Lecture 1 - Justin Rao

Lecture 1: Optimal Pricing for
Monopoly with Multiple Goods
Jacob LaRiviere
Justin Rao
1
maxπ‘₯1,π‘₯2,…,π‘₯𝑛 π‘ˆ π‘₯1 , π‘₯2 , … , π‘₯𝑛
Composite Commodity Theory
Assume there are n goods (e.g., apples,
bananas, carrots, etc…) but we really only
care about one of them (e.g., apples).
How do we handle this problem as
economists?
Lets start with the very general consumer’s
problem: constrained maximization.
Needed assumptions are
completeness, reflexivity and
transitivity.
𝑠. 𝑑. 𝑝1 βˆ™ π‘₯1 + β‹― + 𝑝𝑛 βˆ™ π‘₯𝑛 ≀ 𝐼
Suggestion: Say that we only solved this optimization problem for goods 2, …, n.
Take the first good as a parameter [e.g., like β€˜m’ in y(x)=mx+b]
max π‘₯2,…,π‘₯𝑛 π‘ˆ π‘₯2 , … , π‘₯𝑛 ; π‘₯1
𝑠. 𝑑. 𝑝2 βˆ™ π‘₯2 + β‹― + 𝑝𝑛 βˆ™ π‘₯𝑛 ≀ 𝐼መ
This leads to a bunch of solution functions for all of the parameters…
π‘₯2βˆ— 𝑝2 , … , 𝑝𝑛 , 𝐼መ , … , π‘₯π‘›βˆ— 𝑝2 , … , 𝑝𝑛 , 𝐼መ
መ we’re netting out expenditures on goods 2, 3, …, n.
NOTE: I now is 𝐼;
π‘₯2βˆ— 𝑝2 , … , 𝑝𝑛 , 𝐼መ , … , π‘₯π‘›βˆ— 𝑝2 , … , 𝑝𝑛 , 𝐼መ -> π‘ˆ π‘₯1 , π‘₯2 , … , π‘₯𝑛
Composite Commodity Theory
Idea: with these β€œsolution functions” for the
goods we don’t care about, lets plug them
back in to the original utility function…
Call this new function β€œV” and let x1 vary again:
π‘ˆ π‘₯1 , π‘₯2 , … , π‘₯𝑛 = π‘ˆ π‘₯1 , π‘₯2βˆ— 𝑝2 , … , 𝑝𝑛 , 𝐼መ , … , π‘₯π‘›βˆ— 𝑝2 , … , 𝑝𝑛 , 𝐼መ
መ 𝑝2 , … , 𝑝𝑛
= 𝑉 π‘₯1 , 𝐼;
This reduces the problem to a function of a
bunch of parameters (e.g., m and b) rather
than variables (e.g., y and x).
This is useful; parameters aren’t
complicated but variables are!
Noting that the prices of goods 2, 3, …, n are fixed, consider maximizing V
መ 𝑝2 , … , 𝑝𝑛
maxπ‘₯1,𝐼መ 𝑉 π‘₯1 , 𝐼;
s.t. 𝑝1 βˆ™ π‘₯1 + 𝐼መ ≀ 𝐼
maxπ‘₯1,π‘₯2 𝑉 π‘₯1 , π‘₯2
Composite Commodity Theory
We can call 𝐼መ the composite commodity and
evaluate β€œeverything we don’t care about” as
it and now call it x2.
NOTE:
Effectively we’ve normalized the
cost of the composite commodity
to $1
s.t. 𝑝1 βˆ™ π‘₯1 + π‘₯2 ≀ 𝐼
This problem has a solution where for different levels of p1 and I there are
different solutions for π‘₯1 and π‘₯2 .
π‘₯1βˆ— 𝑝1 , 𝐼 , π‘₯2βˆ— 𝑝1 , 𝐼
Finally, plot π‘₯1βˆ— 𝑝1 , 𝐼 against 𝑝1 and voila! You have a demand curve…
…sum them over all consumers and you have a market demand curve!
How do monopolists price goods?
Monopolists, like all firms,
should price to maximize
profits.
As a result, the demand curve and costs matter
Monopolists, like all firms,
should price to maximize
profits.
As a result, the demand curve and costs matter
Perfect Comp: MB = MC = P
(other firms can undercut price otherwise)
Monopolists, like all firms,
should price to maximize
profits.
As a result, the demand curve and costs matter
Monopolist doesn’t have to worry about
competitors -> set Q such that MC = MR
Monopolists, like all firms,
should price to maximize
profits.
As a result, the demand curve and costs matter
Monopolist doesn’t have to worry about
competitors -> set Q such that MC = MR
How to construct MR?
