Endogenous Coalition of Intellectual Properties: A Three

Transcription

Endogenous Coalition of Intellectual Properties: A Three
Endogenous Coalition of Intellectual Properties:
A Three-Patent Story
Chen Qu
Department of Economics, BI Norwegian Business School, Norway
24 April, 2015
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
1 / 19
Background
Cooperation among intellectual property owners (e.g., patent pools)
is often observed in a variety of industries.
Patent pool: an agreement by multiple patentholders to license a
portfolio of patents as a package to outsiders (or to share intellectual
property among themselves) (New Palgrave)
In 2001, sales of devices wholly or partly based on pooled patents
exceeded $100 billion.
Third Generation Patent Platform Partnership (3G3P): 5 independent
PlatformCos (platform companies), each consisting of patents
essential to one 3G radio interface technology.
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
2 / 19
Research questions
The purpose of this paper is to investigate the endogenous coalition
formation among intellectual property owners in a 3-patent setting:
fragmented pool structure
incomplete pool structure
R
R
complete pool R
R
R
R
R
R
R
4 questions:
(Q1) What are the pro…ts of patent pools in equilibrium under di¤erent
pool structures?
(Q2) Under what circumstances is the (in)complete pool the stable
pool structure?
(Q3) Is a market structure of fragmented patents a possible outcome?
(Q4) What is the welfare e¤ect of a stable pool structure?
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
3 / 19
Theoretical resources and contributions
Lerner & Tirole (2004, AER): a tractable model of patent pools.
p
Non-essential cumulative patents ( )
Stand-alone patents vs. one complete pool ( )
Ray & Vohra (1997, JET): equilibrium binding agreements (EBA).
The protocol of endogenous pool formation.
Incomplete pool may form.
Contributions:
1
2
A full picture of endogenous coalitional behaviors of IP owners;
particularly, the relationship between pool structure outcome and value
accumulation from increasing patents.
An application of theory of coalition formation (in a symmetric and
asymmetric case).
Other related literature: Quint (2014), Aoki & Nagaoka (2006),
Brenner (2009)
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
4 / 19
Plan of the paper
The model and preliminaries
Building block: equilibrium pro…ts under di¤erent pool structures
(Q1)
Stable pool structure: symmetric pro…ts (Q2, Q4)
Stable pool structure: asymmetric pro…ts (Q2, Q3, Q4)
Discussions
1
2
Qu (BI)
Alternative protocol: sequential bargaining
n-patent case (marginal contribution, homogeneous licensees)
Endogenous Coalition of IPs
()
24 April, 2015
5 / 19
Timeline of game
Stage 1. n owners form a pool structure C (a partition of n patents).
Stage 2. Prices are set by simultaneous Nash-like play by the pools.
The pro…t is divided equally within a pool. Asymmetric equilibria are
allowed.
Licensees distributed over [θ ∆, θ ]. Licensee θ’s valuation: θ + V (k ), k:
# of patents θ access, V (k ) % in k.
Stage 3. Each licensee selects the basket B s.t.
maxB C fV (]B ) PB g . (not user-speci…c)
Stage 4. Licensee θ adopts the technology i¤ θ + V (]B )
(user-speci…c)
Qu (BI)
Endogenous Coalition of IPs
()
PB .
24 April, 2015
6 / 19
Equilibrium of subgame: stages 2-4
Lerner and Tirole (2004):
All pools are in the equilibrium basket.
In equilibrium each pool charges
min fcompetition margin, demand marging.
Competition margin: the highest price a pool can charge without being
excluded from the basket.
Demand margin: the optimal price in the absence of co. margin.
9 some pool, s.t. all bigger pools charge same de. margin, and the
rest charge co. margin.
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
7 / 19
3-patent case: pro…ts
3 feasible pool structures:
f1, 1, 1g (fragmented); f1, 2g (incomplete); f3g (complete).
Notations: v1 V (1) + θ, v2 V (2) + θ and v3 V (3) + θ.
Assume licensees uniformly distributed.
Per-owner pro…t π (f3g) =
1 2
12 v3 .
Prop 1.
(a)
(b)
(c)
(a) v3
Qu (BI)
(v1 + v2
3
2 v2
π (f1, 2g)
1 2 1 2
9 v3 , 18 v3
1
v2 ) , 18 v22
2 v2 (v3
1
v3 ) (v3 v2 ) , 2 (v1 + v2
(b) v1 + 12 v2
v3 < 32 v2
Endogenous Coalition of IPs
v3 ) (v3
v1 )
(c) v3 < v1 + 12 v2
()
24 April, 2015
8 / 19
3-patent case: pro…ts cont.
