Clicks and Mortar

Efficiency and the Internet

Price Discrimination
Uniform versus non-uniform pricing
Possibility of arbitrage Uniform pricing
Uniform pricing is linear pricing
Tariff T(q)=pq
Distribution of surplus and efficiency
Types of price discrimination
First degree
Seller extracts full surplus
Second degree
Partial discrimination based on buyer self-selection into pricing
category
Third degree
Discrimination based on signal correlated with preference


Price Discrimination
First degree discrimination
Charge each customer her maximum willingness to pay
Extracts total social surplus from the market
Resulting allocation is efficient:
Let p(q) be the inverse demand function. Then the monopolist receives
p(q) for the q
th
unit sold. This the monopolists marginal revenue. Profit
maximization requires that the monopolist produce and sell to the point
where MR=MC. But this is the same condition that determines the
competitive equilibrium allocation which is efficient.
Implementation in monopoly market by two-part tariff
Let S
c
be the competitive consumer surplus
( ) | |dq p q p S
c
q
o
c
c

}
=
Price Discrimination
Graphically:
Price Discrimination
Suppose there are n buyers each of whom has the
same demand schedule.
The monopolist offers a two-part tariff of the form


The profit per unit sold is then



where C(q) is the monopolists marginal cost
( )
if
if

=
> +
=
0 0
0
q
q
n
S
q p
q T
c
c
( ) ( ) q C
n
S
q p q
c
c
' + = ' t
Price Discrimination
Total profit is obtained by integrating the marginal
profit with respect to q:


But this is just the total surplus in the market.
It is straightforward to show that the profit the monopolist
obtains exceed what she would have gotten at the uniform
monopoly price.
Difficulties with implementing first-degree
discrimination
Lack of knowledge about demand
Heterogeneity of demand
( ) q C S q p
c
c c
+ = t
Price Discrimination
Second-degree price discrimination
Applicable when buyers are heterogeneous and seller has limited
information about preferences
Uses a menu of non-linear tariffs to allow buyers to self-select into
a pricing scheme (personalized pricing)
Two-part tariff is a simple example of non-linear pricing scheme
Digital goods implementation in the form of versioning
Price Discrimination
Tie-in Sales
Bundling of complementary goods or services leads naturally to a
two-tier pricing system
Cameras and film
Amusement parks and rides
Online news subscriptions and access to archived material
Information tracking and analysis capabilities of the web
Flip side of targeted advertising
Track buyer preferences
Conduct price sensitivity experiments
Structure pricing tariffs according to data collected
Dark Side: Privacy Issues

Price Discrimination
Third-degree price discrimination
Monopolist is able to segment the market using external signals
about buyer types
Signals:
Age
Sex
Occupation
Location (or referring site)
New vs. repeat purchases
The monopolist then sets a uniform price in each market segment
to maximize profits from each segment.
Price Discrimination
Model
N market segments
p
i
= price in segment i, q
i
= quantity sold in segment i
D
i
(p
i
) = segmented demand function
q = E
i
D
i
(p
i
)
Assuming a uniform cost function across segments, the monopolists
profit maximization is then to choose prices for each market segment
to solve the problem
| |
( ) ( )|
.
|

\
|

=
i
i i
i
i i i
p p p
p D C p D p
N
,...,
1
max
Price Discrimination
The first-order conditions for this problem can be manipulated into
the form








The optimal pricing rule then is for the monopolist to set the markup
over marginal cost (as a percentage of the price) equal to the inverse
of the elasticity of demand.
( )
( )
( )
i i
i i i
i
i i
i
p D
p D p
p
q C p
'
=
=
'
c
c
where
1
Price Discrimination
Some implications of the markup rule
Market segments with higher demand elasticity will receive a lower
price

Greater price sensitivity market segments get lower prices

Conversely, segments which are less price sensitive will pay higher
prices
Welfare analysis for simple cases shows that the overall effects of
market segmentation are ambiguous. Depending on how price sensitive
different segments are relative to each other, overall consumer surplus
may be larger or smaller with discrimination than without
Price Discrimination
Privacy Issues
Sensitivity of personal information
Medical information and insurance
Access to credit
Protection from job actions
Exposure to spam
Exposure to price discrimination
Information value-added
Customization of products
Targeting of useful information about products
Simplification of transactions
Price Discrimination
The myth of anonymity
Price Discrimination
Internet communications
Complexity of communication protocols requires tracking
information
Packet switching
Message fragmented into uniform size packets
Headers encode information about packet destination using the internet
protocol (IP) address of the recipient
Packets routed through network under control of network transmission
control protocol (TCP)
TCP checks for errors in packets and will request retransmission of
bad packets packets can be traced
Message reconstructed as packets reach destination
Cookies
Internet communication is anything but anonymous
Price Discrimination
Protecting content while revealing identity
Encryption
Secure communications
Online payment systems
Digital signatures
Trust relationships
Legal protections
Privacy guarantees and the First Amendment (freedom of speech) and
Fourth Amendment (freedom from unlawful searches)
Legal restrictions on distribution of personal information disclosed in
transactions
Truth in advertising enforcement of pledges to protect customer
privacy by firms


