Competing Platforms

• Author: Robin S. Lee
• Published date: 01 September 2014
• DOI: http://doi.org/10.1111/jems.12068
• Date: 01 September 2014

Competing Platforms

ROBIN S. LEE

Department of Economics

NYU Stern School of Business

New York,NY 10012

rslee@stern.nyu.edu

Why do competing platforms or networks exist? This paper focuses on instances where the value of

a platform depends on the adoption decisions of a small number of firms, and analyzes the strategic

competition among platforms to get this oligopolistic side on-board. I study a bilateral contracting

game among platforms and firms that allows for general externalities across both contracting

and noncontracting partners, and examine when a market will sustain a single or multiple

platforms. When firms can join only one platform, I provide conditions under which market-

tipping and/or market-splitting equilibria may exist. In particular, even without coordination

failure, congestion effects, or firm multihoming, multiple platforms can co-exist in equilibrium

despite being inefficient from the perspective of the contracting parties. Expanding the contracting

space to include contingent contracts may exacerbate this inefficiency.

1. Introduction

Many important economic markets conduct transactions between agents facilitated by

a “platform” or “network.” Examples are numerous, and include technology standards

and business exchanges which enable interactions between firms, as well as markets

which allow consumers to access the goods and services of other firms.1In certain cases,

market fragmentation across multiple platforms may not be surprising nor warrant

concern, particularly when platforms are sufficiently differentiated.2However, when

there are strong network effects and externalities that favor agglomeration on a single

platform (Katz and Shapiro, 1985), the co-existence of multiple platforms may result

in socially inefficient replication of costs and investment, delayed adoption of new

technology due to uncertainty, and welfare losses due to restricted choice stemming

from product incompatibility (cf. Farrell and Saloner, 1985).3

Though the reason for multiplicity likely varies from industry to industry, potential

explanations include coordination failure (cf. Farrell and Klemperer, 2007, for a survey);

congestion effects (Ellison and Fudenberg, 2003; Ellison et al., 2004); and the ability for

agents to join multiple platforms or “multihome.” To some extent, these analyses are

I wish to thank Attila Ambrus, Susan Athey, Drew Fudenberg, Andrei Hagiu, and Al Roth, and the editors

and referees for for helpful discussion and comments. This is a revised chapter of my 2008 HarvardUniversity

dissertation. All errors are my own.

1. For example, hardware-software markets (computer operating systems, videogame consoles), retail

marketplaces (shopping centers, auction sites, franchises), content delivery systems (television, online music),

and health care (managed care organizations).

2. Heterogeneous consumers may prefer different vertical (price, quality) characteristics or horizontal

(location) attributes. Also, platforms may choose to subsidize different sides of a multisided market as in

Ambrus and Argenziano (2009).

3. Empirical work documenting welfare losses due to incompatibility across platforms include Ohashi

(2003) on VCRs; Rysman (2004) on yellow pages; Ho (2006) on insurer-hospital networks; Ishii (2008) and

Knittel and Stango (2011) on ATM networks; and Lee (2013) in hardware–softwaremarkets.

C2014 Wiley Periodicals, Inc.

Journal of Economics & Management Strategy, Volume23, Number 3, Fall 2014, 507–526

508 Journal of Economics & Management Strategy

incomplete as they are frequently predicated on one or both of the following assump-

tions: (i) agents are atomistic price takers who cannot individually influence market

outcomes; and (ii) platforms are nonstrategic, acting merely as options for others to

choose among. Yet, in many prominent platform markets, neither of these assumptions

hold. First, some agents in the market are oligopolistic, strategic, and can individually

influence profits and shares of all others; second, because a platform’s success will rely

on its ability to initially attract these agents to join, they act strategically and actively by

offering contracts and other incentive schemes. In these settings, it is no longer clear that

the same forces previously identified in the literature lead to the existence of a single

platform (“market-tipping”) or multiple platforms (“market-splitting”).

In an attempt to shed light on this issue, this paper explicitly focuses on the

underlying contracting game between platforms and an oligopolistic set of agents, and

emphasizes how features of this particular form of competition alone can dictate whether

or not a market will tip or sustain multiple platforms. I focus on a setting where agents

are symmetric and can only join a single platform (which may be reasonable for markets

where the costs of multihoming are large, such as the adoption of a particular standard

or technology, or decision to franchise for a particular brand), and show that even

without coordination failure, congestion, or multihoming, outcomes that are worse for

the contracting parties (e.g., multiplicity instead of agglomeration, or vice versa) may

persist in equilibria simply as an outcome of platform competition and contracting with

externalities.4

This paper examines a two-stage bilateral contracting game: platform providers

first make offers to a set of oligopolistic agents—which we will refer to as “firms”—that

specify the payment made to (or demanded of) each firm conditional on that firm joining

the platform; firms then simultaneously choose which platform to affiliate with. I focus

on the strategic game played between only these two sets of players—the platforms

and firms—even if there may exist other sets of players in the market. For tractability, I

assume the eventual market structure and expected payoffs to all parties are determined

solely by the contracting decisions of the oligopolistic set of agents.5These “partition” or

“value” functions—payoffs to agents conditional on the realized network structure—are

allowed to be quite general, and will be assumed to be primitives of the analysis.

As payoffs to any platform or firm will depend on the actions of all agents, the

setting analyzed is a multilateral contracting environment with externalities; as such,

previous analyses of bilateral contracting games are not directly applicable.6

I focus on an environment with only two platforms and assume all firms are

symmetric and must choose a single platform to join; secondly, I impose the solution

concept of Coalition-Proof Nash Equilibrium (CPNE) introduced by Bernheim et al.

4. Although a complete analysis allowing for multihoming and asymmetric firms is beyond the scope of

our paper, insights fromthis paper still apply to these richer environments. For example, Lee and Fong (2013)

study a model of network formation and bargaining in more general networked industries where asymmetric

agents can have multiple contracting partners, and find inefficient outcomes can persist.

5. If there are only firms and platforms in the market, this assumption is not restrictive. In other settings

when there are other agents that act after platforms and firms have contracted, I assume there is a unique

subgame equilibrium following the contracting stage. I discuss this assumption further in the next section.

6. Many of these papers have analyzed a noncooperative setting where one set of players (principals)

typically make take-it-or-leave-it contract offers to the other set of players (agents) in order to induce them

to take certain actions. The seminal paper by Bernheim and Whinston (1986) analyzed common agency,with

many principals attempting to influence the action of a single agent via transfers; Segal (1999) and Segal and

Whinston (2003) studied the case with one principal and many agents; and Prat and Rustichini (2003) explored

the case with multiple principals and multiple agents, but did not allow for agents to exert externalities on

other agents.