CAD COIN UPDATES: PROJECT JASPER BANK OF CANADA
Project Jasper:
Are Distributed Wholesale
Payment Systems Feasible Yet?
James Chapman, Rodney Garratt,1
Scott Hendry, Andrew McCormack2
and Wade McMahon
Distributed ledger technology (DLT)—most commonly known as the foundation
of Bitcoin—offers a fundamentally different way to conduct and
track financial transactions. Researchers are investigating its usefulness
in all corners of the financial system.
Project Jasper is a proof of concept of a DLT-based wholesale payment
system. The experiment provided significant insights into the
relative strengths and weaknesses of using DLT for financial market
infrastructures.
For critical financial market infrastructures, such as wholesale payment
systems, current versions of DLT may not provide an overall net benefit
relative to current centralized systems. Recent versions of DLT have,
however, made advances compared with initial cryptocurrency applications
of DLT.
Benefits for the financial system of a DLT-based wholesale payment system
could likely arise from its interaction with a larger DLT ecosystem of financial
market infrastructures, potentially including cross-border transactions.
Introduction
Financial technology (fintech) is defined as financial innovation enabled by
technology that could result in new business models, applications, processes
or products and that has an associated material effect on financial
markets and institutions or the provision of financial services.3
One such innovation with significant potential is distributed ledger technology
(DLT), or blockchain, as a common variant of it is known (Box 1).
DLT, introduced with the cryptocurrency Bitcoin in 2008 (Nakamoto 2008),
enables the secure validation and recording of transactions. A distributed
ledger is a database shared between multiple parties. It allows those parties
1 University of California Santa Barbara and R3.
2 Payments Canada.
3 See Schindler (forthcoming) for a discussion of the drivers of fintech.
Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017 1 to execute mutually agreed-upon transactions and achieve consensus
on changes to the database. In this way, it ensures consistency between
parties. The key feature of a distributed ledger is that authorized parties,
through the use of a consensus mechanism, share identical versions of the
data without the need for a central database or central administrator.4
A more general-purpose DLT platform called Ethereum was launched in
- It allows any type of digital asset to be defined, created and traded.
It also enables smart contracts, which allow a DLT to execute the terms of
a contract automatically, providing more functionality than simply transferring
one specific type of asset (Buterin 2013). These developments sparked
tremendous interest from the financial sector. The shared nature of the
underlying ledger can offer a number of potential benefits, including process
and cost efficiencies, resilience and interoperability. However, there are also
a number of challenges in adapting DLT to financial sector applications,
including the speed of transacting, achieving finality of the transactions, and
privacy. Recent fintech companies have developed more general DLT systems,
such as Corda5
by R3, to meet the needs of the financial sector.
4 For an overview of DLT and the policy issues surrounding it, refer to CPMI (2017). For a technical but
accessible introduction to some of the concepts in this article, refer to Narayanan et al. (2016).
5 Corda is an open-source distributed ledger platform designed to record, manage and automate legal
agreements between businesses.
What Is a Distributed Ledger System?
Distributed ledger technology became popular following
the introduction of the cryptocurrency Bitcoin in 2009. A
bitcoin (lower case b) is a digital token that represents a
digital money. Bitcoin (upper case B) is a system to transfer
bitcoins between people. this is done via a ledger of transactions
that is visible to all and maintained by a distributed
system of “miners” who operate computers that are nodes
on the system. these nodes update the ledger with new
transactions as they are made. the ledger is designed as a
series of blocks of transactions linked together using cryptography.
this ledger is known as a blockchain.
this system was a breakthrough because it demonstrated a
way to maintain a ledger of information between parties in
such a way that (i) no one oversees the system and (ii) the
ledger can be credibly updated and agreed upon by members
of the Bitcoin system even though no one trusts any
other member to act honestly.
this “trustless” updating of a distributed ledger is achieved
by having miners compete to win the right to validate blocks
of transactions by solving a diffi cult mathematical puzzle.
the fi rst miner who completes a new puzzle broadcasts
the block and solution to all the other miners and in return
“mines” new bitcoins created with that block. Although the
problem miners work on is diffi cult to solve, it is easy to
verify. Once the other nodes have seen and verifi ed a new
solution, the new block is added to the chain, the transactions
in the block are considered settled and miners begin
mining a new set of transactions. the way nodes come to
agreement about the new block is called a consensus mechanism,
and the puzzle is the proof of work (Pow).