Price Discrimination
Market mechanisms for privacy protection
Service for information arrangements
Email
Search
Online file storage
Data analysis engines
Trust relationships
Trusted independent intermediary verifies content and claims
Provision for legal intervention by violators
Better business bureau model
Intermediation
Economic role of intermediaries
Transactional efficiencies
Lower costs in inventory holding, product delivery, insurance,
financing, accounting
Inventory and demand issues
The internet as an information aggregator and transactional role for
intermediaries in markets for digital goods
Intermediaries as Experts
Repeat purchases
Incentive to acquire knowledge about product
Intermediary as Long-term Player
Ongoing benefit to credibility
Intermediation
Intermediaries as information sources
Long-term, multi-product intermediaries and reputational
spillovers
Intermediary has incentive to ensure high quality in any given product
to avoid lost sales in other, unrelated products
Intermediary role provides a punishment mechanism in the form of
exclusion of a sellers product if quality lags
Intermediate production activities
Combining of separate products in retail bundles
Particularly germane in the information industry
News and entertainment content providers combine, package and
distribute work of individual authors
CNN, Napster
Auctions and Contracts
Market Efficiency and Competition
Contracts versus Auctions
Auctions are competitive but costly to hold when all parties to
the transaction must be present in the same place and time to
participate
Contracts are negotiated bilaterally
Less information about costs
Less competitive pricing (Ford-Autolite example)
Less flexibility if terms change
Lower cost since contract governs relationship for an extended
period of time
Auctions and Contracts
Auction Types
Direct vs. Reverse
English vs. Dutch
Sealed bid vs. open outcry
Vickerys Theorem
If buyers have the same information about an object being sold, are
risk-neutral, and have independent valuations of the object, then
any of the above auction formats will achieve maximum revenue
for the seller.
Key points:
Uncertainty about value
Independence of valuations
Auctions and Contracts
Common value auctions
Most common type of auction
Valuations are unknown but closely (or perfectly) correlated
Example: Offshore oil tracts
Example: Procurement contracts for manufactured intermediate
products
The Winners Curse
Experiment: Auctioning off a jar of money
Format
Sealed Bid
First price (i.e. highest price wins)
Auctions and Contracts
Information and the Winners Curse
Distribution of guesses








Mean guess as best estimate of actual value
Auctions and Contracts
Since the winning bid must be higher than the mean (unless all
bids are at the mean), if the mean is an accurate estimate of the true
value, then the winning bid necessarily overstates the value of the
object at auction, and the winner ends up paying too much for the
object.
Optimal bid when faced with the winners curse?
Shave bids below what you believe the true value to be
Reduces revenue to the seller
Auctions and Contracts
Reducing the risk of the Winners curse
Second-price auction
Highest bid wins, but pays second highest price
Eliminates incentive to shave bids
Open outcry auctions
Allows sharing of information among bidders as to the best guess of
the true value of the object
Multi-object auctions
Discriminatory vs. Uniform
Potential inefficiencies in sequential auctions
Auctions and Contracts
Example: 2 units to be auctioned
Buyer 1 values one unit at 10 and 2 at 20
Buyer 2 values one unit at 9 and 2 at 10
Simultaneous auction of both units
Buyer 1 wins with a bid of 10
Sequential auction: Backward induction
Suppose Buyer 1 wins in round 1
To win round 2, Buyer 1 must bid at least 9
Moving back to round 1, since Buyer 2 values one unit at 9, for Buyer
1 to win round 1, she must bid at least 9.
Buyer 1s profit from this is 20-9-9=2.
Auctions and Contracts
Now, suppose Buyer 1 loses in first round.
Buyer 1 can win in round 2 with a bid of 1, yielding a profit of 10-
1=9. Hence, Buyer 1 is better of losing in round 1.
Knowing this, Buyer 2 can win round 1 with a bid of 2. To see why,
we note the following:
Buyer 1 can get a profit of 9 by losing round1 and winning round 2.
Hence, her maximum round 1 bid, if she wins, must yield profit at least
equal to what she gets if she loses, i.e. 9.
Letting this bid be x, we need 20-9-x=9 or x=2 and buyer 2 can win in
round 1 with a bid of 2
Revenue from the sequential auction is then 2+1=3 so the sequential
auction is clearly inefficient.