To sell another unit, must lower the price so
firm loses money on units they were selling
(intensive margin) and gains money from
additional units sold (extensive margin)
MR = extensive gains – intensive losses
maxπ‘ž πœ‹ π‘ž = 𝑃 π‘ž π‘ž βˆ’ 𝑇𝐢(π‘ž)
Monopolist’s math
Maximizes profits (TR – TC) by setting a quantity
and charging needed price to have market clear
e.g., price is a function of quantity: P(q) and
note the TR = P(q)*q
f.o.c.: 𝑃′ π‘ž βˆ— π‘ž βˆ— + 𝑃 π‘ž βˆ— βˆ’ 𝑇𝐢 β€² π‘ž βˆ— = 0
𝑃′ π‘ž βˆ— π‘ž βˆ— + 𝑃 π‘ž βˆ— βˆ’ 𝑀𝐢 π‘ž βˆ— = 0
𝑃′ π‘ž βˆ— π‘ž βˆ— + 𝑃 π‘ž βˆ— = 𝑀𝐢 π‘ž βˆ—
NOTE: P’(q) < 0 since demand slopes downward
𝑃′ π‘ž βˆ— π‘ž βˆ— Intensive margin loss as q increases
𝑃 π‘žβˆ—
Extensive margin gain as q increases
𝑃′ π‘ž βˆ— π‘ž βˆ— + 𝑃 π‘ž βˆ— = 𝑀𝐢 π‘ž βˆ—
Monopolist’s math
𝑑𝑃(π‘žβˆ— ) βˆ—
π‘ž
π‘‘π‘ž
+ 𝑃(π‘ž βˆ— ) = 𝑐
Note that P’(q) is the slope of the demand curve
and that economists think about elasticities rather
than slopes.
𝑑𝑃(π‘ž βˆ— ) βˆ—
𝑃(π‘ž ) βˆ’ 𝑐 = βˆ’
π‘ž
π‘‘π‘ž
Assume MC(q) = c for simplicity (e.g., constant MC)
Divide both sides by P
𝑃(π‘ž βˆ— ) βˆ’ 𝑐
𝑑𝑃(π‘ž βˆ— ) π‘ž
=βˆ’
𝑃(π‘ž βˆ— )
π‘‘π‘ž 𝑃(π‘ž βˆ— )
Lerner Equation: If demand curve is relatively
inelastic, charge a high markup.
NOTE 1: Assume that elasticity is constant along all
portions of the demand curve for convenience.
NOTE 2: β€œConstant Elasticity” makes demand curve nonlinear but a perfectly valid assumption.
NOTE 3: πœ– refers to own price elasticity unless otherwise
noted.
βˆ—
𝑑𝑃(π‘ž βˆ— )
𝑃(π‘ž βˆ— ) βˆ’ 𝑐
% Δ𝑃
1
1
𝑃(π‘ž βˆ— )
=βˆ’
=βˆ’
=βˆ’
=βˆ’
% Ξ”π‘ž
π‘‘π‘ž
𝑃(π‘ž βˆ— )
% Ξ”π‘ž
πœ–
% Δ𝑃
π‘ž
𝑷(π’’βˆ— ) βˆ’ 𝒄
𝟏
=
βˆ’
𝑷(π’’βˆ— )
𝝐
p
MC
Dead
Weight
Loss
pm
Monopoly
Profits
D
q
qm
qc
MR
12
Choosing Quantity
β€’ Marginal Revenue, the increment to revenue from a increase in
quantity sold
MR ο€½ p(q)  qpο‚’(q)
β€’ Elasticity of demand: tells you the % change in quantity for a 1%
change in price
π‘‘π‘ž
%Ξ”π‘ž
π‘‘π‘ž 𝑝
1 𝑝
π‘ž
πœ–=βˆ’
=βˆ’
=βˆ’
=βˆ’
𝑑𝑝
%Δ𝑝
𝑑𝑝 π‘ž
𝑝′(π‘ž) π‘ž
𝑝
13
Examining the elasticity function
1 𝑝(π‘ž)
πœ–=βˆ’ β€²
𝑝 π‘ž π‘ž
First derivative of the
demand curve. At any
point π‘ž0 it gives the
slope at that point.
The demand curve.
Gives the price as
function of q.
Relative levels of price
and quantity
14
Special case: linear demand
1 𝑝(π‘ž)
πœ–=βˆ’
π‘ π‘™π‘œπ‘π‘’ π‘ž
First derivative is
constant
At high prices, q is low.
A 1% change in p is
relatively large,
especially compared to
quantity. Elasticities will
be relatively large.
