Information sharing with blockchain

Christian Riis Flor, Stefan Hirth, Evgeny Lyandres, Alexander Schandlbauer∗

September 2024

Abstract

We analyze benefits and limitations of information sharing among product market rivals via
a smart-contract-enabled blockchain. Firms face uncertain market demand but obtain infor-
mative signals that can be reported to the blockchain, which ensures immutable recording but
does not guarantee truthfulness. Truthful reporting is incentivized by requiring firms to post
collateral (“stake”), which is forfeited if the report is found to be biased ex-post. We derive the
necessary amount of collateral that ensures truthful reporting in equilibrium and examine its
comparative statics with respect to ex-ante demand uncertainty, signal precision, and compet-
itive interaction among firms. While the collateral is returned to firms that report truthfully,
financially constrained firms may struggle to afford the initial stake. Thus, they are potentially
unable to join the blockchain. We investigate conditions under which it is beneficial for firms to
partially subsidize their rivals’ collateral costs. In addition, we examine welfare implications of
information sharing via blockchain and regulators’ incentives for promoting blockchain adoption.

Keywords: blockchain, information sharing, collateral, financial constraints

JEL Classifications: G18, L10, L13, L15

∗
Christian Riis Flor: University of Southern Denmark, Department of Business and Management, Campusvej 55,
5230 Odense M, and Danish Finance Institute, [email protected]; Stefan Hirth: Aarhus University, Department of
Economics and Business Economics, Fuglesangs Allé 4, 8210 Aarhus, and Danish Finance Institute, [email protected].
dk; Evgeny Lyandres: Tel Aviv University, Coller School of Management, P.O. Box 39040, Tel Aviv 6997801, Israel,
[email protected]; Alexander Schandlbauer: University of Southern Denmark, Department of Business
and Management, Campusvej 55, 5230 Odense M, and Danish Finance Institute, [email protected]. Stefan Hirth
gratefully acknowledges funding from Independent Research Fund Denmark, Grant 0133-00087B.
1. Introduction

Firms’ usage of blockchain technology has increased significantly over the last years. In a re-

cent survey by EY in 2023, 46% of U.S. workers reported that their business has started or has

fully adopted blockchain technology in the past three years.1 One example is IBM’s “Food Trust”

blockchain, which allows participants to share records of food provenance and to track products

across the entire supply chain. Launched in 2018, it has attracted several major food industry

companies such as Walmart, Nestle, and Unilever. By encouraging suppliers to join, its aim is

to improve supply chain efficiency and transparency (e.g., Kshetri 2023). Another example is the

“TradeLens” platform, which was established in 2016 by Maersk, the world’s largest shipping com-

pany (Sarker et al. 2021). The main goal was to use blockchain to modernize and automate the

record-keeping and information exchange via paperless trade and thereby to decrease shipping costs.

By 2020, the TradeLens ecosystem included 175 organizations, 10 ocean carriers, and encompassed

data from 600 ports and terminals. However, in 2022, TradeLens abruptly announced discontinua-

tion of operations, which, as analysts argued, was mainly due to the failure to incentivize additional

significant players to join the network.2

While these two examples highlight the potential of blockchain to revolutionize the way infor-

mation is shared, by ensuring greater trust and accountability among participants, one can also see

that significant challenges remain. In this paper we aim to shed light on trade-offs that firms face

when deciding whether to set up and join a blockchain that would facilitate information sharing.

This research question is relevant both for firms along a supply chain, but also, and perhaps es-

pecially, for product market rivals. Moreover, regulators should be aware of those trade-offs when

contemplating whether to support information sharing with blockchain to possibly increase welfare.

At its core, blockchain is a decentralized digital ledger, which securely records transactions

across a distributed network. It allows for immutable and transparent recording of data. Smart

contracts, which are an extension of blockchain technology, are contracts in which the terms of an
1
Additionally, 44% of workers predicted that the technology will be widely used within the next three years. See
https://www.ey.com/, last accessed on August 30, 2024.
2
Moreover, it was argued that the governance system of TradeLens was not optimal: It was meant to be a for-profit
system, and the role of Maersk, as a founding member, was seen as critical. See e.g., www.ledgerinsights.com, last
accessed September 15, 2024.

1
agreement are written directly to a blockchain. These contracts are automatically executed when

predetermined conditions are met, which eliminates the need for intermediaries. Blockchain tech-

nology and smart contracts together offer a potential way to facilitate transactions and information

sharing.

However, the technology is yet to achieve widespread adoption across many industries, and

several obstacles remain (Lacity 2018, Dutta et al. 2020), as also demonstrated in the aforemen-

tioned TradeLens example. Given its potential to enhance transparency and information sharing,

the question thus arises of why blockchain is not used more often. In this paper we argue that one

answer lies in the barriers to adoption.

One challenge of smart-contract-enabled blockchain networks is that the means to achieve truth-

ful reporting of information relies on collateral, known as “staking,” which participants need to post

and which is returned if the report turns out (or is judged to be) truthful ex-post.3 “Skin in the

game” by all participants discourages dishonest and fraudulent behavior and ensures that the infor-

mation shared via the blockchain is truthful. Cryptocurrency payment processors and decentralized

marketplaces, such as BitPay and BitBay, have already implemented similar systems.

We develop a model in which two firms engage in competition in strategic substitutes or com-

plements, while focusing on a setting in which truthful information sharing among product market

competitors enhances their combined value. We begin by assuming that firms can commit to truth-

ful reporting of their demand signals and show that the expected gain from truthful information

sharing is increasing in ex-ante demand uncertainty, in signal precision, and in the extent of com-

petitive interaction between firms. However, assuming truthful reporting is insufficient, as firms

have incentives to deviate from the truthful information sharing equilibrium and misreport the

information that they submit to the blockchain. To ensure truth-telling, firms need to provide

collateral ex-ante, which is sufficiently high that its loss in the case of misreporting would outweigh

the gains from biased reporting. We derive the collateral required for truth-telling in equilibrium

and examine its comparative statics.

Even though the collateral is returned upon honest behavior, financially constrained firms may
3
E.g., Asgaonkar and Krishnamachari (2019) suggest to implement a so-called “double-deposit escrow” smart
contract to incentivize honest actors and truthful information sharing. Each party is required to deposit twice the
item price to the contract and if anything goes wrong, everyone loses the deposited collateral. If not, it is returned.

2
face difficulties in posting the collateral required for joining the blockchain. In such circumstances,

it can be beneficial for rival firms to subsidize the payment of the collateral. We show that such

subsidies are especially likely when firms’ demand signals are precise and when ex-ante demand

uncertainty is high.

Additionally, our model demonstrates that from a social planer’s point of view, blockchain-

based information sharing is welfare improving. Thus, governments may have incentives to adopt

regulations that promote information sharing via blockchain and possibly facilitate participation

by firms in such blockchains. Overall, this paper aims to improve our understanding about benefits

and limitations of information sharing between product market rivals via blockchain.

Our paper contributes to several strands of literature. The first is the literature examining

various benefits of blockchain technology. While initially developed to support cryptocurrencies

such as Bitcoin, applications of blockchain technology have expanded far beyond digital currencies

to include various business and industrial processes (e.g., Swan 2015, Goldstein et al. 2019, Chen

et al. 2019). Earlier papers have examined costs and benefits of decentralized blockchain-based

record keeping (Abadi and Brunnermeier 2019), the role of smart contracts (Cong and He 2019),

advantages of financing via blockchain (e.g., Catalini and Gans 2018, Gan et al. 2021, Chod and

Lyandres 2021, Gryglewicz et al. 2021, Lyandres et al. 2022, Davydiuk et al. 2023, Gan et al. 2023,

Malinova and Park 2023), and potential effects of blockchain on corporate governance (Yermack

2017). Subsequent studies have examined benefits of blockchain in various industries, starting with

supply chain management (e.g., Chod et al. 2020, Dong et al. 2023, Cui et al. 2023, Iyengar et al.

2023), settlement of asset trading (Chiu and Koeppl 2019), asset-backed securities (Chen et al.

2023), environmental monitoring (Cong et al. 2024), and accounting and financial reporting (e.g.,

Cao et al. 2023, Han et al. 2023). Chod and Lyandres (2023) and Shakhnov and Zaccaria (2023)

examine effects of blockchain on product pricing. See Biais et al. (2023) for a recent discussion

of advances in blockchain economics. We add to this literature by highlighting that blockchain

technology and smart contracts can act as a way to ensure truthful information sharing among

non-colluding product market rivals.

Another line of research that we contribute to is the economic literature on information sharing,

pioneered by early papers of Novshek and Sonnenschein (1982), Clarke (1983), Vives (1984), and

3
Kirby (1988). In subsequent work, Raith (1996) develops a general framework highlighting the

importance of both the type of information shared and of products, as well as the structure of

production costs for firms’ decisions to disclose information to their rivals. Malueg and Tsutsui

(1996) show that information sharing among oligopolists may sometimes be optimal. Importantly,

these models rely on the assumption that the information shared is indeed truthful, which we relax.