Prop 2.
(d+f)
(concavity+g)
(e+convexity)
(v1
(d) v3 > 2v1
(e) v3 2v1
π (f1, 1, 1g)
1 2
16 v3 3
((3v2 2v3 ) (v3 v2 ))3
z ) (z, z, v3 v1 z ) with z 2 v2
4
(f) v3
3 v2
(g) v3 < 43 v2
(concavity) v3 < 2v2
(convexity) v3 2v2
v1 , v3 2 v1
v1
v1
In (e+convexity),
z: degree of symmetry. When z = v3 2 v1 , symmetric.
In…nite number of asymmetric equilibria: owner 3 earns high pro…t
and the other two the same low pro…t.
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
9 / 19
Equilibrium binding agreements (symmetric pro…ts)
A pool structure is called an equilibrium pool structure (EPS) if,
under this structure, no owners, individually or as a group, have
incentive to break away from the current pool by EBA:
1
2
Only internal deviations of a subset of an existing pool are allowed:
f3g ! f1, 2g ! f1, 1, 1g;
?
Owners are farsighted: f3g ! f1, 2g 99K f1, 1, 1g;
The coarsest EPS is the stable pool structure.
π (f3g) π (f1, 2g) π (f1, 1, 1g)
1 2
1 2 1 2
1 2
12 v3
9 v3 , 18 v3
16 v3 3
f1, 1, 1g is EPS; f1, 2g is not EPS (f1, 1, 1g blocks f1, 2g);
f3g, comparing π (f3g) with π (f1, 1, 1g), is (coarsest, stable) EPS.
A simple algorithm:
Step I : Is f1, 2g EPS?
If NO, ,!Step II : f3g is stable PS. Done.
If YES, ,!Step III : Is f3g EPS? (Compare π (f3g) with π (f1, 2g))
Example: (a+d+f)
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
10 / 19
Stable pool structure (symmetric pro…ts)
Props 3-5.
stable PS:
a+d+f
b+d+f
b+concavity+g
c+concavity+g
a/c+e+convexity
b+e+convexity
f3g?
Always
f1, 2g?
v3 > x 1
v3 2 [x2 , x3 ]
v3 2
/ (x4 , x5 )
v3 x 1
v3 2
/ [x2 , x3 ]
v3 2 (x4 , x5 )
Always
1
7
6v1 + 3v2
x6
Qu (BI)
Who defects?
Never
v3
x1
Always
Always
Never
f1g
f2g [ / ] f1g
/ (f2g) /
v3 2 (x3 , x6 ]
v3 x 6
/ (f1g]
p
2v2 .
q x1
p
3
3
3 v2 .
2 v2 ; x3
p
(+) 3δ , δ 3v22 2v1 v2 2v12 if δ 0.
p q
2
2v12 v22 , if 2v12 v22 > 0.
2
v3 2
/ (x3 , x6 ]
x2
x4 (x5 )
f1, 2g EPS?
2v1
Endogenous Coalition of IPs
()
24 April, 2015
11 / 19
Stable pool structure (symmetric pro…ts) cont.
stable PS:
a+d+f
b+d+f
b+concavity+g
c+concavity+g
a/c+e+convexity
b+e+convexity
Qu (BI)
f3g?
Always
f1, 2g?
v3 > x 1
v3 2 [x2 , x3 ]
v3 2
/ (x4 , x5 )
v3
Always
v3 x 1
v3 2
/ [x2 , x3 ]
v3 2 (x4 , x5 )
v3 2
/ (x3 , x6 ]
v3 2 (x3 , x6 ]
v3
f1, 2g EPS?
Who defects?
Never
x1
Always
Always
Never
x6
f1g
f2g [ / ] f1g
/ (f2g) /
/ (f1g]
Eg. (a+d+f) V (1) V (2) 0, V (3)
0 )stable f3g.
Eg. (b+d+f) (linear) V (k ) = Ak, A constant, k = 1, 2, 3.
v3 > x1 )stable f3g.
Eg. (b+concavity+g) V (1) 0, V (2) V (3)
0. v3 < x2 )
stable f1, 2g.
Eg. (c+e+convexity) V (1) V (2) V (3) p 0 )stable f3g.
