value capture in hierarchically organized industries: the role of open source inputs
DESCRIPTION
Value capture in hierarchically organized industries: The role of open source inputs. Joachim Henkel, Alexander Hoffmann. Open and User Innovation Workshop, Harvard Business School July 30, 2014. Nokia 770: Open source inputs. *Source: A. Jaaksi , presentation to LinuxWorld 2006. - PowerPoint PPT PresentationTRANSCRIPT
Technische Universität München
Open and User Innovation Workshop, Harvard Business School
July 30, 2014
Joachim Henkel, Alexander Hoffmann
Value capture
in hierarchically organized industries:
The role of open source inputs
Joachim Henkel, Alexander Hoffmann |
Nokia 770: Open source inputs
*Source: A. Jaaksi, presentation to LinuxWorld 2006 2
Joachim Henkel, Alexander Hoffmann | 3
Who appropriates the value that open source inputs contribute to complex, hierarchical systems?
Joachim Henkel, Alexander Hoffmann | 4
A hierarchically organized industry
OEM
2nd tier suppliers
1st tier suppliers
Joachim Henkel, Alexander Hoffmann | 5
Negotiations
The value split between industry participants is determined in negotiations.
Each 1st tier suppliernegotiates with its
2nd tier suppliers
OEM negotiates with1st tier suppliers
Joachim Henkel, Alexander Hoffmann | 6
Bargaining structure and value split
We show: Bargaining structure – who negotiates with whom – affects how the value is split.
Joachim Henkel, Alexander Hoffmann | 7
Model approach:
Hierarchical Shapley Value
Joachim Henkel, Alexander Hoffmann | 8
Hierarchical Shapley Value (1/3)
▪ Idea:
– Value split between level-1 (L1) modules according to Shapley Value
– Value split within a given L1 module acc. to (modified) SV, assuming all other L1 modules are complete and present (reflects limited information)
– Similar to Owen Value (1977), but different in important respect
L1 modules
L2 modules
Joachim Henkel, Alexander Hoffmann | 9
▪ Set of L1 modules: , where , disjoint, and
▪ Level-1 value distribution among L1 modules:
– Induced characteristic function represents value of (complete) L1 modules and their unifications:
– L1 Shapley value distribution:
Hierarchical Shapley Value (2/3)
�̂� :2𝔅→ℝ , �̂� ( 𝐽 ) :=𝑣 (¿𝑀 𝑖∈ 𝐽𝑀 𝑖 )
𝜙𝑀 𝑖
❑ (�̂� )≔ ∑𝑆⊆𝔅¿𝑀𝑖 }
|𝑆|! (𝑘−∨𝑆∨−1 )!𝑘! ( �̂� (𝑆∪𝑀 𝑖 )−�̂� (𝑆 ))
for
Joachim Henkel, Alexander Hoffmann | 10
▪ Level-2 value distribution among modules within
– L2 modules assume presence of all other L1 modules
– Induced characteristic function ():
– L2 Shapley Value distribution within an L1 module:
Hierarchical Shapley Value (3/3)
~𝑣𝑀 𝑖( 𝐽 ) :=
(𝑣 ( 𝐽∪ (𝑀 {𝑀 ¿¿ 𝑖))−𝑣 (𝑀 {𝑀¿¿ 𝑖 ) )𝑣 (𝑀 )−𝑣 (𝑀 {𝑀¿¿ 𝑖) 𝜙𝑀𝑖
❑ (�̂� )
𝜙𝑚 𝑗
𝐻𝑆𝑉 (𝑣 )≔𝜙𝑚 𝑗
❑ (~𝑣𝑀𝑖 ):= ∑𝑆⊆𝑀 𝑖 {{𝑚 𝑗
}¿|𝑆|! (¿ 𝑀𝑖∨−|𝑆|−1 ) !
¿𝑀 𝑖∨!(~𝑣𝑀𝑖 (𝑆∪{𝑚 𝑗 })−~𝑣𝑀𝑖
(𝑆 ))
Joachim Henkel, Alexander Hoffmann | 11
Open Source Inputs
Joachim Henkel, Alexander Hoffmann | 12
OSS modules and value splitWhat if one of the “participants” is an open source module?
Contributes to value creation But: does not claim any value Who appropriates the value contributed by the OSS module?
Joachim Henkel, Alexander Hoffmann | 13
Open source (OSS) modules
Assumptions:
▪ Contributors to the same subsystem as the OSS module are aware of it and use it in their negotiations
▪ Contributors to other subsystems as the OSS module, or higher levels, are not aware of it and do not use it in their negotiation
– “Information hiding” function of modularity (Baldwin and Clark, 2000)
Joachim Henkel, Alexander Hoffmann | 14
Results (1/2)
System
A B&OSS
B OSS
System
A B OSS
Value function , , , ,
, , ,
Since OSS is freely available, we assume that both A and B us it in their value proposition.
B’s value capture acc. to Shapley:
Hierarchical Shapley value:
Split on 2nd level, within “B&OSS”: trivial, all value to B
Split on 1st level: B captures…
Joachim Henkel, Alexander Hoffmann | 15
Results (2/2)
System
A B&OSS
B OSS
System
A B OSS
Through clustering with OSS module, B gains:
,
where denotes the “complementarity gains” of putting A and the OSS module together
Joachim Henkel, Alexander Hoffmann | 16
Conclusions
▪ It is beneficial for firm B to be clustered with the OSS module (compared to a situation where A, B, and the OSS module are all on the top level)
▪ How much B benefits from clustering, depends…– on the complementarity between A and the OSS module
– but not on the complementarity between B and the OSS module
▪ Logic: – Complementarity between B and OSS module benefits B in any case
– The stronger the complementarity between A and OSS module, the more of the value added by OSS (incl. the complementarity gain) is claimed by A in case of no hierarchy, making hierarchy more attractive for B
▪ Generalization of results to larger systems possible
▪ Results extend literature (e.g., Lerner & Tirole 2002, West & Gallagher 2006) showing that complementary products are a way to benefit from OSS
Thanks