auction, the lower the bid should be!
 Be also aware of the winner’s curse as a seller

RWE Managerial Econ 9 Winter term 2020 32 / 38

 Concerns for the seller

1. Collusion:

 Bidders may collude to prevent the seller from extracting surplus.

 Bidders may form a “ring” in which no one bids against the designated bidder.
The goods are afterwards distributed among the members of the ring.
 Colluding bidders do not bid up prices.
 Collusion is more probable in multi-unit auctions (several units of a good are
sold) or if the auction is repeated over time (different highway construction
lots)

RWE Managerial Econ 9 Winter term 2020 33 / 38

 Collusive bidding

FCC spectrum auction: simultaneous ascending auctions for frequencies in

different regions

 Bidders used the last few digits of a bid to encode messages.

 For example, GTE frequently used bids that ended in 483 — this is “GTE” on
the telephone keypad.
 In the same auction, American Portable, a daughter of TDS, signaled interest
by bids that ended in 837 (“TDS”)
 Firms used such prices to indicate their interest in certain markets and to
coordinate their efforts.
 Collusive bidding lowers revenues and efficiency!

RWE Managerial Econ 9 Winter term 2020 34 / 38

 Collusive bidding

German GSM auctions:

 10 equally sized (and valued) blocks of spectrum

 Mannesmann bid 18.18 mio on blocks 1-5 and 20.0 mio on blocks 6-10
 why different bids for equal products?

 T-Mobil increased their bid to 20.0 mio for blocks 1-5

 Both companies ended up sharing the market... “and the auction was over,
before it had gained momentum”

RWE Managerial Econ 9 Winter term 2020 35 / 38

 Concerns for the seller

2. Few bidders:

 If it is clear to all bidders that one particular bidder will win, they are less likely
to participate in the auction.
 If the seller sets a reserve price that is too high, then bidders are less likely to
participate in the auction.
 If the seller sets a reserve price that is too low, it encourages collusion.

RWE Managerial Econ 9 Winter term 2020 36 / 38

 Strategic PR

Telekom Austria was cited in the newspapers, prior to frequency auctions:

“it would be satisfied with 2 out of the 12 blocks of frequencies on offer”,
but “it would bid for a 3rd block if one of its rivals did”

RWE Managerial Econ 9 Winter term 2020 37 / 38

 Further reading

 Auctions and Bidding: A Primer, Milgrom, Paul. 1989.

 What Really Matters in Auction Design, Klemperer, Paul. 2002.
 How Auctions Work for Wine and Art, Ashenfelter, Orley. 1989.
 Economic Insights from Internet Auctions, Bajari, Patrick, and Ali Hortaçsu.
2004.

RWE Managerial Econ 9 Winter term 2020 38 / 38