At low prices, p will be
small and q is large.
Elasticities will
mechanically be low.
15
Some general facts about elasticity
β€’ At high prices, a fixed % change in price is larger in level terms. Since
q will tend to small, a given level change in q, will be a larger in
percentage terms.
β€’ This creates a relationship such that elasticities will tend to be high at
the β€œtop of the demand curve” and low at the bottom.
β€’ When we think of β€œsmall changes” in price, the impact of raising and
lowering will be symmetric. However, for larger changes, e.g. 10%, an
increase and decrease need to have the same magnitude impact
16
Elasticity and total revenue
Total revenue (TR) = p(q)*q
Product Rule
Algebra
Substituting in
TR d  pq  qdp  pdq dp 
pdq οƒΆ
1 
οƒ·οƒ· ο€½ %p1 ο€­ ο₯ 
%TR ο€½
ο€½
ο€½
ο€½
TR
pq
pq
p  qdp οƒΈ
Translating to calculus
Negative of this is elasticity
17
Elasticity and total revenue
Total revenue (TR) = p(q)*q
TR d  pq  qdp  pdq dp 
pdq οƒΆ
1 
οƒ·οƒ· ο€½ %p1 ο€­ ο₯ 
%TR ο€½
ο€½
ο€½
ο€½
TR
pq
pq
p  qdp οƒΈ
β€’ One minus the elasticity translates a price increase in percent to a
revenue increase.
β€’ For example, if the elasticity is 3.5, a 1% price increase causes a -2.5%
impact on revenue (a loss).
18
Elasticity and total revenue
%TR ο€½ %p1 ο€­ ο₯ 
β€’ Elasticity = 1 οƒ  small changes in price do not impact revenue
β€’ Elasticity < 1 οƒ  price drops lower revenue, price increases raise
revenue
β€’ Elasticity>1 οƒ  price drops raise revenue
19
Inverse Elasticity Rule
β€’ Profit Max (MR=MC)
MR
MC

0ο€½
ο€½ p  qpο‚’(q ) ο€­ c' (q )
ο‚Άq
𝑝 βˆ’ 𝑐 β€² π‘ž = βˆ’π‘žπ‘β€² π‘ž
𝑝 βˆ’ 𝑐′(π‘ž)
π‘ž β€²
1
=βˆ’ 𝑝 π‘ž =
𝑝
𝑝
πœ–
20
MR
Inverse Elasticity Rule
β€’ Profit Max (MR=MC)
MC

0ο€½
ο€½ p  qpο‚’(q ) ο€­ c' (q )
ο‚Άq
β€’ Price-cost margin (Lerner index) = 1 over elasticity
p ο€­ cο‚’(q ) ο€­ qpο‚’ 1
ο€½
ο€½ .
p
p
ο₯
β€’ Price minus marginal costs divided by price is referred to as gross
margin.
21
Inverse Elasticity Rule
p ο€­ cο‚’(q ) ο€­ qpο‚’ 1
ο€½
ο€½ .
p
p
ο₯
β€’ Suppose MC=0. Then quantity is chosen so that elasticity is 1.
p
ο€­ qpο‚’ 1
ο€½1ο€½
ο€½
p
p
ο₯
β€’ Intuition: if marginal costs are zero, then optimize for revenue. Total
revenue grows until elasticity = 1.
β€’ If MC>0, one will β€œstop” before reaching elasticity = 1.
22
p
MC
Dead
Weight
Loss
pm
Monopoly
Profits
D
q
qm
qc
MR
23
Digression on Margin Formula
β€’ Perfect competition says price=MC, or zero markup, which implies
elasticity of infinity. In other words, by deviating from market price I
can sell all my units. What are some examples where this is
approximately true in practice?
β€’ In general MC will depend on price. Cannot in general say β€œwhat are
my marginal costs” to get optimal mark up
24
Markup formula cont.
πœ–
𝑝=
𝑐′ π‘ž
πœ–βˆ’1
β€’ Seems to say β€œfixed markup on marginal costs”, but elasticity will depend on
the demand function, so will not be fixed across a firm’s products and across
customers.
β€’ Optimal pricing is much more complicated than a fixed markup
25
Markup formula cont.
