MiniCases

1
Mini-Cases

• Auction Vignettes

2
Two Auctions

• Clinton memoirs
• 8 publishers chose to bid
• Open bidding
• Royalty and fixed fee
• Chose highest and with fixed fee
• 8 million

• Rodriguez
• Blind auction
• Uncertainty about bidders
• 252m deal

3
Questions

• How are the assets in the auction similar or
  different? What are the common and private value
  elements?
• What do these differences mean for the seller and
  buyer?
• How would you compensate the agents?
• Is there evidence on overpaying?

4
On-Line Auction Sites

• On eBay.com, peculiar bidding practices have been
  noticed. In particular,
• Many bidders submit multiple bids in the course
  of the auction (i.e., they submit and later raise
  the maximum price they authorize for proxy
  bidding on their behalf) and
• a non-negligible fraction of bids are submitted
  in the closing seconds of the auction (a practice
  called sniping).
• For example, in a sample of over 1000 completed
  e-Bay auctions (May June 1999), 28 had 0
  bidders, 16 had exactly 1 bidder, and of the
  remaining 585 auctions, 74 showed multiple
  bidding and 18 had bids in the last sixty
  seconds. There is variation in the degree of last
  minute bidding with the highest percentage (56)
  in Antiques Ancient Worlds while the lowest
  (0) was in Collectibles Weird Stuff.

5
Questions

• Would a seller prefer a fixed deadline (eBay) or
  a flexible deadline (amazon)?
• What kind of seller would post a buy it now
  price?
• Does proxy bid system cause shaded bids?
• Why does sniping occur?

6
Petrol Ofisi

• Privatisation of gas station chain in Turkey (50
  market share)
• 10,000 for tender documents
• 20m in collateral
• 22 applications but 4 bidders
• English auction

7
Questions

• What are the benefits of an entry fee?
• What is the role of the indicative round? How
  should a bidder bid?
• Common vs private value?

8
Warner-Lambert/AHP/Pfizer

• W-L and AHP wanted to merge
• Agreed to a break-up fee
• Pfizer wanted to take over W-L but was prevented
  by a prior agreement
• Pfizer argued that rejection of their bid was not
  in the interests of W-L shareholders

9
Questions

• Will break up fees prevent the highest value use?
• Should it have used an open auction?
• How about for MA in general?

10
Case Analysis

• Google

11
Internet Advertising

• Ad revenue from auctions
• Google (2004) 3b
• Yahoo (2004) 1.7b
• Early (pre-1997!) models
• Sold per-impression
• Flat fees to show ads a fixed number of times
• Slow entry, large customer spend

12
Overture (then GoTo and now Yahoo!)

• Generalised First-Price Auctions (1997)
• Advertiser submitted a per click bid
• They paid this every time the ad was clicked.
  Sold one click at a time (rather than 1000
  blocks)
• Highest bids were most prominent
• Ease of use, more entry
• Example
• 2 slots on a page and 3 bidders
• 1st slot gets 200 clicks (per hour), 2nd 100
  clicks
• Bidder values per click of 10, 4 and 2
• If B2 bids 2.01 is gets a slot and so B1 need
  not bid more than 2.02 to get 1st slot. But in
  this case, B2 will find it worthwhile to increase
  their bid to 2.03, etc.
• No pure strategy equilibrium.

13
Googles Rules

• Each bidder can specify a rich rule for
  determining how to bid as a function of the
  search terms and the site from which the search
  originates.
• Google sets a reserve or minimum price for each
  search term.
• Google estimates the click-through rate that
  each bidder would have if it were listed in the
  first spot.
• Google ranks the bids according to the product of
  the clickthrough-rate and the bid it assigns ad
  spots in that order.
• Google is paid only if a bidders link is
  clicked. In that case, it receives the smallest
  price the bidder could have bid to get its
  ranking.
• Generalised second price auction
• Yahoo switched to this.