7 +1
3
Prop 6. Stable PS is f3g if v3
2 v2 or v2 <
3 v1 . (The latter is
irrespective of v3 .)
Endogenous Coalition of IPs
()
24 April, 2015
12 / 19
Welfare analysis (symmetric pro…ts)
We say the stable PS increases welfare if the total price under it <
that under f1, 1, 1g.
Prop 7.
a/b+d+f
b+concavity+g
c+concavity+g
a/b+e+convexity
c+e+convexity
When the coarsest EPS % welfare?
always
p
v3
3/2v2
always & welfare (except if v2 > 54 v1 and v3
always
v3 > 32 v1
x 5)
Except (c+concavity+g) with restrictive v3 , the stable PS (almost) always
increases welfare.
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
13 / 19
Stable pool structure (asymmetric pro…ts)
fa, aAg
! fa, a, Ag.
fA, aag
Subtleties arise of the algorithm of …nding the stable PS.
Prop 8. (Polar asym.) When z = v2 v1 , fA, aag (always EPS) can
be stable in all the cases with (e+convexity).
) fa, a, Ag is never stable.
(In (c+e+convexity), fa, a, Ag is never stable for any z.)
4 feasible PSs: faaAg !
Can fa, a, Ag be stable? YES!
Prop 9. In (a+e+convexity), 9 (z, z )
z
stable PS
[v2 v1 , z ]
fA, aag
v2
(z, z )
fa, a, Ag
v1 , v3 2 v1 s.t.
z, v3 2 v1
faaAg
Modest asymmetry ) …nest market structure. Why?
High asymmetry ; π (ajA, aa) < π (aja, a, A)
High symmetry ; π (AjaaA) < π (Aja, a, A)
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
14 / 19
Welfare implication (asymmetric pro…ts)
By Prop 9, in (a+e+convexity),
z
stable PS
total price
[v2 v1 , z ]
fA, aag
2
3 v3
(z, z )
fa, a, Ag
v3 v1 + z
z, v3 2 v1
faaAg
1
2 v3
Prop 10. In (a+e+convexity), when v3 < 18
11 v1 , 9z 2 (z, z ) s.t.
z 2 (z, z ) leads to a lower total price than the one charged by
18
faaAg; when v3
11 v1 , any z 2 (z, z ) leads to a higher total price.
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
15 / 19
Discussions
Alternative protocol: in…nite-horizon unanimity bargaining
(, sequential game of choosing pool size)
3 symmetric patents: stable PS = SPE of game above.
Marginal contribution of ∆k patents to a size-k pool:
w (k, ∆k ) V (k ) V (k ∆k ).
Prop A2. Consider equilibrium p. Then co. margini = w (n, ni ) i¤
C nni 2 arg max fV (]J )
J C nn i
PJ g .
This holds if V ( ) satis…es w (n, ∆k ) w (k, ∆k ) for any k
any ∆k < k. (weaker than concavity of V ( )).
Eg. V (1) = 1, V (2) = 2, V (3) = 5, V (4) = 6.
Qu (BI)
Endogenous Coalition of IPs
()
n and
24 April, 2015
16 / 19
Discussions cont.
Homogeneous licensees (same θ), no externalities (using Prop.12)
)
) Per-owner pro…t of a size-t pool is w (n,t
.
t
t
1
2
3
4
5
6
V (t )
11.5 14 36 46 53 54
Eg.
w (n, t )/t
1
4
6 10 8.5 9
Multiple stable PS with no stand-alone patents: f2, 4g, f3, 3g.
(EPS f1, 1, 4g blocks f1, 5g. f2, 4g blocks f6g.)
Possible re…nement ) f4, 2g the only outcome:
The coarsest EPS should block some coarser structure;
Alternative protocol of sequential bargaining.
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
17 / 19
Conclusions
Varieties of equilibrium pro…ts under di¤erent structures.
NO straightforward prediction on stable PS!
Symmetric pro…ts: either complete pool or incomplete PS is stable.
Asymmetric pro…ts: fragmented PS can be stable.
Stable PS tends to increase welfare with large V (3); fragmented PS
may increase welfare.
Future research:
1
2
3
Qu (BI)
Up-front fees, per-unit royalties, and combinations of the two.
“Weak” patent: patent litigation, spillovers.
Full characterization of n-patent case.
Endogenous Coalition of IPs
()
24 April, 2015
18 / 19
^O^
Thank you ^O^
Qu (BI)
Endogenous Coalition of IPs
()
24 April, 2015
19 / 19