In our analysis, truthful information sharing can arise as an equilibrium outcome, facilitated by

blockchain technology and smart contracts, and is not an assumption.

Additionally, our paper contributes to the (strategic) management literature on “co-opetition,”

that describes the situation in which firms engage in simultaneous cooperation and competition with

each other (Brandenburger 1996, Lado et al. 1997, Gnyawali and Madhavan 2001). Co-opetition

can be helpful for firms in many dimensions, such as addressing technological challenges and in-

novation (Gnyawali and Park 2011), helping shape industry standards (Mathews 2002, Leiponen

2008), or increasing access to a wider range of resources (Brandenburger and Nalebuff 2021). For

a recent discussion about the interplay and the implications of competition and cooperation see

the discussion paper by Hoffmann et al. (2018). At the same time, however, several studies dis-

cuss challenges that arise when firms simultaneously engage in cooperation and competition (Lavie

2007, Gnyawali and Park 2009, Gnyawali and Ryan Charleton 2018). We contribute to this litera-

ture by highlighting that co-opetition can potentially be facilitated by means of blockchain-based

information reporting.

Our paper highlights a key managerial implication: blockchain can enhance trust and facilitate

collaboration between rival firms, as the decentralized governance structure of blockchain reduces

the need for centralized authorities in managing information sharing agreements. As a consequence,

the costs of sharing of information, such as verifying other firms’ actions and communication, may

be reduced with the help of automated smart contracts, and decision-making can be streamlined.

Hence, the use of blockchain can foster innovation by creating a more open environment for infor-

mation and knowledge exchange.

Beyond managerial considerations, our findings also carry important policy and welfare implica-

tions. Establishing a blockchain to facilitate information sharing between rival firms is not beneficial

from the representative consumer’s point of view. However, it is welfare improving for society, i.e.,

4
from a social planner’s perspective. Hence, a government may have incentives to promote policies

aimed at facilitating blockchain adoption by firms, which could enable truthful information sharing

among them.

The remainder of the paper is organized as follows. Section 2 describes the baseline model

with no information sharing. Section 3 constructs a baseline model in which truthful information

sharing is assumed. Section 4 analyzes how a blockchain can be helpful in establishing truth-

telling and examines conditions for truthful information sharing in equilibrium. Section 5 analyzes

whether it can be optimal for industry participants to subsidize their rivals in order to foster

their participation in the blockchain. Section 6 discusses welfare implications and central planner

considerations. Section 7 concludes. All proofs are found in the Appendix.

2. Baseline model with no information sharing

To set the stage for our analysis of information sharing via blockchain, we outline a baseline

(“textbook”, see, e.g., Vives 1999) duopoly model in which two firms, denoted by i and −i (for a

focal firm and its competitor, respectively) produce differentiated goods purchased by a represen-

tative consumer, whose utility function takes the following form:

1
β(qi2 + q−i
2


U (qi , q−i , pi , p−i ) = αi qi + α−i q−i − ) + 2γqi q−i − (qi pi + q−i p−i ), (1)
2

where qi and pi (q−i and p−i ) are the output quantity and price of firm i (−i); αi (α−i ) proxies for

product i’s (−i’s) desirability and, thus, indirectly, the size of firm i (−i); β ≥ 0 measures the direct

utility; −β ≤ γ ≤ β is the parameter describing substitutability/complementarity of the two firms’

products and thus measures the extent of competitive interaction between the two firms. For γ < 0,

the firms’ products are complements, whereas they are substitutes for γ > 0 (and if γ = β, they

are perfect substitutes). In what follows, we will refer to relative substitutability/complementarity

δ = γ/β and express γ as a fraction of β: γ = δβ, where δ ∈ [−1, 1].

The firms’ output decisions are made simultaneously and non-cooperatively. We assume that

αi are random variables, drawn from two uncorrelated two-point distributions, and can be equal

to ᾱi (1 ± ϵi ), where ᾱi ≥ 0 is the average demand intercept of firm i and ϵi ∈ [0, 1] measures the

5
demand uncertainty of firm i.

The inverse demand function, obtained by differentiating the representative consumer’s utility

in (1) with respect to qi and equating the resulting expression to zero, is given by

pi = αi − βqi − δβq−i , (2)

and firm i’s profit is:

πi = (αi − βqi − δβq−i )qi . (3)

Prior to making its output decision, each firm receives a non-verifiable signal about the real-

ization of its own demand, αi , but no signal about competitor’s demand, α−i . The signal for firm

i, Si , takes the form of Si = H (“high signal”) or Si = L (“low signal”). In the case of Si = H,

1
αi = ᾱi (1 + ϵi ) with probability ϕi ≥ 2 and αi = ᾱi (1 − ϵi ) with probability 1 − ϕi , where ϕi is firm

i’s signal precision. Similarly, in the case of Si = L, αi = ᾱi (1 + ϵi ) with probability (1 − ϕi ) and

αi = ᾱi (1 − ϵi ) with probability ϕi . The expected demand realization, given the signal value, is

then: Eαi (Si ) = ᾱi (1 + I(Si ) (ϵi (2ϕi − 1))) and I(Si ) = 1 if Si = H, whereas I(Si ) = −1 if Si = L.

Firm i’s expected profit, conditional on its signal, is thus:

 
1
1

Eπi (Si ) = Eαi (Si )−βqi (Si )−βδq−i (S−i = H) + Eαi (Si )−βqi (Si )−βδq−i (S−i = L) qi (Si ).
2 2
(4)

Four expected profit functions, two for each of the two firms that can receive two possible

signals, are maximised with respect to the firms’ respective production quantities. Equating the

corresponding partial derivatives to zero and solving the resulting system of four equations leads to

equilibrium output quantities of the two firms conditional on their signals and the firms’ resulting

expected profits ex-ante (i.e. prior to observing their respective signals):

4β 2 ᾱi − 2β 2 δ ᾱ−i + I(Si ) β 2 ᾱi ϵi (4 − δ 2 )(2ϕi − 1)

qino sharing (Si ) = , (5)
2β 3 (4 − δ 2 )
ᾱ2i (δ 2 − 4)2 ϵ2i (1 − 2ϕi )2 + 16 − 16ᾱi ᾱ−i δ + 4ᾱ2−i δ 2


Eπino sharing = . (6)
4β (4 − δ 2 )2

6
3. Truthful information sharing

In this section we examine the desirability of sharing information (demand signals) among

firms, while (temporarily) assuming that both firms report their signals truthfully and that rivals

correctly infer these signals from the submitted reports. Rivals’ reports are available to firms when

they make their output decisions. In the next section, we examine the implementation of truthful

information sharing via a smart-contract-enabled blockchain and analyze conditions under which

truthful reporting of signals on the blockchain is incentive-compatible.

Let us denote by Ri (Si ) the report of firm i given its signal Si , and J−i (Ri ) the inference by firm

−i of the signal received by firm i given the report Ri . Truthful reporting and correct inference

imply Si = Ri (Si ) = J−i (Ri ). Firm i’s expected profit after receiving signal Si and observing rival’s

report R−i is then given by:

Eπi (Si , R−i ) = Eαi (Si ) − βqi (Si , R−i ) − βδq−i (R−i , Ri ) qi (Si , R−i ). (7)

Maximizing each of the two equations (one for each firm) with respect to respective qi (Si , R−i ),

equating the resulting expressions to zero, and solving the resulting system of two equations leads

to equilibrium output quantities of the two firms conditional on their own signal and their rival’s

truthfully reported signal:

2ᾱi 1 + I(Si )(ϵi (2ϕi − 1)) − δ ᾱ−i 1 + I(S−i )(ϵ−i (2ϕ−i − 1))
qitruthf ul sharing (Si , R−i ) = . (8)
β(4 − δ 2 )

The resulting ex-ante expected profits are:

−4ᾱi ᾱ−i δ + 4ᾱ2i 1 + ϵ2i (1 − 2ϕi )2 + ᾱ2−i δ 2 1 + ϵ2−i (1 − 2ϕ−i )2

Eπitruthf ul sharing = . (9)
β(4 − δ 2 )2

To make the intuition more transparent, we will focus on the analysis of the symmetric case

hereafter, that is, we assume identical parameters for both firms: ᾱi = α¯−i ≡ ᾱ and ϵi = ϵ−i ≡ ϵ.4

The equilibrium expected profit of each of the two firms5 under truthful information sharing, (9),
4
The model is fully analytically tractable for the case of asymmetric firms as well.
5
From here on, we remove the subscript i when denoting the two firms’ equilibrium outcomes, as they are sym-

7
and the one without information sharing, (6), can then be simplified as:

(δ − 2)2 + ϵ2 (2ϕ − 1)2 (4 + δ 2 )

Eπ truthf ul sharing = ᾱ2 , (10)
β(4 − δ 2 )2
4 + ϵ2 (2ϕ − 1)2 (2 + δ)2
Eπ no sharing = ᾱ2 . (11)
4β(2 + δ)2

The expected benefit of truthful information sharing to each of the two firms is given by:

α2 δ 2 12 − δ 2 ϵ2 (1 − 2ϕ)2


truthf ul sharing no sharing
∆Eπ = Eπ − Eπ = . (12)
4β (4 − δ 2 )2

The properties of ∆Eπ are summarized in the following proposition.