14
Example (no CTR adjustment)

• If 3 bidders, act truthfully, bid 10, 4 and 2.
• Payments will be 4 and 2 so their total
  payments are 800 and 200 respectively.
• No incentive to change their bids so it is an
  equilibrium.

15
Example (with CTR)

• (from Paul Milgrom) I bid 1 and have an
  estimated click-through rate of .50.
• You have bid 2 and have an estimated click
  through rate of .2.
• The reserve price is 0.1.
• My score is .5 yours is .4, so my ad ranks
  first.
• I could have won with a bid as low as .81, so
  that is what I pay if my link is clicked.
• You could have had your spot for as low as .1, so
  that is what you pay if your link is clicked.

16
Questions

• What are the key properties of this mechanism?
• Why is Google paid only for clicks?
• By comparison, television ads are priced
  according to impressions (how many times they
  are seen).
• By comparison, consignment stores are paid
  according to sales.
• Incentives
• Under what assumptions does Googles pricing
  scheme lead to a dominant strategy of bidding
  honestly?
• Under what assumptions is the auction outcome
  efficient?
• Under what assumptions is the auction
  revenue-maximising?
• Are these assumptions realistic?

17
Dynamics
18
Botnets strangle Google Adwords

• (by John Leyden at The Register) Security
  researchers have discovered a way to shut down or
  seriously impair a Google Adwords advertising
  campaign by artificially inflating the number of
  times an ad is displayed. By running searches
  against particular keywords from compromised
  hosts, attackers can cause click-through
  percentage rates to fall through the floor.
• This, in turn, causes Google Adwords to
  automatically disable the affected campaign
  keywords and prevent ads from being displayed. By
  disabling campaign keywords using the technique,
  cybercrimals could give their preferred parties
  higher ad positions at reduced costs, according
  to click fraud prevention specialists Clickrisk.
• "By disabling targeted keywords across many
  advertisers' campaigns simultaneously by
  artificially inflating the number of times an ad
  is displayed an attacker can secure a higher ad
  position," explains Clickrisk.com chief exec Adam
  Sculthorpe. The attack - dubbed keyword hijacking
  - is difficult to prevent because it takes
  advantage of a design feature of Google Adwords
  rather than a flaw, he added. Clickrisk came
  across the attack in investigating why the click
  through rates of one of its clients - which had
  been running at a steady rate - dropped to zero
  for no apparent reason. Subsequent monitoring and
  forensic testing revealed that a botnet made up
  of open proxies in China was responsible for the
  attack.
• Highcost-per-click (CPC) advertisers in niche
  markets are particular vulnerable to the keyword
  hijacking attack. "Once keywords are disabled
  they can't be re-enabled and attacks can go
  undetected for some time," Sculthorpe told El
  Reg. When keywords are disabled an advertiser
  must erase all campaigns featuring the affected
  keywords and create a new campaign as a
  workaround.
• Although the true scope of the problem remains
  unclear, Clickrisk security analysts believe the
  keyword hijacking attack may be widely exploited.
  Clickrisk advises users to monitor click-through
  rates and traffic levels, log into Google Adwords
  campaign frequently and check that keywords are
  not disabled.
• The incidence of click fraud risk exposure in
  general is on the rise. According to Clickrisks
  chief risk officer, Jack Bensimon, "our clients
  have experienced substantial losses ranging from
  20 65 per cent of their total click costs."
  Bensimon believes that "managing business risk is
  a critical component of online advertising" and
  further recommends that "online marketers should
  be vigilant and regularly monitor keywords".

19
Towards a Better Design

• If the assumptions are unrealistic,
• how might a bidder exploit the differences?
• how might a competitor create a better design?
• What constraints, if any, would you expect to be
  imposed on your design
• by competition from other search engines?
• by legal considerations?
• How might Google accommodate market power of
  advertisers?
• Quantity discounts?