while the Bitcoin system has proven to be quite resilient,
a number of aspects undermine its suitability for fi nancial
market infrastructures: (i) all transactions are visible to
everyone, which may, for example, violate banking laws and
put certain parties involved in transactions at a disadvantage;
(ii) Pow is very costly in terms of time and energy
and its benefi ts are not typically needed in trusted environments;
and (iii) the system is open to anyone who would like
to join and participants are anonymous.
to address these issues, fi nancial technology companies
have been developing alternatives to the Bitcoin system.
these new distributed ledger systems allow access only to a
restricted set of trusted counterparties. in some systems the
consensus mechanism is replaced by other methods to
reach agreement. in the Corda platform used in Phase 2 of
Project Jasper, this is done via a notary node that is tr usted
by everyone and replaces the Pow function. Finally, these
systems dispense with the concept of a blockchain and
replace it with a ledger that is still distributed among the
nodes, but where each node has access only to necessary
data. this aff ords less transparency across the system
and allows more privacy for participants.
2 Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017
Financial sector participants are interested in this distributed ledger technology
for several reasons. It has the potential to reduce back-office costs
by automating various settlement processes. It can increase the reliability
and traceability of information stored in the ledger, since the consensus
mechanism puts limits on who can change records and how they can
change them. Finally, with decentralized processes, settlement of transactions
could be faster—reduced to hours or minutes instead of days.
One area of interest has been the potential implications of DLT for financial
market infrastructures (FMIs). FMIs act as the trusted third party between
financial institutions, tracking and recording transactions in centralized
ledgers. Operators of FMIs, participants and central banks are all interested
in the efficiencies and opportunities that a DLT-based system could provide
relative to current centralized systems. As a result, many recent DLT
advancements have focused on ways for traditional operators of centralized
systems to realize the benefits of DLT while mitigating its disadvantages.
For example, a common trend has been toward creating DLT systems
that restrict access to a group of trusted entities. This contrasts with open
arrangements like Bitcoin, where any entity can participate. To date, central
banks have implemented DLT only in proofs of concept, and further examination
of potential DLT applications can be expected.
One of the areas being investigated is the possible application of DLT to
wholesale payment systems. Canada’s existing wholesale payment system
is the Large Value Transfer System (LVTS), operated by Payments Canada.
The LVTS processes an average of $175 billion in payments each business
day. It has been designated a systemically important FMI and is overseen by
the Bank of Canada in accordance with the Principles for Financial Market
Infrastructure (PFMIs).6
Wholesale payment systems make sense as an early potential application
of DLT because they are relatively simple. They are also critical for financial
stability. It is therefore important that overseers, like the Bank of Canada,
understand how the use of DLT could change the way centralized systems
are structured and operate, whether a DLT system could meet existing
international standards, and any potential implications for payment system
policy.
In 2016, Payments Canada, along with the Bank of Canada, R3 and
Canadian commercial banks that are members of the R3 consortium, initiated
an experimental project, code-named Project Jasper, to explore a DLTbased
wholesale payment system.7
The immediate goal of Project Jasper
was to build a proof-of-concept system (with no intention of advancing to
a production-level system) that leveraged a settlement asset issued and
controlled by a central bank. In the first phase (Phase 1), participants built a
settlement capability on an Ethereum platform and demonstrated its ability
to exchange a settlement asset between participants. The second phase
(Phase 2), built on a Corda platform, incorporated a liquidity-saving mechanism
(LSM) that allows participants to coordinate their payments to reduce
liquidity needs. As part of Phase 2, the participants are preparing a longer
white paper, to be published by the end of June 2017, outlining the detailed
technical and policy implications of the work.
6 The PFMIs are a set of international standards for systemically important payment systems established
by the Bank for International Settlements (CPSS-IOSCO 2012).
7 R3 is an international consortium of large banks with the goal of investigating and developing applications
of DLT for the financial sector. Participating Canadian members are BMO Bank of Montreal,
Canadian Imperial Bank of Commerce, HSBC, National Bank of Canada, Royal Bank of Canada,
Scotiabank and TD Canada Trust. These seven institutions are also members of Payments Canada,
and they are all participants in the LVTS.
Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017 3
One of the main lessons from this experiment is that the versions of distributed
ledger currently available may not provide an overall net benefit when
compared with existing centralized systems for interbank payments. Core
wholesale payment systems function quite efficiently. There may, however,
be net benefits for the broader group of payment system participants and
the entire financial system from a DLT-based wholesale payment system
in terms of savings from reduced back-office reconciliation and improved
interaction with a larger DLT ecosystem of financial market infrastructures.
Below is a high-level overview of the project and the preliminary findings.
Key Features of Project Jasper
Project Jasper provided vital insights into how a central bank and participating
financial institutions can complete interbank payments on a distributed
ledger.8
The project also offered an understanding of the functioning
of a wholesale payment system using different DLT platforms and how
modern payment system features, such as queues, could be incorporated
to increase efficiency by reducing collateral needs. Finally, developing a
working prototype improved awareness of potential risks associated with
DLT-based systems and how they can be mitigated.
The first key challenge in developing Project Jasper was establishing how
to transfer value. The PFMIs require that an FMI settle in central bank
money whenever practical and available. This usually means settling using
accounts at the central bank. To do this, the concept of a digital depository
receipt (DDR) was used to represent Bank of Canada deposits. A DDR is a
digital representation of currency that is issued by the Bank of Canada; it
could be one approach for a wider use of central bank money in the future
(Garratt 2017). DDRs are issued in the system by the Bank of Canada and
are backed one for one by cash pledged to the Bank by participants. The
exchange of DDRs for central bank money means there is no increase in
money circulating in the banking system.
The DDRs are used by participants in the system to exchange and settle
interbank payments. The processing cycle of Project Jasper achieved
ultimate settlement finality on the books of the Bank of Canada after
exchanging DDRs with the Bank of Canada for Canadian dollars transferred
into their respective settlement accounts. For all intents and purposes, these
DDRs functioned as cash in the system.
The second key challenge was how to most efficiently settle payments with
the minimum amount of DDRs or liquidity. Historically, interbank payments
were settled using systems that conduct end-of-day netting between participants.
But as volumes and values increased in these systems, central
banks became concerned about the risks inherent in netting. In response,
most central banks have opted for the implementation of real-time gross
settlement (RTGS) systems (see Bech and Hobijn 2007). With RTGS, payments
are processed individually, immediately and with finality throughout
the day. Phase 1 of Project Jasper was implemented as a pure RTGS
system with every individual payment on the ledger being prefunded by
DDRs in the participant’s wallet.
RTGS systems eliminate settlement risk at the cost of an increased need for
liquidity. Liquidity demands on RTGS systems can be enormous, given the
large values that are settled in these systems—typically up to one-fifth of a
8 Further information about work on e-money at the Bank of Canada can be found on the Bank’s
e-money page.
4 Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017
country’s gross domestic product on a daily basis. To make RTGS systems
less liquidity-demanding, operators around the world have implemented
LSMs.9
The most effective LSMs are those that support settlement by periodically
matching offsetting payments that have been submitted to a central
payments queue and settling only the net obligations.10 However, offsetting
algorithms cause delay in settlement, which is unacceptable for some types
of payment. Banks therefore need a way to make these time-critical payments.
Phase 2 of Project Jasper explored the possibility of giving banks the
choice of entering payments for immediate settlement or into a queue for
netting and deferred settlement. Project Jasper appears to be the first public
instance of implementing an LSM algorithm on a distributed ledger platform.
Technical Aspects of Project Jasper
The rise of Bitcoin spurred the interest of FMI developers in DLT. Bitcoin
uses a proof-of-work (PoW) protocol that provides decentralized validation
of transactions. PoW protocols are designed to deter a participant
from taking over an open DLT system and double-spending or rewriting
the ledger. This is done by requiring costly work from each node verifying
transactions. This protocol can be very computationally expensive, however,
and requires some level of transparency of all transactions. On the Bitcoin
blockchain, for example, all the identities of participants are masked, but all
transactions are visible to everybody. This expense and transparency stem
from the anonymous and open nature of DLTs such as Bitcoin.