Proposition 1 The expected benefit to each firm from truthful information sharing and correct

inference relative to no information sharing, ∆Eπ, is always non-negative, and it is strictly positive

unless one of the following occurs: ϕ = 1/2, ϵ = 0, δ = 0, or ᾱ = 0. Furthermore, if ∆Eπ > 0,

then:

∂∆Eπ
> 0, (13)
∂ϵ
∂∆Eπ
> 0, (14)
∂ϕ
∂∆Eπ
has the same sign as δ. (15)
∂δ

Proposition 1 highlights that truthful information sharing with correct inference is beneficial for

each of the two firms. Moreover, the expected gain from truthful information sharing is increasing

in both the ex-ante demand uncertainty, ϵ, and in the signal precision, ϕ. If the firms’ products

are strategic complements (δ < 0), the benefit of truthful information sharing increases in the

degree of product complementary (i.e., the advantage of information sharing is larger the lower

the δ). Figure 1 depicts illustrative numerical implementations of these comparative statics with

the parameter values specified in Table 1. Panels (a) and (b) demonstrate that ϕ and ϵ have

interchangeable effects. In the case of symmetric firms, this is evident from a casual observation

of ∆Eπ in (12). Panel (c) shows that the gain from truthful information sharing is increasing in
metric.

8
 Δπ Δπ
0.10 0.10

0.08 0.08

0.06 0.06

0.04 0.04

0.02 0.02

0.00 ϵ 0.00 ϕ
0.0 0.2 0.4 0.6 0.8 1.0 0.5 0.6 0.7 0.8 0.9 1.0

(a) Comparative statics for ϵ. (b) Comparative statics for ϕ.

Δπ
0.10

0.08

0.06

0.04

0.02

δ
-1.0 -0.5 0.0 0.5 1.0

(c) Comparative statics for δ.

Figure 1: Expected gain from truthful information sharing. This figure illustrates ∆Eπ,
i.e., the difference in each firm’s equilibrium expected profit under truthful information sharing and
the equilibrium expected profit under no information sharing as a function of ϵ, ϕ, and δ. Baseline
parameter values are specified in Table 1.

the extent of competitive interaction among firms (|δ|), regardless of whether the firms compete

in substitutes or complements. Overall, the conclusion from Proposition 1 and Figure 1 is that

truthful information sharing is more beneficial in cases in which more can be learned from the

reports (i.e., higher the higher the demand uncertainty, ϵ, and/or signal precision, ϕ) and in which

the effects of a firm’s strategy on its rival are more pronounced (i.e., higher |δ|).

9
Table 1: Parameter values. This table presents our parameter values for the base case of the
model.

Parameter choices

Expected demand intercept ᾱ 2

Direct utility β 0.7
Substitutablilty γ 0.35
Relative substitutability δ = γ/β 0.5
Signal precision ϕ 0.75
Demand uncertainty ϵ 0.6

4. Blockchain implementation of truthful information sharing via

staking

In this section, we relax the assumption of truthful information sharing and examine conditions

under which truthful reporting occurs in equilibrium. In doing so, we assume that the firms can

join a blockchain that stores and shares their reports of demand signals. Notably, while blockchain

technology with incentive-compatible consensus mechanism ensures immutability of recorded re-

ports, it does not necessarily ensure truthful reporting. Truthfulness of reporting depends on the

strength of firms’ incentives to (mis)report their signals. Truthful reporting can be ensured by

means of staking a collateral, which is foregone (i.e. transferred to the rival firm or to a third

party) if the signal that is eventually (after output quantities are chosen) revealed and recorded

on the blockchain differs from the initially (before output quantities are chosen) submitted report

of that signal. The collateral is returned to the firm if the report turns out to be truthful and is

foregone otherwise. Such “staking” can be implemented via a smart contract, which is automat-

ically executed when predetermined conditions are met. In our case, the smart contract would

compare the report with the signal and transfer the collateral either back to the firm or elsewhere

(to the rival or to an external party) based on the result of this comparison. Thus, examining the

conditions for truthful reporting on a blockchain amounts to analyzing the minimal collateral that

exceeds the potential gain from misreporting the demand signal.

10
 We begin by establishing that misreporting is only profitable if it takes the form of an inflated

demand signal (i.e. a report of H while the signal is L), whereas it is unprofitable for falsely

reporting demand that is lower than the signal. Assume, without loss of generality, that firm i is

the misreporting one. Assume also that firm −i believes firm i’s report and chooses its quantity

accordingly, whereas firm i takes firm −i’s reaction to firm i’s (potentially biased) report into
−iL
account when deciding on its production quantity. Let ∆πiH,L be the expected profit of firm i if

Si = L and Ri = H (i.e. firm i reports H after receiving signal L), and firm −i receives the signal
−iH
S−i = R−i = L, which it truthfully reports. ∆πiH,L has a similar interpretation except that firm

−i receives and reports S−i = R−i = H.

Proposition 2 Deviation from truthful reporting in a Cournot game can only be profitable for the

misreporting firm, if it reports H instead of L. Specifically,

−iL −iH
∆πiH,L ≥ 0, ∆πiH,L ≥ 0,
−iL −iH
∆πiL,H ≤ 0, ∆πiL,H ≤ 0.

Form i’s gains from reporting H after receiving signal L conditional on firm −i’s signal are:

ᾱ2 δ 2 ϵ(2ϕ − 1) 2(2 − δ) − (4 − 2δ − δ 2 )ϵ(2ϕ − 1)

−iL
∆πiH,L = , (16)
β (4 − δ 2 )2
ᾱ2 δ 2 ϵ(2ϕ − 1) 2(2 − δ) − (4 + 2δ − δ 2 )ϵ(2ϕ − 1)


−iH
∆πiH,L = . (17)
β (4 − δ 2 )2

A firm benefits more from misreporting if the other firm receives a low (high) signal, if δ > 0 (δ < 0).

Specifically,

−iL −iH 4ᾱ2 ϵ2 (1 − 2ϕ)2 δ 2

∆πiH,L − ∆πiH,L = δ. (18)
β (δ 2 − 4)2
| {z }
≥0

−iL −iH
This implies that for δ > 0 (δ < 0), it is sufficient to consider ∆πiH,L (∆πiH,L ) in order to derive the

lowest sufficient collateral that firm i must post (stake) on the blockchain to ensure truthful reporting

of its signal. Since the firms are symmetric, the potential gain for firm −i from misreporting and

11
resulting collateral that is required for firm −i to report truthfully are the same. Hence, the required

collateral, Ci , of firm i is (16) when firms’ products are substitutes, δ > 0, and (17) if they are

complements, δ < 0.

The intuition for the only profitable misreporting being inflation of a low signal is as follows.

In a Cournot game, if the two firms’ products are substitutes (δ > 0), a higher signal by a firm

lowers the optimal output of its rival (since firms’ reaction functions are downward-sloping), raising

the misreporting firm’s profit. Reporting a lower-than-truthful signal, on the other hand, would

lead to an opposite effect – raising the rival’s optimal quantity, which would would hurt the misre-

porting firm. If the two firms’ products are complements (δ < 0), reporting a higher-than-truthful

signal increases the rival’s output (since reaction functions are upward-sloping), benefiting the mis-

reporting firm. On the contrary, reporting a lower-than-truthful signal lowers the rival’s optimal

output, hurting the firm. Overall, reporting a lower-than-truthful signal would be detrimental for

the misreporting firm, both for substitute and complement products.

A related argument is the intuition behind the last part of Proposition 2. Given that a firm

misreports a higher-than-truthful signal, it benefits under both strategic substitutes and comple-

ments. When products are substitutes, the fact that the competitor receives a low signal works to

the benefit of the focal firm and amplifies the misreporting benefit relative to when the competitor

receives a high signal. When competition is in strategic complements, the opposite is true.

The bottom line from Proposition 2 is that truthful reporting, as assumed in Section 3, is not

achieved automatically, i.e. without properly incentivizing the firms. In the case of blockchain-based

information sharing, truthful reporting is achieved by requiring firms to post collateral (stake) that

is foregone in cases in which a firm’s report turns out untruthful, such that the value of the stake

exceeds the benefit of misreporting. If the size of the collateral is chosen correctly (i.e. equalling

or exceeding the one specified in Proposition 2), firms would optimally report truthful signals.

In the next proposition we examine the comparatice statics of required collateral, Ci .

