Data sharing agreements in vertically differentiated two‐sided markets

• Author: Juan‐Manuel Sánchez‐Cartas,Gonzalo León
• DOI: http://doi.org/10.1111/ijet.12192
• Date: 01 September 2020
• Published date: 01 September 2020

Data sharing agreements in vertically differentiated

two-sided markets

Juan-Manuel S

anchez-Cartas



and Gonzalo Le

on



We study the impact of data sharing agreements between two platforms with vertically

differentiated consumers. We assume several utility functions to address the impact of those

agreements under different market configurations. The results show that the follower always

wants to share its network with the leader, while the leader has no incentive to share its network.

We also prove that vertical two-sided markets are quite prone to monopolies. Lastly, we show

that regimes in which the leader shares its network are worse than not sharing at all.

Key words two-sided market, data sharing, vertical differentiation, compatibility

JEL classification D43, K21 L11, L13, L15

Accepted 27 March 2018

1 Introduction

We study the effects of data sharing agreements and expectations in vertically differentiated

two-sided markets and their implications on price competition. Many real-life markets operate with

these features, indeed many digital markets follow this basic description. For example, apps such as

Google Fit and Garmin Connect can be described with these models. Those apps are platforms in

which users can follow their lifestyle, and developers can use users’ data to create new apps. So both

users and developers value the presence of the others on the platform. But other markets can also be

described, such as wearables and fitness trackers. In this case, some users may value the quality of the

device but may not care about the presence of advertisers/developers on the other side of the market.

However, other users such as runners and athletes may compete with each other, and in that sense

they value the presence of other athletes (direct network effect).

These examples share three features. First, all of them are prone to generating data sharing

agreements. For example, in the case of fitness platforms such as Endomondo and MyFitnessPal,

users can import/export their data from one platform to another but only use one of them.

1

In other

words, they singlehome, but they ‘‘are’’ on two platforms. From the developers’ point of view,

the Endomondo database consists not only of Endomondo users but also of some MyFitnessPal

users. The same is true in the fitness tracker market. For example, Fitbit and Withings have data



Centre for Technology Innovation (CAIT), Madrid Polytechnic University, Montegancedo Campus, Pozuelo de Alarc

on,

Spain. Email: juanmanuel.sanchez@upm.es

This work has been supported by Project H2020 FI-WARE, and especially by the Joint Research Unit between the

Universidad Polit

ecnica de Madrid (UPM) and Telef

onica R&D. We also thank the anonymous reviewer for valuable

comments.

1

https://goo.gl/V4K9S5

doi: 10.1111/ijet.12192

International Journal of Economic Theory xxx (2018) 1–22 ©IAET 1

International Journal of Economic Theory

International Journal of Economic Theory 16 (2020) 260–281 © IAET

260

sharing agreements that allow them to synchronize their devices with competitors’ platforms, which

means that they have larger databases to sell to advertisers or developers.

2

A second interesting fe ature is that not all users val ue the presence of other us ers (or the

quality of the product) in t he same way. At a given price, so me users put more value on the

presence of other users (o r the quality). Not all Ga rmin users value the pos sibility of getting in

touch with other users i n the same way, not all Goog le Fit users compete with other users i n

the same way, and not all Fitb it users value the techn ology of the device they are wearing in the

same way.

A third feature is that all of the aforementioned examples are digital markets.

We consider three different cases. The first is related to Gabszewicz and Wauthy (2004), which is

adopted as the framework for comparing. In the second we assume agents’ heterogeneity comes

from exogenous features (quality). In the third we assume heterogeneity comes from direct network

effects. In this last case we assume not only responsive expectations (or beliefs) but also passive

expectations.

We compare the three different scenarios with and without data sharing agreements. We show

that platforms’ profits are higher when the follower platform shares its data with the leader relative

to the case where there is no data sharing. In contrast, the case in which the leader shares its data

with the follower leads to a Bertrand equilibrium. As a matter of fact, from the leader’s point of view,

it is not profitable to share its data because it reduces the differentiation. We also find that multiple

equilibria are common to all frameworks. For instance, we find the coexistence of an asymmetric

interior equilibrium with the dominant firm equilibrium in which a company monopolizes the

market in a two-sided market with direct network effects, and in a market with exogenous quality

levels (when those quality levels are low and similar). Finally, this work provides an interesting

contrast with Hagiu and Halaburda’s (2014) results. In our work, platforms with more market power

prefer responsive users.

2 Literature review

Data sharing agreements are not a new concept.

3

However, the concept has new implications in the

digital economy. These new implications are mainly related to compatibility. As a matter of fact, data

sharing agreements can be considered as a kind of compatibility because they allow companies to

‘‘access’’ to the network of other companies.

Compatibility has a long tradition in the industrial organization literature, and it is a broader

concept than data sharing. In that sense, it is useful to analyze that literature to know how to address

the data sharing phenomenon.

We can identify two ways of addressing compatibility: one is the classical way based on one-sided

models, such as Katz and Shapiro (1985), Farrell and Saloner (1985, 1986), and Matutes and

Regibeau (1988), among others. These works have established the lines along which compatibility is

addressed.

However, as Wright (2004) points out, conventional knowledge from one-sided markets may

lead to mistakes in two-sided markets. In that sense, conclusions from the one-sided literature may

not be robust in two-sided markets. This suggests that policy-makers have to be careful not to base

2

https://goo.gl/mtRnZu or https://goo.gl/FMxKHT

3

We define a data sharing agreement as a contract by which platforms or companies allowother companies to access to

specific customer data for example, free access to the information about the lifestyle of users.

Data sharing agreements Juan-Manuel S

anchez-Cartas and Gonzalo Le

on

2International Journal of Economic Theory xxx (2018) 1–22 ©IAET

International Journal of Economic Theory 16 (2020) 260–281 © IAET 261