πœ–
𝑝=
𝑐′ π‘ž
πœ–βˆ’1
β€’ Markup > 1
β€’ Elasticity will generally depend on q, costs depend on q
β€’ With constant elasticity, firm passes on more than 100% of cost or tax (a
tax is like a marginal cost, firm β€œmarks up the tax”)
β€’ Works at firm level, with elasticity measured at firm, not industry
β€’ Does not capture competitor reactions
26
Monopoly Pricing Formula
β€’ Prices depends on elasticity, which depends on the product, customer
characteristics
β€’ Offer discounts to elastic (price sensitive) customers
β€’ Discounts offered on basis of factors correlated with price sensitivity
β€’ Pricing can often be understood by how decision variables correlate
with price sensitivity
27
Pricing Multiple Goods
28
Basic idea
β€’ One firm selling multiple goods
β€’ A firm’s goods will, in general, β€œcompete with each other” to some
degree
β€’ A rational firm takes this into account when setting price. Optimal
price will depend on own price elasticity (what we just learned about)
and cross price elasticities
29
Pricing Related Goods
Review: substitutes and complements
β€’ Complements and Substitute Products (Relationship)
ο‚§
ο‚§
Sales of a good rise when the price of a complement falls
β€’
Console and games
β€’
Drinks and food at a restaurant (e.g. happy hour to attract customers)
Sales of a good fall when the price of a substitute falls
β€’
Games vs. other games
β€’
Food at a restaurant
ο‚§
Lower price of substitute cannibalizes demand from other product
ο‚§
Lower price of complement promotes sales of other product
ο‚§
Prices of substitutes (complements) are higher (lower) than standalone profit-maximizing prices
30
Inverse Elasticity Rule 2
β€’ Suppose we sell n goods indexed i=1,…,n
β€’ Demands xi(p)
β€’ Profit
n
 ο€½ οƒ₯i ο€½1  pi ο€­ mc(i ) xi (p).
β€’ If we assume constant marginal cost, this simplification is an example of selling
the same good in multiple markets or to multiple customer β€œtypes”
p j dxi
β€’ Cross-price elasticity
ο₯ij ο€½
xi dp j
.
β€’ Note no minus sign.
β€’ Positive οƒ  substitutes; Negative οƒ  complements
31
Representative Consumer Assumption
β€’ If there is a representative consumer maximizing utility:
max u(x)-px, so u ο‚’(x) ο€½ p and u ο‚’ο‚’(x)dx ο€½ dp
β€’ Thus there are symmetric cross-derivatives
ο‚Άxi ο‚Άx j
ο€½
ο‚Άp j ο‚Άpi
Recall this rule from
multivariate calculus
From the total
derivative of FOC
This rule need not hold in
practice, but is a commonly
made assumption
32
In Matrix Notation
pi ο€­ mc(i )
,
Price cost margin: Li ο€½
pi
ο‚Άx j
ο‚Άxi

n
n
0ο€½
ο€½ xi  οƒ₯ j ο€½1 ( p j ο€­ mc(i ))
ο€½ xi  οƒ₯ j ο€½1 ( p j ο€­ mc(i ))
ο‚Άpi
ο‚Άpi
ο‚Άp j

οƒΆ
n ( p j ο€­ mc (i ))
οƒ·
ο€½ xi 1  οƒ₯ j ο€½1
ο₯
ij

οƒ·
p
j

οƒΈ
0 = 1 + E L, and thus L = - E-1 1
33
Rule for inverting a
2x2 matrix
Two Good Formula
β€’ L = - E-1 1 yields
π‘³πŸ
π‘³πŸ
π’†πŸπŸ
= 𝒆
𝟐𝟏
π’†πŸπŸ
π’†πŸπŸ
π’†πŸπŸ βˆ’ π’†πŸπŸ
π‘³πŸ = βˆ’
π’†πŸπŸ π’†πŸπŸ βˆ’ π’†πŸπŸ π’†πŸπŸ
=βˆ’
π’†πŸπŸ
Factor out π’†πŸπŸ
βˆ’πŸ
𝟏
𝟏
Divide top
and bottom
by π’†πŸπŸ
π’†πŸπŸ
π’†πŸπŸ
=
π’†πŸπŸ π’†πŸπŸ
π’†πŸπŸ βˆ’
π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ
π’†πŸπŸ
πŸβˆ’
πŸβˆ’
𝟏
π’†πŸπŸ
π’†πŸπŸ
=βˆ’
π’†πŸπŸ π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
𝒆
𝟏𝟏
πŸβˆ’
πŸβˆ’
π’†πŸπŸ π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
Multiply top
and bottom
by π’†πŸπŸ
34
Two Good Formula
β€’ L = - E-1 1 yields
𝟏
L𝟏 = βˆ’
π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ π’†πŸπŸ
will be between 0 and 1 because π’†πŸπŸ π’†πŸπŸ < π’†πŸπŸ π’†πŸπŸ
This is because cross price elasticities have to be smaller than the
relevant own price elasticities.