Proposition 3 The collateral required to ensure truthful reporting, Ci , is increasing in ϵ if ϵ or ϕ

are not too high, or if |δ| is not too low. Specifically, the required collateral is increasing in ϵ for

12
 1
ϵ< 2 or ϕ < 34 , and otherwise we have for δ > 0

∂Ci
/ AL ∪ ∀(δ, ϕ) ∈ AL : ∀ϵ < ϵ̂L (δ, ϕ),
> 0, ∀(δ, ϕ) ∈ (19)
∂ϵ
∂Ci
< 0, ∀(δ, ϕ) ∈ AL : ∀ϵ > ϵ̂L (δ, ϕ), (20)
∂ϵ

and for δ < 0

∂Ci
/ AH ∪ ∀(δ, ϕ) ∈ AH : ∀ϵ < ϵ̂H (δ, ϕ),
> 0, ∀(δ, ϕ) ∈ (21)
∂ϵ
∂Ci
< 0, ∀(δ, ϕ) ∈ AH : ∀ϵ > ϵ̂H (δ, ϕ), (22)
∂ϵ

where the sets AL and AH as well as the functions ϵ̂L (δ, ϕ) and ϵ̂H (δ, ϕ) are defined in the proof.

Furthermore, the required collateral is increasing in |δ|. Specifically

∂Ci
> 0, δ > 0,
∂δ
∂Ci
< 0, δ < 0.
∂δ

The gain from misreporting (and, as a result, the required collateral) is increasing in the ex-ante

demand uncertainty, ϵ, for most parameter combinations. However, if the signal precision, ϕ,

and the demand uncertainty, ϵ, are high, it is possible that the gain from signal misreporting is

decreasing in demand uncertainty. This result is a combination of two effects. On the one hand,

the benefit of misreporting is increasing in the information content of the report, which is higher

when the ex-ante-demand uncertainty and the signal precision are higher. On the other hand,

since the benefit of misreporting is driven by increasing a firm’s profit conditional on a low signal,

the absolute increase in profit is relatively low when the demand intercept in the low demand

realization is low (i.e. when demand uncertainty is high). The combination of these two effects

leads to the non-monotonic relation between the required collateral and demand uncertainty. To

elaborate on this intuition, we observe that higher demand uncertainty has two effects. Since firm

i has received a low signal, there is a direct negative effect on firm i’s expected profit because the

market demand is lower with higher demand uncertainty. On the other hand, firm −i also has to

13
take the higher demand uncertainty into account implying that there is more slack for firm i to

benefit from a misleading report. This is an indirect positive effect on firm i’s expected profit. Now

consider an example with ϕ close to 1 and ϵ close to 1. A low signal implies that firm i’s posterior

belief regarding its market demand is close to 0 and this severely limits the value of exploiting firm

−i. Therefore, the direct effect dominates and thus the value of a higher demand uncertainty is

decreasing. In contrast, if ϕ is close to 0.5, then a low signal is not very likely to imply a low market

demand. As a result, there is more slack to profit from misreporting and the positive indirect effect

dominates. The relation between Ci on one hand and ϵ and ϕ on the other hand is depicted in

Figures 2a and 2b for cases of strategic substitutes (δ > 0) and strategic complements (δ < 0),

respectively.

Figure 2c shows the gain from misreporting for different values of δ. Misreporting is more

beneficial the higher the extent of competitive interaction between the firm under both types of

competition. If the firms’ products are strategic complements, δ < 0, the benefit of misreporting

for a given |δ| is larger than in the case of strategic substitutes, δ > 0. The reason is scale:

Equilibrium output quantities and profits are larger under competition in complements, leading to

larger benefits from misreporting.

5. Optimal subsidization

It is possible that a firm is unable to post the collateral required for ensuring truthful reporting

in equilibrium, for example if it is partially financially constrained. One potential way to overcome

this limitation of collateral-based information sharing is to allow for transfers from an unconstrained

firm to a constrained one. Intuitively, in the absence of transfers and if one firm is unable to stake

sufficient funds, the equilibrium would be that of no information sharing. However, if the benefit to

the unconstrained firm from moving to the equilibrium involving both firms reporting their signals

truthfully exceeds the transfer sufficient for the constrained firm to post the required collateral,

the unconstrained firm would benefit from posting partial collateral on behalf of the constrained

firm. We now examine conditions for optimal cross-subsidization of firms’ stakes required for

truthful information sharing in equilibrium. To examine the effects of cash shortfalls on equilibrium

14
 (a) Changes in ϵ and ϕ for a positive value of δ (b) Changes in ϵ and ϕ for a negative value of δ
Ci

0.7

0.6

0.5

0.4

0.3

0.2

0.1

δ
-1.0 -0.5 0.5 1.0

(c) Changes in δ

Figure 2: Collateral required for truthful reporting. The figure depicts the minimal collateral
required for truthful reporting, Ci , as a function of ϵ, ϕ, and δ. Baseline parameter values are
specified in Table 1. In Figure (a) δ = 0.2 and in Figure (b) δ = −0.2.

outcomes, we denote firm i’s (−i’s) liquid funds as Li ≥ 0 L−i ≥ 0.

Proposition 4 Truthful information sharing on a blockchain can be supported depending on the

size of the required collateral compared to firms’ liquid funds. Specifically,

(a) If Li > Ci and L−i > C−i , then the firms join the blockchain and stake the required collateral,

(b) If Li < Ci and L−i < C−i , then the firms cannot stake collateral sufficient for ex-post truthful

reporting and do not join the blockchain,

(c) If Li < Ci and L−i ≥ C−i + (Ci − Li ) and if ∆Eπ−i ≥ Ci − Li , then firm −i subsidizes firm i

with Ci − Li (i.e. stakes Ci − Li on firm i’s behalf ), and the firms stake the collateral and join

15
 gains gains
0.20

0.025
0.15
0.020

0.10 0.015

0.010
0.05
0.005

ϵ ϵ
0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0

(a) Changes in ϵ given a high value of ϕ (b) Changes in ϵ given a low value of ϕ
gains
0.4

0.3

0.2

0.1

δ
-1.0 -0.5 0.5 1.0

(c) Changes in δ

Figure 3: Optimal subsidization. This figure illustrates potential subsidization by unconstrained

firm i of collateral of constrained firm −i, as a function of ϵ and δ. The sub-figures show firm i’s
−iL −iH
potential gain from ex-post misreporting, max{∆πiH,L , ∆πiH,L } (blue dashed line), and firm −i’s
gain from truthful information sharing relative to no sharing, ∆Eπ−i (orange solid line). Baseline
parameter values are as in Table 1. In Figure (a), δ = 0.50 and ϕ = 0.90, in Figure (b), δ = 0.50
and ϕ = 0.55, and in Figure (c), ϵ = 0.6 and ϕ = 0.75.

the blockchain. If ∆Eπ−i < Ci − Li or L−i < C−i + (Ci − Li ), the firms cannot stake sufficient

collateral and do not join the blockchain. (This result is symmetric in the case of L−i < C−i

and Li ≥ Ci + (C−i − L−i ) and if ∆Eπi ≥ C−i − L−i .)

Figure 3 examines the determinants of the optimal level of subsidization. We consider the case

in which firm i is financially constrained, whereas firm −i has deep pockets and may subsidize

firm i’s collateral to achieve truthful reporting. For illustrative purposes, we focus on competition

in strategic substitutes in Panels (a) and (b); the results for strategic complements are similar.

As established in Proposition 2, when competition is in substitutes, δ > 0, the highest gain from

misreporting occurs if the rival firm receives a low demand signal. Therefore, we focus on the case

16
 −iL
in which both firms receive signal L, and firm i may gain ∆πiH,L from reporting H, as indicated by
−iH
the dashed blue line. In Panel (c), we focus on ∆πiH,L , as this is the highest gain from misreporting
−iL
for δ < 0 and on ∆πiH,L for δ > 0. The solid orange line in all Panels depicts the gain that the

deep-pocketed firm −i can achieve by establishing truthful information sharing compared to the

no-sharing scenario.

Panel (a) examines the case of of a rather precise demand signal, ϕ = 0.90. For sufficiently

high values of ϵ, the deep-pocketed firm, −i, may fully subsidize firm i’s collateral. In particular,

for ϵ > 0.66, where the solid orange line is above the dashed blue line, firm −i would be willing

to post the full Ci in behalf of firm i. In other words, the firms join the blockchain even if one

of them is penniless and cannot post even a partial collateral. For ϵ < 0.66, the unconstrained

firm, −i, would subsidize firm i up to the orange line (∆Eπ−i ), so that firm i has to come up

with Li = Ci − ∆Eπ−i in order for both firms to join the blockchain and ensure truthful reporting.