In Phase 1 of Project Jasper, the system was built on the Ethereum platform,
which uses a PoW consensus protocol. The public version of Ethereum is an
unrestricted system that shares a full copy of the ledger with all participants;
Jasper used a version that shared the ledger among R3 members only. In
a closed, private network, like a wholesale payment system, PoW protocols
are neither necessary nor desired. Restricting access to trusted counterparties
enables developers of DLT protocols to use alternative efficient
protocols to perform the validation and recording functions.
The Corda platform on which Phase 2 of Project Jasper is built uses a
notary function instead of PoW. A key feature of Corda is that updates to
the ledger are achieved through two functions: a validation function and
a uniqueness function.11 The validation function, performed by the parties
involved in the transaction, ensures that all details of the transaction are
correct and that the sender has the required funds. The uniqueness function
is performed by a notary. For the Project Jasper system, this is the Bank of
Canada. In this role as notary the Bank has access to the entire ledger so
that it can verify that the funds involved in a transaction are available.
Liquidity-saving mechanisms in Project Jasper
The Jasper LSM is a payment queue with periodic multilateral payment
netting. Conceptually, the way it works is quite simple. If a bank has a nonurgent
payment, the payment can be put in a holding queue. After the bank
9 In the early 1990s approximately 3 per cent of the largest payment systems in the world used liquiditysaving
features; by 2005 this proportion had risen to 32 per cent (Bech, Preisig and Soramäki 2008).
This trend has continued, and nearly all major payment systems now use some form of LSM.
10 The liquidity savings from offsetting algorithms arise from the fact that liquidity is needed only for the
net difference between payments to allow settlement. Suppose Bank A needs to make a payment to
Bank B for a value of $100, and Bank B needs to make a payment to Bank A for a value of $90. The
amount of liquidity required to settle these two payments, if they were entered into a queue operating
an offsetting algorithm, would be $10. In contrast, without an LSM, the liquidity requirement to settle
these two payments would be at least $100.
11 See the non-technical Corda white paper.
Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017 5
submits a notification of the payment to the queue, the submitted payment
waits with other queued payments until the beginning of a matching cycle.
The queue is then locked temporarily while an algorithm combines all the
submitted payments, determines each bank’s net obligations and assesses
each bank’s liquidity position.12
A payment queue is inherently centralized. A key challenge was implementing
it in a DLT system, rather than using traditional account-based
centralized ledger systems. These technical issues introduced significant
complexity and served to highlight the challenges inherent in building
decentralized systems that rely on some level of centralized control or centralized
information.
The innovative solution developed in Project Jasper was the incorporation
of an “inhale/exhale” routine onto the Corda platform. Before the matching
cycle begins, banks may submit payments to the queue. However, these
payments do not immediately go through the two-stage validation and
uniqueness process necessary to add a transaction to the ledger in the
Corda system. Instead, the payment instructions sit in the queue until the
matching cycle begins. At that point, a sequence of events occurs. First,
during the “inhale” phase, a notification is sent to all banks participating in
the matching cycle requesting that they send DDR to the Bank of Canada.
Each of these individual payments is then validated and added to the ledger.
Then, in the “exhale” phase, the matching algorithm determines a subset
of payments to clear, on a net basis, given available funds. The Bank of
Canada sends DDR payments back to all participating banks equal to the
amounts they contributed, plus or minus any money they are owed or owe
following the completion of the matching algorithm.
To illustrate, suppose that only two banks, A and B, place payments to each
other in the queue with values equal to $100 and $90, respectively. In addition,
each bank sent $15 to the queue as part of the inhale phase. After netting the
two payments, the algorithm would charge $10 to Bank A and credit $10 to
Bank B. Given their initial contributions from the inhale phase, this would
mean the exhale phase payments are $5 to Bank A and $25 to Bank B.
These transactions are then validated and added to the ledger. Payments not
matched by the algorithm remain in the queue. At this point a new matching
cycle begins. Banks are free to enter or remove payments from the queue until
the end of the next matching cycle, and the process continues to repeat.
Efficiency and Financial Stability Risks of Project Jasper
The efficiency and financial stability risks of Project Jasper were evaluated
through the lens of the PFMIs that apply to the operation of a wholesale
payment system. Of these, only those relevant to a proof-of-concept system
were considered. Principles that would apply only to a production-level
FMI—such as those relating primarily to governance and legal aspects—
were excluded.13 Thus, the examined principles can be grouped in terms
of the risks they address: credit and liquidity risk, settlement risk and operational
risk.