35
Two Good Formula
β€’ L = - E-1 1 yields
𝟏
L𝟏 = βˆ’
π’†πŸπŸ
𝟏
=βˆ’
π’†πŸπŸ
π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ π’†πŸπŸ
+
πŸβˆ’
βˆ’ =βˆ’ 𝟏 >𝟏
++
π’†πŸπŸ 𝟎 βˆ’ 𝟏
πŸβˆ’
βˆ’βˆ’
36
Two Good Formula for Substitutes
β€’ L = - E-1 1 yields
𝟏
L𝟏 = βˆ’
π’†πŸπŸ
𝟏
=βˆ’
π’†πŸπŸ
π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ π’†πŸπŸ
+
πŸβˆ’
βˆ’ =βˆ’ 𝟏 >𝟏
++
π’†πŸπŸ 𝟎 βˆ’ 𝟏
πŸβˆ’
βˆ’βˆ’
37
Two Good Formula for Substitutes
β€’ L = - E-1 1 yields
𝟏
L𝟏 = βˆ’
π’†πŸπŸ
π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ
π’†πŸπŸ π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ π’†πŸπŸ
38
Two Good Formula for Complements
β€’ L = - E-1 1 yields
π’†πŸπŸ
𝟏 πŸβˆ’ βˆ’
L𝟏 = βˆ’
π’†πŸπŸ 𝟎 βˆ’ 𝟏
39
Two Good Formula Review
β€’ L = - E-1 1 yields
𝟏
=βˆ’
π’†πŸπŸ
π’†πŸπŸ
πŸβˆ’
π’†πŸπŸ
𝒆 𝒆
𝟏 βˆ’ 𝟏𝟐 𝟐𝟏
π’†πŸπŸ π’†πŸπŸ
π’†πŸπŸ > 𝟎, goods are substitutes. A price decrease on product 2 decreases sales on product
1 (go in same direction)
π’†πŸπŸ < 𝟎, goods are complements. A price decrease on product 2 increases sales on
product 1 (go in opposite directions)
π’†πŸπŸ & π’†πŸπŸ will be negative due to law of demand (note before we β€œembedded the
negative sign)
40
Bundling
β€’ Pure bundling: only sell the bundle
β€’ Cars & tires
β€’ Cable TV
β€’ Cars + feature β€œmodels”
β€’ Mixed bundling: sell separately with a discount for bundle
β€’ Video games w/ console
β€’ Sports passes (clubs, ski resorts)
41
Enormous
Bundle
42
Utilities with Independent Values
Action
Utility
Buy Nothing
0
Buy Good 1
v 1 – p1
Buy Good 2
v 2 – p2
Buy Both
v1+v2 – pB
43
v2
Buy neither, but would buy a
bundle. Starting from the top right
of the square, there is always
some bundle I want to offer
pB
Buy Good 2
Buy Both
p2
Buy Nothing
Buy Good 1
p1
v1
44
v2
If p2 optimal, this price reduction
doesn’t affect profits – sales gains just
balance price cut
pB
Buy Good 2
Buy Both
p2
Buy Nothing
Buy Good 1
p1
v1
45
v2
pB
Buy Good 2
Buy Both
p2
If p1 optimal, this price reduction
doesn’t affect profits – sales gains just
balance price cut
Buy Nothing
Buy Good 1
p1
v1
46
v2
pB
Buy Good 2
Buy Both
p2
Reducing bundle price gives the
additional sales of both goods with a
single price cut
Buy Nothing
Buy Good 1
p1
v1
47
Conceptualizing bundles
β€’ A bundle can be thought of as a β€œconditional discount”. E.g. If you buy
good 1, I’ll give you a discount on good 2.
β€’ This lets me give β€œtargeted offers”
β€’ Especially powerful when my valuation of good 2 is much lower if I
already have good 1. E.g. gym memberships bundle many partner
locations for a small increase in price because otherwise people
would rarely buy more than 1.
48
Bundling as a part of corporate strategy
β€’ Rethink Product
β€’ Jack Walsh noticed GE made more profit on engine service than
aircraft engines
β€’ Redefined product: sell engines in order to sell service
β€’ Rather than selling service to make engines more attractive
β€’ Very profitable
β€’ IBM pivoted from providing software and services to sell hardware
β€’ IBM Global Services
β€’ Hardware margins typically low except Apple
49
Bundling Insights
β€’ Mixed bundling is always more profitable than no bundling
β€’ With independent or negatively correlated goods
β€’ Better for consumers as well!
β€’ Often a β€œgrand bundle” does well for firms, but can be bad for
consumers
β€’ Skims out the most willing-to-pay with a β€œsuper good”
β€’ Bundles can be used to help customers
50