Notably, as long as Li ≥ Ci −∆Eπ−i , firm i will choose to post the required collateral in equilibrium,

as it will be the largest beneficiary of truthful reporting. Panel (b) depicts the case of lower signal

precision, specifically ϕ = 0.55. In this case, the benefit of misreporting exceeds the benefits of

truthful information sharing for all levels of demand uncertainty ϵ. As a result, the cash-rich firm

−i would subsidize firm i only partially, while the remainder of the collateral would need to be

covered by the liquid funds Li of firm i. Panel (c) examines how the optimal subsidization depends

on relative substitutability/ complementarity of the firms’ products, δ. While the gain from ex-post

misreporting (dashed blue line) lies above the benefit of truthful information sharing (solid orange

line), partial subsidization can still be beneficial as long as ∆Eπ−i ≥ Ci − Li .

The conclusion from this analysis is that firms may join the blockchain with truthful information

sharing in equilibrium even if one of the firms does not have the necessary collateral at hand as

long as a) the joint liquid holdings of the two firms exceed the joint collateral required for truthful

reporting and b) the subsidy of one firm’s collateral by another firm does not exceed the latter

firm’s gain from truthful information sharing relative to information sharing.

17
6. Welfare and policy

After having investigated possible transfers between the firms that could occur in equilibrium

in which a cash-rich firm’s gain from truthful information sharing exceeds a constrained firm’s

collateral shortfall, we next examine welfare effects of truthful information sharing. This analysis

could shed light on whether a social planner or government would be willing to subsidize or otherwise

facilitate information sharing among firms. We proceed to compare the representative consumer’s

utility without information sharing, U no sharing , with the utility under truthful information sharing,

which is guaranteed for high enough posted stakes, U truthf ul sharing .

Lemma 1 The representative consumer’s equilibrium utility without information sharing is:

ᾱ2
 
no sharing (1 + δ) 2 2
U = 4 + ϵ (2ϕ − 1) . (23)
4β (2 + δ)2

The representative consumer’s equilibrium utility with truthful information sharing is:

ᾱ2 (2 − δ)2 (1 + δ) + ϵ2 (2ϕ − 1)2 (4 − 3δ 2 )

U truthf ul sharing = . (24)
β (4 − δ 2 )2

The representative consumer’s utility gain from truthful information sharing is:

∆ δ 2 (4 + δ 2 )
∆U = U truthf ul sharing − U no sharing = −ᾱ2 ϵ2 (2ϕ − 1)2 ≤ 0. (25)
4β(4 − δ 2 )2

Lemma 1 establishes that truthful information sharing among producers is not beneficial from the

representative consumer’s point of view. Consequently, for information sharing to be beneficial for

the society at large, the increase in producer surplus should outweigh the reduction in consumer

surplus.

Proposition 5 Truthful information sharing is welfare improving. Specifically,

20 − 3δ 2
∆W = ∆U + 2∆Eπ = ᾱ2 ϵ2 (2ϕ − 1)2 δ 2 ≥ 0, (26)
4β(4 − δ 2 )2

18
and it is increasing in ϵ, ϕ, and |δ|, i.e.,

∂∆W
> 0,
∂ϵ
∂∆W
> 0,
∂ϕ
∂∆W
has the same sign as δ.
∂δ

The effects of ϵ, ϕ, and δ on welfare and its components are illustrated in Figure 4. The main

takeaway is that the joint gain to the two firms from truthful information sharing, 2∆Eπ, is indeed

positive and larger than the absolute value of the representative consumer’s utility loss, ∆U . Thus,

if a social planner evaluates the entire welfare by summing up consumer and producer surplus, it

should favor blockchain-based implementation of truthful information sharing among firms. For

higher values of ex-ante demand uncertainty, ϵ, and signal precision, ϕ, truthful information sharing

via blockchain-based reporting is increasingly welfare improving. If the two firms’ products are

strategic substitutes, higher substitutability raises welfare gain from information sharing. If the

products are complements, then welfare gain is increasing in complementarity.

7. Conclusions and future research

In this paper we study how a smart-contract-enabled blockchain can facilitate truthful informa-

tion sharing among rival firms. Competing firms can benefit from sharing of their demand signals,

as long as they can be sure that their counterparty’s report is truthful. However, while an ex-ante

truthful information sharing agreement is beneficial, firms have ex-post incentives to deviate from

the agreement, i.e., misreport (some of) their signals.

One way to ensure truthful reporting is by requiring firms to post collateral, which is re-

turned if the eventually revealed signal matches the report, and confiscated otherwise. Such a

“staking-based” mechanism can be implemented via a smart contract, which automatically sends

the collateral back to its staker or elsewhere depending on whether the reported demand deviates

from the realized one. We examine the required collateral and its comparative statics with respect

to demand uncertainty, demand signal precision, and the degree of competitive interaction among

19
DeltaUtilityWelfare DeltaUtilityWelfare

0.20 0.20

0.15 0.15

0.10 0.10

0.05 0.05

ϵ ϕ
0.2 0.4 0.6 0.8 1.0 0.6 0.7 0.8 0.9 1.0

-0.05 -0.05

(a) Varying ϵ (b) Varying ϕ

DeltaUtilityWelfare

0.4

0.3

0.2

0.1

δ
-1.0 -0.5 0.5 1.0

-0.1

(c) Varying δ

Figure 4: Welfare gain from truthful information sharing. This figure illustrates how truthful
information sharing impacts overall welfare, ∆W in Proposition 5 (dotted green curve), as well as its
components, representative consumer’s utility, ∆U (dashed blue curve) and the two firms’ expected
profits, 2∆Eπ (solid orange curve), for various ϵ and δ. Parameter values are specified in Table 1.

20
product market rivals.

One limitation of such a staking-based mechanism is that the collateral required for ensuring

truthful information sharing in equilibrium can be larger than firms’ liquid reserves. Financially

constrained firms may thus be unable to post sufficiently high collateral to ensure ex-post truthful

reporting, limiting firms’ ability to share demand signals. Our analysis highlights that it can be

beneficial for competitors to (partially) subsidize their rivals’ collateral required for ex-post truthful

sharing, as the gain from covering the collateral shortfall can outweigh the cost of such subsidization.

We also show that in cases in which the combined liquid holdings of firms are insufficient

for posting the required collateral – rendering truthful information sharing impossible – a central

planner/government may want to fill the collateral shortfall. The reason is that truthful information

sharing is welfare-improving (relative to no information sharing). Falling short of advocating for

direct subsidies from the government to financially constrained firms, we can safely say that a

social planner should be interested in facilitating systems that allow firms to subsidize each other’s

collateral in an efficient and practical way. As long as firms have incentives to subsidize each other,

it may be enough for a regulator to assist with the legal and technical framework that would allow

firms to commit to truthful information sharing. In some cases, truthful reporting of signals may

not be achievable without direct government subsidization. This happens when the combined cash

at the disposal of the two firms is lower than the combined collateral required for incentivizing

truthful reporting. In cases like this, direct subsidies aimed at relaxing firms’ financial constraints

may be conceivable.6

In our analysis, we make several simplifying assumptions that could potentially be relaxed in

future work. First, we assume that firms are symmetric. Thus, we abstract from the effects of

firms’ sizes and/or risk profiles on equilibrium outcomes. Second, extending our duopolistic model

to an oligopoly setting would allow examining additional potential equilibria. In particular, an

interesting question is whether all firms would join a blockchain in equilibrium (with or without

cross-subsidization), or if only some would participate while others not.

6
However, it is important to take into account that government’s subsidies aimed at relaxing financially constrained
firms’ collateral constraints could lead to adverse incentives for the firms, e.g. various agency considerations.

