Value appropriation between the platform provider and app developers in mobile platform mediated networks

Abstract

The mobile ecosystem has recently experienced a transition in platform leadership from network operators to mobile operating system providers. In each system the platform provider exerts effort in order to attract other firms for generativity and profitability. In this paper, we identify and analyze the working mechanism of one business practice that significantly influences the ecosystem’s generativity and platform provider’s profitability via value appropriation. Revenue sharing has become a common practice in the mobile ecosystem following NTT DoCoMo’s radical revenue-sharing model contributing toward mobile service success in Japan. Studies further argue that offering a wide portfolio of services through an attractive or innovative revenue-sharing model is one of key success factors in the mobile ecosystem. However, app developers have continuously claimed that they do not receive their fair share and the press reports a substantial number of disputes concerning revenue sharing between the platform provider and app developers. We propose a new bargaining model, the modified apex game, that investigates how value is likely to be appropriated between the platform provider and app developers within a given mobile platform mediated network. We support our theoretical predictions using data collected from the early mobile ecosystem by a network operator as well as the iOS and Android mediated networks.

Appendices

Appendix A

Outcomes of traditional bargaining models for a 3-entity MAG

The core

From the definition of the core given in Eichberger (1993), the core satisfies:

(i) v({1})⩽π₁⩽v({1, 2, 3})−v({2, 3}),

(ii) v({2})⩽π₂⩽v({1, 2, 3})−v({1, 3}) and

(iii) v({3})⩽π₂⩽v({1, 2, 3})−v({1, 2}).

We accordingly have:

Nash bargaining solution

From Compte and Jehiel (2010):

• i) If v({1, 2}), v({1, 3})⩽ (2)/(3)v({1, 2, 3}), then π ₁=π ₂=π ₃=(1)/(3)v({1, 2, 3}).

• ii) If v({1, 2})>(2)/(3)v({1, 2, 3}) and 1−(1)/(2)v({1, 2})⩾v({1, 3}), then π ₁=π ₂=(1)/(2)v({1, 2}) and π ₃=v({1, 2, 3})−v({1, 2})

• iii) If v({1, 3})>(2)/(3) v({1, 2, 3}) and 1−(1)/(2)v({1, 3})⩾v({1,2}), then π ₁=π ₃=(1)/(2)v({1, 3}) and π ₂=v({1, 2, 3})−v({1, 3})

• iv) If v({1, 3})+(1)/(2)v({1, 2})>v({1, 2, 3}) or v({1, 2})+(1)/(2)v({1, 3})>v({1, 2, 3}), then π ₁=v({1, 2})+v({1, 3})−v({1, 2, 3}), π ₂=v({1, 2, 3})−v({1, 2}), and π ₃=v({1, 2, 3})−v({1, 3}).

We accordingly have:

Nucleolus

Solving (Lemaire, 1984):

yields δ=0. We accordingly have:

Shapley value

From (Shapley, 1953):

we have:

Appendix B

Table B1

Table B1 Summary statistics