12 The design is similar to the LSM added to the United Kingdom’s wholesale payment system, the Clearing
House Automated Payment System (CHAPS), in April 2013. In CHAPS, the time between each matching
cycle is two minutes and payments are frozen for 20 seconds during each matching cycle while the
matching algorithm runs. The United Kingdom reports liquidity savings of around 20 per cent (Davey and
Gray 2014).
13 Other legal questions outside of the PFMIs, such as anti-money laundering requirements, were also
excluded.
6 Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017
Credit and liquidity risk
The Jasper platforms were designed without credit risk because all payments
represent a claim on deposits at the central bank—a riskless asset.
Participants transfer cash to the Bank of Canada, through the LVTS, which
then creates DDRs that can be exchanged on the distributed ledger platform.
Overall, nothing in the proof-of-concept design was identified to be
fundamentally incompatible with the credit-risk principle.
As outlined above, Project Jasper incorporated an LSM that imitated the
functionality of existing RTGS systems to mitigate liquidity risk, the risk
that a participant would have insufficient DDRs to make a payment. The
performance of Jasper’s LSM is currently being tested using simulated data.
While it is too early to predict the results of these simulations, we can report
that to date we see no evidence that implementing the LSM on a distributed
ledger would change its use or performance relative to a centralized system.
The LSM would likely generate liquidity savings similar to existing LSMs.
Settlement risk
Settlement is defined as the irrevocable and unconditional transfer of an
asset. Defining the conditions under which settlement is final is foundational
to financial stability.
Two aspects related to settlement finality are relevant to the application
of DLTs like Project Jasper: operational settlement—or the certainty of the
process by which a decentralized ledger is updated—and legal settlement,
which is how settlement finality is defined in relevant system rules and associated
laws.
To ensure legal settlement finality, Project Jasper was structured so that
a transfer of DDR was equivalent to a full and irrevocable transfer of the
underlying claim on central bank deposits. This design feature relates to the
issuance of DDR and is therefore independent of the platforms upon which
Jasper was built.
In contrast, to ensure operational settlement finality, issues related to the
underlying technology of the DLT platforms used would need to be resolved.
In the case of Ethereum, a PoW consensus mechanism is used to validate
payments. But PoW settlement is probabilistic. The payment is therefore
never fully settled because there is always a small probability that the
payment could be reversed. Settlement becomes increasingly certain as
the recorded transaction becomes more immutable over time, but it never
reaches the point of being irrevocable. In the Corda platform, the role of a
trusted notary would, in theory, eliminate this uncertainty because transactions
could not be revoked once completed. However, this system has not
been stress tested, and thus some risk may still be associated with settlement
finality.
Overall, the move from Ethereum to Corda reduces settlement risk and
improves the likelihood that a production system would comply with the
settlement-risk principle. However, a final assessment requires further testing.
Operational risk
Resilience, security and scalability are the core operational risk considerations
in wholesale payment systems. Given that Project Jasper is not a
production-level platform, a detailed assessment of all of these operational
risks was not possible. That said, the focus of Jasper was on resilience and
scalability.
Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017 7
In terms of resilience, a key question was whether a DLT-based wholesale
payment platform could provide more cost-effective resilience by having no
single point of failure. Phase 1 of Project Jasper demonstrated a lower cost
for high availability14 because the nodes operated by all of the participants
essentially served to back each other up insofar as their shared data were
concerned. This guaranteed high availability without extra risk-proofing
of each node. However, once additional functionality, such as an LSM, is
added to the system, the susceptibility to a single point of failure can return.
Resilience must therefore be carefully considered in the implementation
design, for three reasons.
First, additional technology components—such as key, identity and system
access management—are currently based on centralized models and the
assumption of single trusted operators (there are early-stage attempts to
devise distributed versions of these). Thus, these important components
suffer from the same typical challenges associated with a single point of
failure that existing centralized systems face. For example, the digital keys
are bound to individual participants and are used to prove these participants’
right to perform transactions on specific assets; any operator of
a blockchain node needs to have system components to store its digital
keys securely and not share them with others in the network. Thus, system
components that store digital keys should be made highly available to avoid
single point-of-failure risk and backed up for disaster recovery since this
information cannot be recovered from another participant’s node.
Second, the single-point-of-failure comparison of DLT systems with existing
systems can be taken a step further with a notary system, such as Corda.
