Pricing and commitment by two-sided platforms.

I study pricing and commitment by platforms in two-sided markets with the following characteristics: (i) platforms are essential bottleneck inputs for buyers and sellers transacting with each other; (ii) sellers arrive before buyers; and (iii) platforms can charge both fixed fees and variable fees (royalties). I show that a monopoly platform may prefer not to commit to the price it will charge buyers at the same time it announces its seller price if it faces unfavorable seller expectations. With competing platforms, commitment makes the existence of an exclusive equilibrium (in which sellers register with only one platform) less likely, but it has no impact on multi-homing equilibria (in which sellers support both platforms) whenever these exist.

Olaf [Olafsson] was about two-thirds of the way through his speech when he said, "I would like to call up Steve Race to tell you a little bit more about the Sony Playstation." So I walked up. I had a whole bunch of sheets of paper in my hands, and I walked up, put them down on the podium, and I just said, "$299," and walked off stage to this thnderous applause.

Steve Race, President of Sony Computer Entertainment of America, at the first E3 (Electronic Entertainment Expo), May 11-13, 1995, Los Angeles, more than 6 months before the actual release of the Playstation.

1. Introduction

* Most of the recent literature on two-sided markets has modelled the two "sides" or categories of agents as arriving at the same time and therefore playing a simultaneous coordination game. While no one would argue that all agents in such markets arrive at the same time in a literal sense, the idea is that the equilibria of the simultaneous-move game represent the equilibria that would arise in a more realistic, sequential-move game. This is true as long as agents of the two sides arrive in a "sufficiently" alternated fashion. Some two-sided markets fit this description: credit cards (merchants and consumers); yellow pages directories, TV, newspapers (advertisers and viewers/readers); real-estate agents and other intermediaries (buyers and sellers), etc.

However, there are several prominent categories of two-sided markets for which this stylized representation does not seem particularly well suited, owing to a natural and well-defined order of arrival of the two sides, in the sense that most members of one side of the market arrive before most members of the other side. For example, in the software and video game markets, most application and game sellers join platforms (operating systems and game consoles) before most buyers do. This is for technological reasons: application and game development are long and costly processes, so platform vendors in these markets have to start courting sellers more than a year before the platform is ready to go on the market, in order to ensure that enough application support will be available for it at launch. An operating system or game console cannot be launched simultaneously to both buyers and sellers, because no buyers would purchase it without enough applications or games, and by the time any became available, buyers would be gone to another platform.

In this article I depart from previous literature by focusing on the strategic issues arising in two-sided markets in which the two sides arrive in a clear and well-defined order. The key issues that arise are whether platforms find it profitable to credibly commit to the price they will charge the side arriving later when trying to attract the side arriving earlier and how multi-homing on one side of the market--i.e., the possibility that its members support more than one platform--affects the pricing and commitment decisions.

I propose a model for studying Bertrand competition among platforms in two-sided markets with the following characteristics: (i) platforms are essential bottleneck inputs for buyers (buyers) to access the products offered by sellers (sellers); (ii) sellers arrive before buyers; and (iii) platforms can charge fixed fees on both sides and variable fees (royalties) on each buyer-seller transaction.

Perhaps the main characteristic of two-sided markets is the presence of bilateral indirect network effects giving rise to a "chicken-and-egg" problem. In my case, however, since sellers arrive before buyers, this implies that indirect network effects are asymmetric: once sellers have decided which platform to support, the coordination problem on the buyer side vanishes. Buyers will simply adopt the platform offering the largest surplus, taking into account the price it charges and the number of supporting sellers. Thus, the only coordination game is played by sellers. It would then seem natural that platforms should concentrate all their efforts on attracting sellers, or, in other words, on capturing "chicken" But this is not necessarily true when credible commitment to buyer prices is feasible, since then platforms have the option to commit to attract a large share of buyers ex post by announcing a low buyer price ex ante, which enables them at the same time to charge higher seller prices. For example, as illustrated by the quote at the beginning of the article, it is common for video game console manufacturers to announce (attractive) price tags for their upcoming consoles well in advance of their actual release, in order to attract the support of independent game developers (and justify charging them around $8 royalties per game sold).

Commitment and the determination of variable fees (royalties) are interdependent in my model. Royalties link the pricing game for sellers and the subsequent pricing game for buyers. Specifically, positive royalties announced in the first stage act as negative "marginal costs" for platforms when they compete for buyers in the second stage. There are two conflicting effects of royalties: on the one hand, absent commitment to buyer prices, high royalties put platforms in a better position vis-a-vis buyers in the second stage by allowing them to price more aggressively, but on the other hand, from an ex ante (i.e., prior to stage one) perspective, a platform expecting to extract most of the surplus created by sellers' products has an incentive to set low royalties in order to maximize this surplus by curbing seller market power over buyers.

Although there is full and complete information in my model, I show that even a monopoly platform can sometimes find it optimal not to commit to buyer prices at the time it announces prices for sellers. This result is rather counterintuitive: one would be inclined to believe that commitment always does weakly better in the absence of uncertainty, since then one can always commit to the action one would end up choosing anyway. The reason why this is not necessarily true here is the two-sidedness of the pricing problem, i.e., the interdependencies between buyers' and sellers' platform adoption decisions. If, for instance, sellers always coordinate on not adopting the platform whenever this is an equilibrium given the prices announced, the platform finds it optimal not to commit. This credibly communicates to sellers that the platform will attract all buyers irrespective of seller support, since the platform has the flexibility to price low enough to buyers if seller support is not forthcoming. But then sellers will in fact sign up, so that the platform can subsequently extract all the buyer surplus created, whereas commitment would have forced it to set a much lower buyer price.

With competing platforms under symmetric seller expectations, when commitment is not feasible, a pricing equilibrium always exists. If sellers cannot multi-home (i.e., have to support only one platform) then the only possible equilibrium involves exclusivity with either platform. If multi-homing is possible, then a multi-homing (respectively, exclusive) equilibrium exists if and only if it is constrained socially optimal. Introducing the possibility for platforms to commit makes the existence of exclusive equilibria less likely but has no bite on the multi-homing equilibria whenever the latter exist.

The rest of the article is organized as follows. In Section 2, I set up the modelling framework, and Section 3 is devoted to the analysis of the monopoly platform case. In Section 4, I analyze platform competition in a duopoly with symmetric seller expectations. Section 5 concludes.