Project Jasper

An Experiment in Large Value Payments with Distributed Ledgers

Jasper National Park | Source: Wikipedia

Christopher Reid
Bank of Canada
bank-banque-canada.ca
What is Project Jasper?
 An ongoing collaboration initiated by Payments Canada and the Bank of
Canada to explore the possibility of issuing, transferring and settling
central bank-issued assets on a distributed ledger network

bank-banque-canada.ca
What is Project Jasper?
 Proof-of-concept application of DLT to wholesale payments.
– Phase 1: Permissioned Ethereum platform
• Atomic settlement only
– Phase 2: Corda platform (R3)
• Added LSM settlement on DLT

 Objectives:
– Gain insight into interbank payments on DL and how modern payment
features such as an LSM can be introduced.
– Improve awareness of potential risks and how they can be mitigated.
bank-banque-canada.ca
Approach

Pose Design and Conduct

Hypothesis/ build proof analysis
Questions of concept

bank-banque-canada.ca
Functional Lifecycle
Financial Institution

Request DDR against

cash collateral

Redeem DDR Issue DDR to

for cash LSM Atomic FIs wallet

Exchange DDR

Financial Institution B Financial Institution A

bank-banque-canada.ca
Architecture – Ethereum Platform
Designed for the internet and
pseudonymous access
 Data and function replicated
everywhere
 Little confidentiality
All ledgers
 High resilience are
 Any changes are validated and identical
recorded by a consensus of
whole network
 The consensus method (PoW)
means changes become more
permanent over time
bank-banque-canada.ca
Architecture – Phase 1, Proof-of-Concept on Ethereum
 Although the RBC wallet exists on all nodes,
only RBC can invoke the function to spend
from it (only it has the private key)
 Similarly, although all ledgers have the RBC wallet
function to issue new DDR, only Bank of is a smart
Canada can invoke that function contract
 If an FI’s node fails, they should be able to that holds
continue transacting on the network balance
and
 Settlement does not have a definite point- enables
in-time finality credit and
 Everyone on the network an see all trades debit

bank-banque-canada.ca
Architecture – Corda Platform
Designed for Financial Institutions
and identified (permitted) access
 Data only shared and replicated
on nodes of parties to trade Identical
 Good confidentiality
 More costly resilience Identical
 Changes validated only by
parties and trusted notary
 Changes are deterministic Identical
 Some centralised functions
Centralized Notary: Validates and records Doorman: Identity and
functions all transactions it receives authorization service
bank-banque-canada.ca
Architecture – Phase 2, Proof-of-Concept on Corda
 Hybrid: decentralised /
centralised
 FIs do not receive information
on trades they are not party to
 Since data and functions are not
replicated, an FIs node must be
available for it to transact
 Requires ‘traditional’ high
availability and disaster
recovery architecture
 As with current systems,
central function failure =>
system wide failure
 Settlement has definite finality

bank-banque-canada.ca
LSM Netting Functions – Privileged Access in Centralised Systems
The CB has
access to
Ledger of all accounts Ledger of all
FIs FI accounts FI accounts
Before After

Pending CB
CB
Pending
Compute Adjust
net books
Pending
Pending

bank-banque-canada.ca
LSM Netting Functions – A Challenge in DLT

FIs

Pending
CB
X The CB cannot
withdraw
Pending
Compute directly from
CB
net FIs ‘wallets’
Pending Adjust
Pending books

bank-banque-canada.ca
LSM Netting Functions – A Solution in DLT

Financial Financial Settlement

Institution A Institution B agent
Add payment to queue
Add to queue
Add payment to queue
Add to queue
Time

Freeze queue
Send me DDR to cover your payments
Send me DDR to cover your payments
 Send DDR
‘inhale phase’
Send DDR
Time

Run netting
algorithm to
compare net
positions bound
by DDR received
Send net DDR
‘exhale phase’
Send net DDR

Financial Financial Settlement

Institution A Institution B agent

bank-banque-canada.ca
Findings – Against PFMIs
Principle LVTS Phase 1 (Ethereum based) Phase 2 (Corda based)

Credit & Liquidity risk None – all DDR backed 1:1 with cash collateral on LVTS

None (requires further

Settlement Risk None Probabilistic finality
investigation)
Central system
Tolerant to partial failure – Requires engineering for HA
Operational Risk – engineered for HA
rely on network, not individual and DR at central node and
resilience and DR. Minimal
nodes. High redundancy. participant’s nodes
participant work.
Many examples of Ethereum limited to about No inherent limitation.
Operational Risk –
central systems with 12 tx/sec. Ongoing efforts to Bottlenecks likely to be central
scalability
high throughput scale this functions (e.g. notary)
bank-banque-canada.ca
Other Lessons
 There are unlikely to be net benefits using DLT over a
centralised system for the sole purpose of payments
 Benefits more likely to come from enabling efficiencies in
the broader FMIs where potential for cost saving by
simplifying processes (e.g. back office reconciliation)
 The LSM implementation is complex – inherently
centralised functions in decentralised systems is tricky
 Phase 2 achieves both privacy and controlled transparency.
 Platforms and use-cases are still evolving.
Our work is a point in time.
bank-banque-canada.ca
Ongoing Work
 Costs: hard to ascertain overall operational costs including IT costs.
Savings would also need to be costed
 Barriers: assuming a permissioned system, is technology the issue?
Centralised system may be the smallest technology burden for
participants.
 Security: Limited scope for analysis in a PoC since so much of security
depends on the operational model and production infrastructure.
 Modern central systems: How does a DLT based system measure
against a modern centralised, accessible, service-oriented payment
system?

bank-banque-canada.ca
Questions?

bank-banque-canada.ca