21
References
Abadi, Joseph and Markus Brunnermeier (2019) “Blockchain economics,”Technical report, National Bureau
of Economic Research.
Asgaonkar, Aditya and Bhaskar Krishnamachari (2019) “Solving the buyer and seller’s dilemma: A dual-
deposit escrow smart contract for provably cheat-proof delivery and payment for a digital good without
a trusted mediator,” in 2019 IEEE international conference on blockchain and cryptocurrency (ICBC),
262–267, IEEE.
Biais, Bruno, Agostino Capponi Capponi, Lin William Cong, Vishal Gaur, and Kay Giesecke (2023) “Ad-
vances in blockchain and crypto economics,” Management Science, 69 (11), 6417–6426.
Brandenburger, Adam M (1996) Co-opetition: Doubleday.
Brandenburger, Adam and Barry Nalebuff (2021) “The rules of co-opetition,” Harvard Business Review, 99
(1), 48–57.
Cao, Sean, Lin William Cong, and Baozhong Yang (2023) “Financial Reporting and Blockchains: Misre-
porting, Auditing, and Regulation,” Working Paper.
Catalini, Christian and Joshua Gans (2018) “Initial coin offerings and the value of crypto tokens,” Working
Paper.
Chen, Mark A, Qinxi Wu, and Baozhong Yang (2019) “How valuable is FinTech innovation?” Review of
Financial Studies, 32 (5), 2062–2106.
Chen, Xia, Qiang Cheng, and Ting Luo (2023) “The economic value of blockchain applications: Early
evidence from asset-backed securities,” Management Science, 70 (1), 439–463.
Chiu, Jonathan and Thorsten Koeppl (2019) “Blockchain-based settlement for asset trading,” Review of
Financial Studies, 32 (5), 1716–1753.
Chod, Jiri and Evgeny Lyandres (2021) “A theory of ICOs: Diversification, agency, and information asym-
metry,” Management Science, 67 (10), 5969–5989.
(2023) “Product market competition with crypto tokens and smart contracts,” Journal of Financial
Economics, 149 (1), 73–91.
Chod, Jiri, Nikolaos Trichakis, Gerry Tsoukalas, Henry Aspegren, and Mark Weber (2020) “On the financing
benefits of supply chain transparency and blockchain adoption,” Management Science, 66 (10), 4378–
4396.
Clarke, Richard N (1983) “Collusion and the incentives for information sharing,” Bell Journal of Economics,
383–394.
Cong, Lin William and Zhiguo He (2019) “Blockchain disruption and smart contracts,” Review of Financial
Studies, 32 (5), 1754–1797.
Cong, Lin William, Yuanyu Qu, and Guojun Wang (2024) “Blockchain for environmental monitoring: Theory
and empirical evidence from China,” Working Paper.
Cui, Yao, Vishal Gaur, and Jingchen Liu (2023) “Supply chain transparency and blockchain design,” Man-
agement Science, 70 (5), 3245–3263.
Davydiuk, Tetiana, Deeksha Gupta, and Samuel Rosen (2023) “De-crypto-ing signals in initial coin offerings:
Evidence of rational token retention,” Management Science, 69 (11), 6584–6624.
Dong, Lingxiu, Yunzhe Qiu, and Fasheng Xu (2023) “Blockchain-enabled deep-tier supply chain finance,”
Manufacturing and Service Operations Management, 25 (6), 2021–2037.
Dutta, Pankaj, Tsan-Ming Choi, Surabhi Somani, and Richa Butala (2020) “Blockchain technology in supply
chain operations: Applications, challenges and research opportunities,” Transportation research part e:
Logistics and transportation review, 142, 102067.
Gan, Jingxing, Gerry Tsoukalas, and Serguei Netessine (2021) “Initial coin offerings, speculation, and asset
tokenization,” Management Science, 67 (2), 914–931.
(2023) “Decentralized platforms: Governance, tokenomics, and ICO design,” Management Science,
69 (11), 6667–6683.

22
Gnyawali, Devi R and Ravindranath Madhavan (2001) “Cooperative networks and competitive dynamics:
A structural embeddedness perspective,” Academy of Management Review, 26 (3), 431–445.
Gnyawali, Devi R and Byung-jin Park (2009) “Co-opetition and technological innovation in small and
medium-sized enterprises: A multilevel conceptual model,” Journal of Small Business Management,
47 (3), 308–330.
Gnyawali, Devi R and Byung-Jin Robert Park (2011) “Co-opetition between giants: Collaboration with
competitors for technological innovation,” Research Policy, 40 (5), 650–663.
Gnyawali, Devi R and Tadhg Ryan Charleton (2018) “Nuances in the interplay of competition and cooper-
ation: Towards a theory of coopetition,” Journal of Management, 44 (7), 2511–2534.
Goldstein, Itay, Wei Jiang, and G Andrew Karolyi (2019) “To FinTech and beyond,” Review of Financial
Studies, 32 (5), 1647–1661.
Gryglewicz, Sebastian, Simon Mayer, and Erwan Morellec (2021) “Optimal financing with tokens,” Journal
of Financial Economics, 142 (3), 1038–1067.
Han, Hongdan, Radha K Shiwakoti, Robin Jarvis, Chima Mordi, and David Botchie (2023) “Accounting
and auditing with blockchain technology and artificial Intelligence: A literature review,” International
Journal of Accounting Information Systems, 48, 100598.
Hoffmann, Werner, Dovev Lavie, Jeffrey J Reuer, and Andrew Shipilov (2018) “The interplay of competition
and cooperation,” Strategic Management Journal, 39 (12), 3033–3052.
Iyengar, Garud, Fahad Saleh, Jay Sethuraman, and Wenjun Wang (2023) “Economics of permissioned
blockchain adoption,” Management Science, 69 (6), 3415–3436.
Kirby, Alison J (1988) “Trade associations as information exchange mechanisms,” RAND Journal of Eco-
nomics, 138–146.
Kshetri, Nir (2023) “Blockchain’s role in promoting quality, safety and sustainability in the food and beverage
industry,” in 32nd European Conference of the International Telecommunications Society (ITS).
Lacity, Mary C (2018) “Addressing key challenges to making enterprise blockchain applications a reality.,”
MIS Q. Executive, 17 (3), 3.
Lado, Augustine A, Nancy G Boyd, and Susan C Hanlon (1997) “Competition, cooperation, and the search
for economic rents: A syncretic model,” Academy of Management Review, 22 (1), 110–141.
Lavie, Dovev (2007) “Alliance portfolios and firm performance: A study of value creation and appropriation
in the US software industry,” Strategic Management Journal, 28 (12), 1187–1212.
Leiponen, Aija Elina (2008) “Competing through cooperation: The organization of standard setting in
wireless telecommunications,” Management Science, 54 (11), 1904–1919.
Lyandres, Evgeny, Barardino Palazzo, and Daniel Rabetti (2022) “Initial coin offering (ICO) success and
post-ICO performance,” Management Science, 68 (12), 8658–8679.
Malinova, Katya and Andreas Park (2023) “Tokenomics: When tokens beat equity,” Management Science,
69 (11), 6568–6583.
Malueg, David A and Shunichi O Tsutsui (1996) “Coalition-proof information exchanges,” Journal of Eco-
nomics, 63, 259–278.
Mathews, John A (2002) “The origins and dynamics of Taiwan’s R&D consortia,” Research Policy, 31 (4),
633–651.
Novshek, William and Hugo Sonnenschein (1982) “Fulfilled expectations Cournot duopoly with information
acquisition and release,” Bell Journal of Economics, 214–218.
Raith, Michael (1996) “A general model of information sharing in oligopoly,” Journal of Economic Theory,
71 (1), 260–288.
Sarker, Suprateek, Stefan Henningsson, Thomas Jensen, and Jonas Hedman (2021) “The use of blockchain
as a resource for combating corruption in global shipping: an interpretive case study,” Journal of
Management Information Systems, 38 (2), 338–373.
Shakhnov, Kirill and Luana Zaccaria (2023) “Utility tokens, network effects, and pricing power,” Manage-
ment Science, 69 (11), 6625–6640.

23
Swan, Melanie (2015) Blockchain: Blueprint for a new economy: ” O’Reilly Media, Inc.”.
Vives, X (1999) Oligopoly Pricing: Old Ideas and New Tools: The MIT Press.
Vives, Xavier (1984) “Duopoly information equilibrium: Cournot and Bertrand,” Journal of Economic
Theory, 34 (1), 71–94.
Yermack, David (2017) “Corporate governance and blockchains,” Review of Finance, 21 (1), 7–31.

24
Appendix: Proofs
Proof. Proposition 1
From 4 and 7 we note that

∆Eπ = Eπ BC − Eπibase (27)

(δ − 2)2 + ϵ2 (2ϕ − 1)2 (4 + δ 2 ) 4 + ϵ2 (2ϕ − 1)2 (2 + δ)2
= ᾱ2 2 2
− ᾱ2 (28)
β(4 − δ ) 4β(2 + δ)2
ϵ2 (2ϕ − 1)2 δ 2 (12 − δ 2 )
= ᾱ2 ≥ 0, (29)
4β(4 − δ 2 )2

because all factors are non-negative. All factors are strictly positive if ϕ > 1/2, ϵ > 0, |δ| > 0, and
ᾱ > 0. The sensitivities in (13) and (14) follow easily by inspection. Taking the derivative of (29)
with respect to δ and rearranging terms yields

∂∆Eπ (2ϕ − 1)2 (12 + δ 2 )

= 2ᾱ2 ϵ2 δ. (30)
∂δ β 2 (4 − δ 2 )3

Assuming ∆Eπ ̸= 0, it follows that the right-hand side has the same sign as δ.