20
VCG Auctions

• Vickery-Clarke-Groves Auction
• With one slot, identical to GSP auction
• VCG charges bidder i the externality they impose
  on others (their decrease in value of clicks
  because of is presence).
• Example
• B2 still pays 200. B1 now pays 600 (rather than
  800) 200 (externality on B3, pushed out)
  400 (externality on B2, pushed to second)
• VCG has lower revenues when bidders tell truth
• But under GSP bidders may not tell truth.

21
Sample Assumptions

• Bidders value clicks
• Without regard to the source of the click
• Without regard to the position of the ad
• The auction is honest and trusted
• Click through rates are genuine
• No shill bidders, false clicks, manipulated
  prices
• Private values (no adverse selection)
• Searcher behavior
• Searcher clicks only on the first listing
• Searcher clicks on the first relevant listing
• Searcher inspects the top two listings equally

22
Additional Applications
23
Uses of Novel Auctions

• Simultaneous design adopted for spectrum auctions
  in Australia, Canada, Germany, Guatemala, Mexico,
  US
• Variant with price-hour-quantity bids and a new
  activity rule adopted for day-ahead electric
  power in California
• Variant proposed for standard offer electrical
  service auctions
• Procurement auction with endogenous market
  structure proposed by GTE for universal
  telephone service in the US.

24
Wholesale Electricity Markets
25
Resistance to Change

• Cannot use decentralised allocation mechanisms
• Market theory supply does not equal demand.
  Particularly costly in electricity.
• But does this really mean you cant use a
  market-based process of decision-making?

26
The Balance Requirement

• Technical requirement that supply onto
  transmission exceed demand
• Why is this an economic problem?
• Losses from mismatches
• Comparing apples and ambulances
• Comparing ambulances and tow trucks

27
Characteristics of Electricity

• Nonstorability
• Lack of substitutes
• Difficulty in using price signals
• Localised effects

28
Achieving Balance

• Informational inputs
• Demand forecasts
• Costs and generating plant availability
• Transmission loss information
• Weather forecasts?
• Network operator forms dispatch schedule
• Determines reserve capacity requirement

29
Economic rewards

• Who provides the information?
• What are their incentives to report accurately?
• What are their incentives to behave efficiently?

30
Vertical Integration

• Internal arrangements to gather relevant
  information
• Internal accounting arrangements link costs to
  individual performance -- no necessary economic
  conflict in cost apportionment
• Not readily amenable to competition

31
Contractual Arrangements

• Network operator requires information from
  participants
• Procedures for re-balancing
• Procedures for handling constraints
• Generators and retailers agree separately to how
  costs are apportioned using long and medium-term
  contracts

32
Electricity Pools

• Simply another form of contractual arrangement
• Fixed rules
• like contractual arrangements
• unlike vertical integration
• Lots of cost information
• unlike contractual arrangements
• like vertical integration

33
Pools vs. Vertical Integration

• Information problems under vertical integration
• Incentives to minimise costs
• Incentives to negotiate toughly on input supply
  contracts
• Fiscal constraints
• Difficulties in financing new investments

34
Pools vs. Contracts

• Information provided to operator
• richer amount under pools
• only quantity information with contracts
• Pool can be augmented by contractual arrangements
• no compulsion to supply information
• can make long-term financial arrangements if so
  choose -- to minimise risk

35
Pool Bidding

• Each generator bids in supply schedule
• How much it will supply at each price
• ISO uses bids to form industry supply schedule to
  meet demand
• Minimise costs treating bids as cost
• Generates dispatch schedule
• Takes into account inter-regional transmission
  constraints
• System marginal price is the bid of the last unit
  dispatched
• All dispatched units get this price
• This is done every 5 minutes

36
Simple Pool Model
37
Truthful Cost Revelation

• Suppose demand is q 3
• Optimal dispatch A 2, B 1
• Marginal cost bidding A 2, B 1
• Payments to each of 3 per unit.
• Achieves allocative and productive efficiency.

