.
Exclusionary discount contracts under
asymmetric information
Enrique Ide (Stanford) & Juan-Pablo Montero (PUC-Chile)
CRESSE 2015, Rethymnon-Greece
The Problem
entrant
Incumbent dominant supplier
,
≪1
0
Retail buyer
…….
…….
final consumers
,
1
Settings where exclusive contracts foreclose efficient entry
1. "Rent Shifting" models: uncertainty about ’s cost
• Aghion & Bolton (1987), Choné & Linnemer (2015)
2. "Naked Exclusion" models: exploit some buyers due to scale economies
• Rasmussen-Ramseyer-Wiley (1991), Segal & Whinston (2000), Spector (2011)
3. "Downstream Competition" models: exploit final consumers
• Simpson & Wickelgren (2007), Asker & Bar-Isaac (2014)
Exclusives and asymmetric information: exclusion more likely
• Aghion & Bolton (1987):  signals his private information with higher
breaching penalties (Ziss 1996)
• Giardino-Karlinger (2015): a "strong"  can hide behind a "weak"  in
a pooling equilibrium
• Calzolari & Denicolo (2015): exclusives are used to better screen buyers
• Johnson (2012), others?
Discount (e.g. rebate) contracts
• What is a rebate contract? a contract ( ) where
—  is the list price  pays if she buys from both  and 
—  is the percent-off-list-price in all units if  buys only from 
• can this contract be anticompetitive, yet profitable?
−

    − 

• Rebates:  rewards  for the exclusivity (ex-post)
• Exclusives:  compensates  for the exclusivity (ex-ante), which is
enforced with penalties for breach
Rebates in antitrust cases
• EU Commission v. British Airways (2003)
• EU Commission v. Michelin II (2003)
• AMD v. Intel (2005)
• Allied Orthopedic v. Tyco (2010)
• ZF Meritor v. Eaton (2012)
Rebate contracts 6= Exclusive contracts
• Simpson & Wickelgren (2007): Exclusive contracts cannot exclude if
penalties for breach are limited to expected damages, e.g., ( −  )
• Rebates don’t face such constraint⇒larger exclusionary potential
• However, rebates suffer from an "easy terminability" problem (Ide-MonteroFigueroa 2015):
— the exclusivity must be implemented ex-post with too large rewards
— rewards that cannot be recouped with high inframarginal prices ( ≤
)
• IMF (2015): a rebate contract ( ) can never anticompetitive in any
of the three settings above....unless...
Motivation for this paper
• unless the contract is completed with up-front payments from  to :
( ) −→ (  )
• But we don’t see upfront payments () documented in any of the antitrust cases above (we do see some from  to )
• Any reasons why we don’t see them?
— financial constraints (as in Ordover & Shaffer 2013)
— lack of first-mover advantage (as in Calzolari & Denicolo 2015)
— contracts ( ) are enough to foreclose inefficient entry, i.e.,  +
  
This paper: asymmetric information
• the use of upfront payments (’s) is also problematic if there is asymmetric information
• when  knows more than  about final demand (either high or low)
— both  and  like lower ’s
— contract to low-type is less exclusionary (lower  and higher effective
price)
— and not exclusionary at all ( = 0) if asymmetric info is high enough
— rebates (   = 0) are still offered to prevent inefficient entry
• similarly, when  knows more about final demand he signals a highdemand with less exclusionary contracts
Rest of the presentation
• focus on a rent-shifting (A&B) model: one retail monopolist ()
• at date 1,  makes a take-or-leave-it contract offer to 
— at this time  is unknown to both  and :  ∼ (·) over [0 ]
• at date 2, and having observed a contract,  makes a take-or-leave-it
offer to  for the contestable fraction  of the demand
— if ’s offer is accepted,  enters by paying a fixed cost  → 0
— ( and  don’t renegotiate their contract)
• at date 3, if  and  fail to sign a contract at date 1,  and  compete
in (non-linear) prices in the spot market ( is known at this stage)
Demand
•  is better informed of whether demand is likely to be high or low
(
1 +  with probability 
1
with probability 1 − 
where  ∈ {  }, with    = 0
Demand =
• ’s prior:  = Pr( =  ) and 1 −  = Pr( =  = 0).
• ( learns about the true demand before  makes his offer)
•  can sell up to (1 + ) units when demand is 1 +  and up to  units
when demand is 1
•  will consider menus (     ) and (  ), where  is the
up-front payment
Payoffs in the absence of a rebate contract (q=1)
• these are  and ’s outside options
• If  ≤  , efficient spot competition:
—   = (1 − )( −  )  = ( −  )
—   = ( −  )
• If    ,  enters with a price offer to  of  − 
—   = (1 − )( −  )  = ( −  )
—  = 0
Outside options and inefficient entry
• since ( ) is the probability that    , outside options are equal
to
̄  = (1 − )( −  )
̄  = ( −  )( )  ( −  )
• which add to
̄  + ̄ =  −  − [1 − ( )]( −  )   − 
• coalition of  and  could always sign something to secure  − 
Full-information problem
• let  =  −  denote the effective price  needs to charge to enter
• the rebate offer to buyer  =   solves
max E = (1+)[(1−)( − )+( − ) {1 − ()}]+
  



• subject to
 ≤ 
(1 + )( −  + ) −  ≥ ̄ = (1 + )̄
Full-information exclusionary solution
• the A&B solution
(̃ )
∗
 = ̃ ≡  −
 
(̃ )
• and
∗ ∈ [∗  ]
∗ = (1 + )[ − ∗ + ∗ − ̄ ]  0
• note the flexibility all the way to a two-part tariff (∗ = ∗ and ∗ = 0)
Second-best problem
• find the menu {  } and {     } that maximizes
E = [(1 +  )[(1 − )( −  ) + ( −  ) {1 − ( )}] +  ]
+ (1 − )[(1 − )( −  ) + ( −  ) {1 − ()} + ]
• subject to   ≤  and
[ ]
[]
[ ]
[]
(1 +  ) −  ≥ (1 +  )̄ 
 −  ≥ ̄ 
(1 +  ) −  ≥ (1 +  ) − 
 −  ≥  − 
where  =  −  −  is ’s ex-post profit per-unit of demand under
contract 
Second-best solution: always serving the low type
∗
• there is no distortion at the top: ∗∗
 =  = ̃  
• information rents of high-type not always increasing in  (Jullien 2000)
∗∗ − ∗∗ − ̄ ] ≥ 0
I() =  [ − (1 − )


=  [ − ∗∗
 − ̄  ] ≥ 0
• and then solve for ∗∗
 according to the FOC
 + (1 − )[()( − ) − ()] = 0
that yields ∗∗
  ̃ (which is increasing in ).
• finally, the up-front payment to the low-type is obtained directly by
setting the  condition to equality
∗∗ = ( − ∗∗) − ̄ ≥ 0



Concluding remarks
• Rebate contracts not equipped with up-front payments (like in EU Commission v. Michelin II ) are not anticompetitive in any of the settings
where exclusives are (IMF 2015)
• up-front payments restore the exclusionary potential of rebates, but
costly to use when the incumbent must screen privately informed buyers
(same for signalling)
• even for large information asymmetries the incumbent finds it optimal to
offer menus with non-exclusionary rebates (i.e., rebates without up-front
payments)
• ....as they help foreclose inefficient entry