Proof. Proposition 2
To calculate the benefit of misreporting, we consider the difference of firm i’s expected profit when
it misreports and when it reports its signal truthfully. To do so, we derive the optimal quantities
for the firms both with truthful reporting and with misreporting (assuming the other firm sends a
truthful report). We report the result in (8) for the former case. The latter case follows by similar
calculations. We plug these quantities into the expected profit for firm i and obtain:

ᾱ2 δ 2 ϵ(2ϕ − 1) 2(2 − δ) − (4 − 2δ − δ 2 )ϵ(2ϕ − 1)

−iL
∆πiH,L = , (31)
β (4 − δ 2 )2
ᾱ2 δ 2 ϵ(2ϕ − 1) 2(2 − δ) − (4 + 2δ − δ 2 )ϵ(2ϕ − 1)


−iH
∆πiH,L = , (32)
β (4 − δ 2 )2
−iL ᾱ2 δ 2 ϵ(2ϕ − 1)(−2(2 − δ) − (4 + 2δ − δ 2 )ϵ(2ϕ − 1))
∆πiL,H = , (33)
β(4 − δ 2 )2
−iH ᾱ2 δ 2 ϵ(2ϕ − 1)(−2(2 − δ) − (4 − 2δ − δ 2 )ϵ(2ϕ − 1))
∆πiL,H = . (34)
β(4 − δ 2 )2
−iL
We focus on the benefit of misreporting in (31). We observe that ∆πiH,L is a product of a linear
function, f1 , and an affine function, f2 , of ϵ, where

ᾱ2 δ 2 (2ϕ − 1)
f1 = ϵ, (35)
β(4 − δ 2 )2
f2 = 2(2 − δ) − 4 − 2δ − δ 2 (2ϕ − 1)ϵ.


(36)

Given the parameter restrictions, f1 is positive (disregarding the “= 0” cases). Since |δ| < 1, the
slope of f2 is negative and the intercept is positive. Thus, we need to consider when f2 becomes

25
negative. f2 ≥ 0 iff.

− 4 − 2δ − δ 2 (2ϕ − 1)ϵ + 2(2 − δ) ≥ 0,

(37)

i.e.,

2(2 − δ) ∆
ϵ≤ 2
= ϵ0 . (38)
(4 − 2δ − δ ) (2ϕ − 1)

Now we analyze ϵ0 to see whether it is below 1. If it is not, then it is not a binding constraint.
Hence, suppose

ϵ0 ≤ 1 (39)

i.e.,

2(2 − δ) ≤ 4 − 2δ − δ 2 (2ϕ − 1).

(40)

The right-hand side is least restrictive for ϕ = 1, so we consider this case:

2(2 − δ) ≤ 4 − 2δ − δ 2 (41)

and this is true iff

0 ≤ −δ 2 , (42)
−iL
which is contradiction unless δ = 0.Thus ∆πiH,L > 0.
−iH
For ∆πiH,L we can similarly define the function

f2H = 2(2 − δ) − 4 + 2δ − δ 2 (2ϕ − 1)ϵ,

(43)

and by inspection we end up with a condition similar to (41)

2(2 − δ) ≤ 4 + 2δ − δ 2 (44)

iff

0 ≤ (4 − δ)δ, (45)
−iH
which is a contradiction. Thus ∆πiH,L > 0.
By inspection of (33) and (34) we observe that the intercept in the corresponding f2 and f2H
−iL −iH
functions are negative. It trivially follows that ∆πiL,H < 0 and ∆πiL,H < 0.
To prove the last assertion, note that

ᾱ2 δ 2 ϵ(2ϕ − 1)
 
−iL −iH 2 2
∆πiH,L − ∆πiH,L = − (4 − 2δ − δ )ϵ(2ϕ − 1) + (4 + 2δ − δ )ϵ(2ϕ − 1) , (46)
β (4 − δ 2 )2
ᾱ2 δ 2 ϵ(2ϕ − 1)
 
2 2
= − 4 + 2δ + δ + (4 + 2δ − δ ) ϵ(2ϕ − 1), (47)
β (4 − δ 2 )2
ᾱ2 δ 2 ϵ2 (2ϕ − 1)2
= 4δ, (48)
β (4 − δ 2 )2

26
which is equal to (18).

Proof. Proposition 3
Suppose δ > 0. The collateral required to ensure truthful reporting, Ci , i.e., the relevant benefit
−iL −iL
of misreporting is ∆πiH,L , and from (16) we observe that ∆πiH,L is a product of a linear function,
f1 , and an affine function, f2 , of ϵ, where

ᾱ2 δ 2 (2ϕ − 1)
f1 = ϵ, (49)
β(4 − δ 2 )2
f2 = 2(2 − δ) − 4 − 2δ − δ 2 (2ϕ − 1)ϵ.


(50)

To prove (19) we consider the partial derivative with respect to ϵ

∂ −iL ∂ ∂f1 ∂f2

∆πiH,L = (f1 f2 ) = f2 + f1 ≥0 (51)
∂ϵ ∂ϵ ∂ϵ ∂ϵ
iff
1 1
f2 ∂f2 ≤ −f1 ∂f1 = −ϵ, (52)
∂ϵ ∂ϵ

because f1 is linear. Inserting the expression for f2 the assertion holds iff
2−δ ∆
ϵ≤ = ϵ̂L (δ, ϕ). (53)
(4 − 2δ − δ 2 ) (2ϕ − 1)

The auxiliary function ϵ̂L (δ, ϕ) is helpful, because if this is above 1, then ϵ always increases the
value of misreporting. The function is clearly decreasing in ϕ, so to find the strongest restriction
on ϵ we consider ϵ̂L (δ, 1):

2−δ
ϵ̂L (δ, 1) = , (54)
4 − 2δ − δ 2
and taking the derivative we get

d L (4 − δ)δ
ϵ̂ (δ, 1) = , (55)
dδ (4 − 2δ − δ 2 )2

and since δ ∈ (0, 1) the expression in (55) is positive (and 0 when δ = 0). From this we observe
that the minimum of ϵ̂L (δ, ϕ) is ϵ̂L (0, 1) = 1/2. Whence it follows that if ϵ < 1/2, then the benefit
of misreporting is increasing in ϵ. We need to analyze when ϵ̂L (δ, ϕ) = 1. We can do so by either
considering ϕ as a function of δ or the opposite. The first method yields

6 − 3δ − δ 2
ϕ̂(δ) = , (56)
2(4 − 2δ − δ 2 )

with
′ (4 − δ)δ
ϕ̂ (δ) = > 0, (57)
2(δ 2+ 2δ − 4)2

thus ϕ̂ has a minimum at δ = 0. For a given δ we now have that the benefit of misreporting is

27
 ϕ
1.0

0.9

0.8

0.7

0.6

δ
-1.0 -0.5 0.0 0.5 1.0

Figure 5: The figure illustrates the effect of δ on ϕ̂.

increasing in ϵ for ϕ < ϕ̂(δ). If ϕ > ϕ̂(δ), the benefit of misreporting is increasing for ϵ < ϵ̂L (δ, ϕ)
and decreasing otherwise.
The second method is if we consider δ for a given ϕ. Denote the relation as δ̂, then δ̂ is a
correspondence:
 √
 −4ϕ+3− 80ϕ2 −104ϕ+33 = ∆ L
δL (ϕ)
2(2ϕ−1)
√
δ̂(ϕ) = 2
(58)
 −4ϕ+3+ 80ϕ −104ϕ+33 = ∆ L
δ (ϕ).
2(2ϕ−1) H

It is only relevant to consider both boundaries when δL (ϕ) ≤ δH (ϕ). It is easily seen that the
11
boundaries coincide when ϕ = 20 or ϕ = 34 . The former case implies δ = 4, the latter case implies
δ = 0. As δ ∈ (0, 1), it follows that ϕ = 43 is the relevant level. That is, if ϕ < 34 , then ϵ̂(δ, ϕ) > 1 in
which case the benefit of misreporting is always increasing in ϵ (because ϵ is bounded from above by
1 per assumption). For ϕ ≥ 43 , the benefit of misreporting is decreasing in ϵ when δ ∈ (δL (ϕ), δH (ϕ))
and ϵ > ϵ̂L (δ, ϕ). It is increasing otherwise. We depict this relationship in Figure 5. From the above,
we define:

AL = {(δ, ϕ) ∈ (0, 1) × ( 34 , 1]|ϕ > ϕ̂(δ)}, (59)

= {(δ, ϕ) ∈ (0, 1) × ( 34 , 1]|δ ∈ (δL (ϕ), δH (ϕ))}, (60)

−iH
Suppose δ < 0. From (17) we observe as above that ∆πiH,L is a product of two functions:

ᾱ2 δ 2 (2ϕ − 1)
f1 = ϵ, (61)
β(4 − δ 2 )2
f2H = 2(2 − δ) − 4 + 2δ − δ 2


(2ϕ − 1)ϵ. (62)

28
To prove (21) we consider the partial derivative with respect to ϵ

∂ −iH 1
∆πiH,L ≥ 0 ⇐⇒ f2H H ≤ −ϵ, (63)
∂ϵ ∂f 2
∂ϵ

and inserting the expression for f2H the assertion holds iff

2−δ ∆
ϵ≤ = ϵ̂H (δ, ϕ). (64)
(4 + 2δ − δ 2 ) (2ϕ − 1)

If the auxiliary function ϵ̂H (δ, ϕ) is above 1, then ϵ always increases the value of misreporting. The
function is clearly decreasing in ϕ, so to find the strongest restriction on ϵ we consider ϵ̂H (δ, 1):