Unlike PoW, participants’ individual nodes must be operational to send or
receive payments, reducing the resilience of the system. The Corda DLT
platform examined in Project Jasper partitions data so that each participant’s
node has access to and maintains only a subset of that data. While
this approach resolves data privacy issues, it introduces significant challenges
for data replication across the network.15 Unlike public blockchain
schemes, where all nodes have a copy of the exact same database (e.g.,
Jasper Phase 1), these restricted systems have a point of failure at every
node; that is, each node requires data replication and archiving to ensure
business continuity, rather than each node providing resilience to the
system, as in the case with the Ethereum blockchain.
Third, a single point of failure is more likely in a notary system, where nodes
are relatively more specialized than they are in a PoW system. In Phase 2
of Project Jasper, the role of notary in Corda is performed by the Bank of
Canada, so an outage at the Bank would prevent any processing of payments.
This is important because it highlights that operational resilience is
related to the function being performed by each node.
An overall evaluation suggests that, when compared with both centralized
platforms and an open DLT platform, restricted distributed ledger schemes
may decrease operational resilience if they are not carefully designed. This
need for operational resilience may make Jasper Phase 2, based on Corda,
more expensive than the current centralized system in terms of meeting the
PFMIs. In Jasper Phase 2, it is therefore likely that each participant would
have to invest in a high-availability node to reduce the chance of an outage.
14 A payment system is said to be highly available if it operates a very high percentage of the time it is
supposed to, for example, 99.99 per cent of the time.
15 It is important to note that Corda queues pending requests to nodes as part of the design, so that when
a participant with an outage is back, online transactions may still be processed.
8 Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017
Another key aspect of operational risk in the PFMIs is scalability. Currently,
the LVTS processes 32,000 transactions daily, with a peak throughput of
roughly 10 transactions per second. In DLT arrangements, there is a computational
cost to distributing functionality. In PoW platforms like Ethereum,
there is limited capacity to scale. In Phase 1 this was approximately
14 transactions per second because Ethereum was designed for the public
Internet, where speed limitations would challenge information flow between
nodes. While this speed is sufficient to process current daily LVTS volumes,
it could create future peak volume constraints, such as in times of market
stress or volatility. In contrast, scalability would not be a constraint in the
Corda platform because Corda does not have a consensus method based
on a fixed time and requires only nodes of the involved parties and a notary
to verify transactions.
Transparency and Privacy
A fundamental requirement for a wholesale payment system is the need for
participants to keep their transactions private from parties not involved in
the transaction. This is necessary to prevent other participants from being
able to take advantage of this information. A participant’s clients may also
prefer or require this privacy. By implication, PoW systems are ill-suited
for these types of large-value systems because they operate under the
assumption that all transactions in the system are, at a certain level, publicly
observable.
In contrast, notary-based DLT systems, such as Corda, permit increased
privacy because a trusted third party (e.g., the Bank of Canada) helps
validate all transactions. But the lack of transparency in the Corda system
implies that no node in the system, with the possible exception of the
notary, has all the information. Therefore, if the information at one or more
nodes is corrupted, it may not be possible to reconstruct the entire network
since even the notary does not have a full copy of the ledger. This creates
the need for backups of individual nodes and a loss of the economies of
scale associated with centralized systems. Further, it raises the question of
whether the proposed operational-resilience benefits of DLT are possible
under the constraint that transactions remain private.
Conclusion
Project Jasper enabled a better understanding of the roles and responsibilities
of the operator of a DLT wholesale payment system, its participants
and the central bank. In a DLT framework, the operator’s role would likely
be closer to that of a rule maker or standard setter rather than a traditional
IT infrastructure operator. DLT has implications for the roles of operators as
well as for how the PFMIs should be applied or revised. It may be necessary
at some point to update the PFMIs to include principles outlining regulatory
authorities’ requirements for structuring a DLT for a market infrastructure.
In addition, the work on Project Jasper has allowed the stakeholders of the
wholesale payment system to jointly develop the platform. Both private and
public sector partners learned a great deal about the technical aspects of
DLT from the project. They found this improved their mutual recognition of
the complexity of the processes involved and cultivated collaboration to
overcome technical obstacles. It also allowed for a comprehensive comparison
of different DLT technologies from all perspectives (i.e., overseer,
operator and participant).
Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017 9
A pure stand-alone DLT wholesale payment system is unlikely to match the
net benefits of a centralized wholesale payment system. This is because
some parts of a viable wholesale payment system are inherently centralized,
such as the LSM discussed above. This added complexity could lead to
further operational risk when compared with current centralized systems.
Instead, the benefits of a DLT-based wholesale payment system likely lie
in its interaction with the broader FMI ecosystem. Such benefits may be
obtained by integrating other assets on the same ledger as payments—
which could greatly simplify collateral pledging and asset sales—reaping
economies of scope and reducing costs to participants by integrating backoffice
systems.
Cost savings or efficiency gains may also be possible sector-wide. This
could occur if a DLT-based core interbank payment system can serve as
the basis for other DLT systems to improve clearing and settlement across
a range of financial assets. For example, exchange-traded assets already
clear and settle through safe and efficient systems. But gains would be
possible if these systems could be integrated by having cash on the same
ledger as payments to settle the cash leg of each transaction. Over-thecounter
markets (for stocks, bonds and derivatives), syndicated loans and
trade finance are much more decentralized systems with long settlement
times. These could be significantly improved using a DLT-based platform if
they could be integrated with a core wholesale payment system, resulting in
the transfer of cash payments using central bank money.
Distributed ledger platforms offer potential cost savings by lowering the
costs of reconciliation. If a DLT-based system allows banks to validate their
transactions at the very beginning, it could reduce back-office reconciliation
work and potentially achieve major cost savings for the financial sector.
These cost savings depend on the nature of the DLTs: a PoW system like
Ethereum, for example, would be relatively more expensive to operate
because of the computational cost of the consensus mechanism.
Project Jasper has provided valuable insights to all the parties involved.
Several paths could be explored further. One possible future extension
could be to think about how to pledge general collateral instead of cash
collateral to the Bank of Canada. Another would be to explore the potential
integration between Project Jasper and other types of DLTs, either domestically
or internationally. This could help determine potential efficiency
increases from better connections, improved automation of cross-border
payments or the ability to settle multiple assets (e.g., bonds or money
market instruments) on the same ledger.
http://www.bankofcanada.ca/wp-content/uploads/2017/05/fsr-june-2017-chapman.pdf
References
Bech, M. L. and B. Hobijn. 2007. “Technology Diffusion Within Central
Banking: The Case of Real-Time Gross Settlement.” International
Journal of Central Banking 3 (3): 147–181.
Bech, M. L., C. Preisig and K. Soramäki. 2008. “Global Trends in LargeValue
Payments.” Federal Reserve Bank of New York Economic Policy
Review (September): 59–81.
10 Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017
Buterin, V. 2013. “A Next-Generation Smart Contract and Decentralized
Application Platform.” Ethereum White Paper. Available at
https://github.com/ethereum/wiki/wiki/White-Paper.
Committee on Payments and Market Infrastructures (CPMI). 2017.
“Distributed Ledger Technology in Payment, Clearing and Settlement:
An Analytical Framework.” Bank for International Settlements (February).
Committee on Payment and Settlement Systems and Technical Committee
of the International Organization of Securities Commissions (CPSSIOSCO).
- “Principles for Financial Market Infrastructures.” Bank for
International Settlements (16 April).
Davey, N. and D. Gray. 2014. “How Has the Liquidity Saving Mechanism
Reduced Banks’ Intraday Liquidity Costs in CHAPS?” Bank of England
Quarterly Bulletin (Q2): 180–189.
Garratt, R. 2017. “CAD-Coin versus Fedcoin.” R3 Report (5 April).
Nakamoto, S. 2008. “Bitcoin: A Peer-to-Peer Electronic Cash System.”
Available at https://bitcoin.org/bitcoin.pdf.
Narayanan, A., J. Bonneau, E. Felten, A. Miller and S. Goldfeder. 2016.
Bitcoin and Cryptocurrency Technologies: A Comprehensive
Introduction. Princeton: Princeton University Press.
Schindler, J. Forthcoming. “FinTech and Financial Innovation: Drivers and
Depth.” Washington: Board of Governors of the Federal Reserve System.
Project Jasper: Are Distributed Wholesale Payment Systems Feasible Yet? BANK OF CANADA • Financial System Review • June 2017 11