Econ 5700
Lecture 10: Cartels
Econ 5700 SP20
Prof. Adam Dearing
The Ohio State University
1 / 15
Agenda
1. Cartels
2. Detecting Collusion
2 / 15
Last Time
I Last time: Stackelberg and Repeated Games
I We saw that collusion can be sustained in an infinitely repeated game
I Requires threat of future punishment
I Firms must be sufficiently “patient”
I We used Grimm-Trigger punishment strategy
I But how do firms determine the strategy?
3 / 15
X
Tacit Collusion
I Tacit collusion: Firms collude without explicitly communicating
I Both independently think of the punishment strategy
I They just happen to agree upon it
I Then we observe the collusion
I Tacit collusion is not illegal in the U.S.
I But explicit coordination is illegal
4 / 15
Cartel
I Cartel: explicit coordination by firms to raise prices
I Why do it? To raise profits!
I Repeated games analysis from last time applies to both cartels and tacit collusion
I We only care about how they arrive at the strategies for legal reasons
I Economic consequences are the same either way I This is the idea of “cheap talk”
5 / 15
Cartel Formation
I What facilitates cartel formation?
1. Ability to raise market price
I Relatively inelastic demand I No/limited close substitutes I No/limited entry by non-members I [Think about long-run]
2. Low expectation of severe legal punishment
3. Low organizational cost
6 / 15
] ↳wcstsnfssau
- in
Low Organizational Cost
I Factors contributing to low organizational cost:
1. Few firms involved
2. Highly concentrated market
3. (Nearly) homogenous products
4. Trade association
7 / 15
Antitrust Laws (U.S.)
I Sherman Act (1890): price fixing is “per-se” illegal I “Per-se”: just doing it is illegal I “Rule of reason”: must prove harm to consumers/competition
I Clayton Act (1914): strengthened the Sherman Act I Deals with other practices, such as exclusive dealing.
I Federal Trade Commission Act (1914): bans “unfair competition” I Established the FTC I Deals with false advertising
I Robinson-Patman Act (1936): bans wholesale price discrimination I Established due to the rise of chain stores I Follows “rule of reason” I Used to be “per-se” until 1990s
8 / 15
Detecting Cheating
I Four factors help detect cheating
1. Few firms I Easier to observe rivals’ behavior
2. Low independent price fluctuations I Are price changes due to demand/MC changes or due to
“cheating”? I Harder to answer if demand/MC vary wildly over time
3. Observable prices I Ex: database (govt. or private)
4. Firms sell to same types of consumers I Same vertical structure
9 / 15
f transparency )
Low Incentive to Cheat
I Some factors contributing to low incentive to cheat
1. Rapidly increasing MC I Expensive to expand output
2. Many consumers, small purchases I Idea: little incentive to cut prices to get loyalty
3. Single sales agent I Has little incentive to favor one firm
4. Few firms I Smaller gains from deviation I Think about Bertrand Grim-Trigger from last time with n firms
. . .
10 / 15
Finn with is west price
gets whole NLT - - -
Ties split equally
Unirrtrigyer :
• everyone plays pm until
deviation
. play I - shoe rush I e
-
. no )
forever after dev .
Rican : In
-
- sooo
A- Nash =p
tsfcowde :
get F- Mnt forever
NPV 's "
= bond t 8 bond t 82 - n
= Sq ( Its t 82 t . . . )
NN""=s-' Ifdeviatt
!
get T -
- sooo today
To - - o forever after
NPV d "
= Sooo t S - O t 82 . Ot rn
NPVdN=soo#
Louw de if
NPV " "
d Npv der
Sone . Ig I 8000
800ns E Sooo ( r - 8)
In I I - 8
s± s e Er
n
as n grows , rewire Cos I
Preventing Cheating
I There are a few ways to prevent cheating
1. Trigger strategy in a repeated game
I Ex: last time
2. “Most-favored nation” clause and a repeated game
I Offer rebates to past consumers if you lower prices in the future
3. Price-matching guarantee
I Offer to match rival’s price I Can sustain collusion without a repeated game I [Work through this]
11 / 15
if ririrely !
Start with P , =p z = pm- → what it firm 2 undercuts
?
i ) No grantee- : firm 23 Price h a Little ,
qty basicanydovbles
ii ) withgvarwtee !
Firm 2 's Price I a Little
- . . firm I matches
→ they soil split wet
→ cover a than
when
crowding ⇒ No incentive
to lower
price !
Non-Price Collusion
I Auctions I “Bid rigging” is basically just price collusion I But can also collude on entry I Prof. John Asker, UCLA (2010): stamp auctions
I Product Line Decisions I Prof. Chris Sullivan, UW-Madison (2017): super-premium ice
cream
I “When the smooth get chunky, the chunky get smooth.” - Ben Cohen
I Not clear if this harms consumers
12 / 15
Detecting Collusion
I How can economists detect collusion? I Use econometrics!
I Not always going to have a quote like Ben Cohen’s
I Let’s focus on price collusion in homogenous products
I Consider the following pricing equation:
p � ✓ p
|"| = MC
13 / 15
PII - Fei ) all
Detecting Collusion
p � ✓ p
|"| = MC
I ✓ is the conduct parameter I ✓ = 0: Bertrand or perfect competition I ✓ = 1: Monopoly I ✓ = 1
n : Cournot competition (identical firms)
I Idea: two-step procedure 1. Observe price/quantity in multiple periods (or markets)
I Note: need instrumental variable for price
2. Estimate demand using observations 3. Estimate ✓ and MC simultaneously using results from steps 1
and 2
14 / 15
PLI - Fa ) are
Detecting Collusion
I In the real world, we have heterogenous firms, product differentiation, etc.
I But the technique is essentially the same. I Ex: Miller and Weinberg (2017) I This is how they showed the Miller/Coors merger increased
collusion
I Interpreting generic values of ✓ is difficult (if not impossible) I Generally, ✓ closer to 1 interpreted as “more collusive”
I Cutting-edge work is trying to provide alternatives to the conduct parameter approach
15 / 15