2−δ
ϵ̂H (δ, 1) = , (65)
4 + 2δ − δ 2
and taking the derivative we get

d H 8 − 4δ + δ 2
ϵ̂ (δ, 1) = − , (66)
dδ (4 + 2δ − δ 2 )2

and since δ ∈ (−1, 0) the expression in (66) is negative with a minimum when δ = 0. From this we
observe that the minimum of ϵ̂H (δ, ϕ) is ϵ̂H (0, 1) = 1/2. Whence it follows that if ϵ < 1/2, then the
benefit of misreporting is increasing in ϵ. We need to analyze when ϵ̂H (δ, ϕ) = 1. We can do so by
either considering ϕ as a function of δ or the opposite. The first method yields

H 6 + δ − δ2
ϕ̂ (δ) = , (67)
2(4 + 2δ − δ 2 )

with
H 8 − 4δ + δ 2
ϕ̂ ′ (δ) = − < 0, (68)
2(4 + 2δ − δ 2 )2
H
thus ϕ̂ has a minimum at δ = 0. For a given δ we now have that the benefit of misreporting is
H
increasing in ϵ for ϕ < ϕ̂(δ). If ϕ > ϕ̂ (δ), the benefit of misreporting is increasing for ϵ < ϵ̂H (δ, ϕ)
and decreasing otherwise.
H H
We now consider δ for a given ϕ. Denote the relation as δ̂ , then δ̂ is a correspondence:
 √
4ϕ−1− 25−88ϕ+80ϕ2 ∆ H
H

2(2ϕ−1) = δL (ϕ)
δ̂ (ϕ) = √ (69)
2 ∆ H
 4ϕ−1+ 25−88ϕ+80ϕ = δ (ϕ).
2(2ϕ−1) H

It is only relevant to consider the boundaries when they are below 0 (because we consider the case
δ < 0). But δH H (ϕ) ≤ 0 iff

p
4ϕ − 1 + 25 − 88ϕ + 80ϕ2
≤ 0, (70)
2(2ϕ − 1)

and since ϕ ∈ ( 12 , 1] this is not possible. Hence δH

H (ϕ) > 0 always and does not provide a restriction

29
on δ. We now consider δLH (ϕ). Note that
p
3 − 4ϕ − 25 − 88ϕ + 80ϕ2
δLH ′ (ϕ) = p . (71)
(2ϕ − 1)2 25 − 88ϕ + 80ϕ2

This is negative if
p
(3 − 4ϕ)2 < ( 25 − 88ϕ + 80ϕ2 )2 , (72)

i.e., if

9 + 16ϕ2 − 24ϕ < 25 − 88ϕ + 80ϕ2 , (73)

iff

16(2ϕ − 1)2 > 0, (74)

which always holds for ϕ > 1/2. Hence, δLH (ϕ) is decreasing in ϕ. Now δLH (ϕ) = 0 iff
p
4ϕ − 1 − 25 − 88ϕ + 80ϕ2
=0 (75)
2(2ϕ − 1)

iff
p
4ϕ − 1 − 25 − 88ϕ + 80ϕ2 = 0 (76)

iff

ϕ ∈ { 12 , 43 }, (77)

and since ϕ > 1/2 we have that δLH (ϕ) = 0 iff ϕ = 34 . That is, if ϕ < 34 , then ϵ̂H (δ, ϕ) > 1 in which
case the benefit of misreporting is always increasing in ϵ. For ϕ ≥ 43 , the benefit of misreporting is
decreasing in ϵ when δ ∈ (δLH (ϕ), 0) and increasing otherwise. From the above, we define:
H
AH = {(δ, ϕ) ∈ (−1, 0) × ( 34 , 1]|ϕ < ϕ̂ (δ)}, (78)
= {(δ, ϕ) ∈ (−1, 0) × ( 34 , 1]|δ ∈ (δLH (ϕ), 0)}. (79)

This proves the sensitivity effects wrt. ϵ.

We now turn to the sensitivity wrt. δ. Let δ > 0. We consider the derivative
−iL
∂∆πiH,L 2ᾱ2 δϵ(2ϕ − 1)
 
3 2 3 2


= − δ + 4δ − 12δ + 16 + ϵ(2ϕ − 1) δ + 4δ + 12δ − 16 . (80)
∂δ β (4 − δ 2 )3

The first factor has the same sign as δ, so it suffices to study the sign of the last factor. We
conjecture:

−δ 3 + 4δ 2 − 12δ + 16 +ϵ(2ϕ − 1) δ 3 + 4δ 2 + 12δ − 16 ≥ 0.

(81)
| {z } | {z }
∆ ∆
=h1 >0 =h2

To show that h1 > 0 we note that h1 = (δ − 2)(δ 2 − 2δ + 8). Hence δ = 2 is the only real-valued

30
solution (because the parabola has a minimum in δ = 1 with a value equal to 7). Since h1 = 16 > 0
for δ = 0, it follows that h1 > 0 for δ < 2. Consider h2 . If h2 ≥ 0, then h1 + h2 > 0 and we are
done. If h2 < 0, then we require

h1 ≥ −ϵ(2ϕ − 1)h2 > 0 (82)

and it suffices if

h1 ≥ −h2 (83)

iff

−δ 3 + 4δ 2 − 12δ + 16 + (δ 3 + 4δ 2 + 12δ − 16) ≥ 0 (84)

i.e.,

8δ 2 ≥ 0, (85)
∂ 2L ∂ 2L |
which is true. Whence it follows that ∂δ ∆π1H,L has the same sign as δ. Since ∂δ ∆π1H,L δ=0 = 0,
the proposition is proved.
Let δ < 0. We consider the derivative
−iH
∂∆πiH,L 2ᾱ2 δϵ(2ϕ − 1)
 
3 2 3 2


= − δ + 4δ − 12δ + 16 + ϵ(2ϕ − 1) −δ + 4δ − 12δ − 16 . (86)
∂δ β (4 − δ 2 )3

The first factor has the same sign as δ, so it suffices to study the sign of the last factor. We
conjecture:

−δ 3 + 4δ 2 − 12δ + 16 +ϵ(2ϕ − 1) δ 3 + 4δ 2 + 12δ − 16 ≥ 0.

(87)
| {z } | {z }
∆ ∆
=h1 >0 =hH
2

We know that h1 > 0 from above. Consider hH H H

2 . If h2 ≥ 0, then h1 + h2 > 0 and we are done. If
hH
2 < 0, then we (as before) require

h1 ≥ −hH
2 (88)

which holds iff

−δ 3 + 4δ 2 − 12δ + 16 + (−δ 3 + 4δ 2 − 12δ − 16) ≥ 0 (89)

i.e.,

−δ 3 + 4δ 2 − 12δ ≥ 0, (90)
∂ −iH
which is true. Whence it follows that ∂δ ∆πiH,L has the same sign as δ.

Proof. Proposition 4
Case (a) in Proposition 4 holds because ∆Eπi = ∆Eπ−i > 0 and thus either firm has an incentive
to provide the collateral and they have the means to do so. Case (b) holds because the firms cannot

31
provide the necessary funds. Case (c) follows by first observing that firm i cannot submit Ci ; it is
short Ci − Li in liquid funds. Firm −i can provide the collateral C−i and is left with L−i − C−i in
liquid funds. If what is left is high enough, i.e., L−i − C−i ≥ Ci − Li , then firm −i can potentially
pay the remaining collateral for firm i. This will be a sunk cost to firm −i, and thus firm −i
is only willing to do so if its benefit of a blockchain is large enough to compensate for this, i.e.,
∆Eπ−i ≥ Ci − Li . This is symmetric in case firm −i cannot submit C−i .

Proof. Proposition 5
The sensitivity effects are trivial except for the effect in δ. To consider the sensitivity effect in ∆W ,
we first define ξ = δ 2 and observe that

∂ 1 ∂ 20ξ − 3ξ 2
∆W = ᾱ2 ϵ2 (2ϕ − 1)2 , (91)
∂ξ 4β ∂ξ (4 − ξ)2
1 (4 − ξ)2 (20 − 6ξ) − (20ξ − 3ξ 2 )2(4 − ξ)(−1)
= ᾱ2 ϵ2 (2ϕ − 1)2 , (92)
4β (4 − ξ)2
1 (4 − ξ)(20 − 6ξ) + (20ξ − 3ξ 2 )2
= ᾱ2 ϵ2 (2ϕ − 1)2 , (93)
4β (4 − ξ)
1 80 − 4ξ
= ᾱ2 ϵ2 (2ϕ − 1)2 , (94)
4β (4 − ξ)

and an application of the chain rule implies that

∂ 20 − δ 2
∆W = 2ᾱ2 ϵ2 (2ϕ − 1)2 δ , (95)
∂δ β(4 − δ 2 )2

which has the same sign as δ.

32