38
Equilibrium Behaviour

• Suppose demand 2 units
• MC Bidding either A 2 or (A 1, B 1)
• Price equals 3
• A earns 1 and B earns 0
• Can either do better?
• If A raises bid on second unit to 4, means only
  dispatched for one unit.
• If B raises bid, then is not dispatched at all
• Neither can do better -- an equilibrium!

39
Market Power

• Suppose (again) that demand 3 units
• MC Bidding
• A earned 1 and B earned 0
• Can either generator do better?
• If A bids second unit at 4, then earns 2 extra.
• MC bidding not an equilibrium
• New equilibrium
• A bids (2, 5.9) B bids (3, 6)
• Resulting price equals 5.9

40
Value of Contracting

• In economics, consider value of contracting by
  comparing it to the appropriate counterfactual
  ...
• 1. Pool price exposure if do not contract with
  that generator, what pool price will you face?
• 2. Generator exit if pool prices insufficient to
  cover fixed generation costs, generator may exit
  without contract.

41
Pool Price Exposure

• If you do not contract with the generator what
  will pool prices be?
• Pool market equilibria at the contract time, how
  many generating companies can bring pool price to
  VoLL by not bidding in capacity?
• If none, pool prices will equal system marginal
  cost
• If one or more, pool prices may go to VoLL
• If generator can bring pool to VoLL, want to
  contract with that generator.

42
Maximum Contract Price

• You should pay a generator no more than it brings
  to value in the market.
• This requires calculating the reduction in
  value that occurs if generator is not in the
  market.

43
Graphically

Industry Supply
VoLL
Demand
Quantity
44
Added Value

Industry Supply
VoLL
Gens Added Value
Demand
Quantity
45
Added Value

• The added value of a generating company is the
  maximum amount of profits it can expect to earn
  from either contracting or the pool market.
• Here a retailer may contract with the generator
  for VoLL price on all of its capacity. This may
  prevent it paying a VoLL price to other
  generators.

46
An Example

• Two generators A and B
• A B
• Capacity 20 10
• Marginal Cost 2 5
• VoLL 100

47
Added Value (Demand 10)

As AV 30 Bs AV 0
100
5
2
10
20
Quantity
48
Added Value (Demand 15)

As AV 520 Bs AV 0
100
5
2
10
20
Quantity
49
Added Value (Demand 25)
As AV 1485 Bs AV 475

100
5
2
10
20
Quantity
50
Exit Option

• Sometimes, a generator may not have sufficient
  value added to keep operating.
• Nonetheless, it may bring sufficient value to
  retailers even if it does not bring sufficient
  value to the market.
• A generator has some value if it reduces the
  added value of other generators.

51
Entry Value (Demand 15)

As AV 520 Bs EV 950
100
5
2
10
20
Quantity
52
Additional Considerations

• Demand uncertainty
• Impact of outages
• Financial analysis determining option values in
  an electricity market

53
Rail Auctions
54
Rail Reform

• Sweden wanted to use a decentralised rail use
  allocation processes.
• Consultants objected (Coopers Lybrand)
• These train paths cannot be treated as
  independent units, since they are not
  interchangeable, and depend on the specification
  of all other paths in the integrated timetable.
  There is therefore no common unit of capacity on
  a mixed-use railway which can be allocated to
  owners, prices and traded among a number of
  buyers and sellers.
• However, a simple free auction cannot be used for
  railway capacity since there are no independent
  units of capacity to bid for. The viability of
  every bid to operate a train service depends on
  the specification of every other train service
  which has been bid for.

55
Proposed Times
C
D
SYD
CBR
MEL
H
G
A
B
E
F
time
56
Feasible Schedules
C
D
SYD
CBR
MEL
H
G
A
B
E
F
time
57
Feasible Schedules
C
D
SYD
CBR
MEL
H
G
A
B
E
F
time
58
Which schedule should be chosen?

• Can work out what is feasible? But what is most
  efficient?
• Answer use simultaneous ascending/combinatorial
  auction.
• Bid for each particular route
• Bid for route combinations
• Choose schedule among these that maximises revenue