Ripple (amongst others) promises to disrupt correspondent banking by driving out delay, cost, complexity, and uncertainty. Will it deliver on this promise?

Ripple has recently announced Series B funding of $55 million from some pretty prestigious players. After announcing the Ripple Global Settlement Network, they have 15 of the world’s 50 largest banks signed up, and over 30 bank-to-bank pilots completed. Ripple promises a transformation of correspondent banking, and low-cost, instantaneous and secure payments anywhere in the world.

What is the Ripple Global Settlement Network?

Technically speaking, the Ripple network is a public permissioned Distributed Ledger Technology (DLT) network. (One well-known, though not necessarily well-understood, form of DLT is “blockchain”, and a particular application that rides on top of a public permissionless blockchain is bitcoin.)

The Ripple protocol, which underlies the settlement network, has also been adopted in regional bank consortia, and for internal trials, in private networks. The Ripple protocol was designed to be deployed as a replacement and linking infrastructure for national and regional Real-Time Gross Settlement (RTGS systems) which typically are used to clear and settle urgent payments. In an RTGS, settlement occurs typically through the movement of funds within the ledgers of the central bank, meaning that all members of the RTGS must maintain sufficient balances with the central bank to facilitate payments. This in turn creates an intraday market for local currency balances (such as the Fed Funds market). Examples of RTGS include FedWire, CHAPS, and TARGET2. There are over 90 RTGS implementations worldwide.

Other forms of clearing and settlement network exist too, including those that net off bilateral obligations, with settlement of the netted balances occurring on a periodic basis. An example is the bank-owned CHIPS system in the US.

Globally, there is no such settlement capability, because there is no global entity that can maintain currency balances on behalf of the member banks. Ripple does not propose a solution to this challenge, but does offer what should more properly be termed a global clearing network. Settlement still needs to occur through bilateral transfers of funds between banks, even though outstanding obligations are recorded on the Ripple distributed ledger.

Management of correspondent accounts

Currently, with no global settlement authority, banks that transact payments across borders need to maintain Nostro – Vostro (or in the US “Due From” – “Due To”) relationships. One bank opens an account, usually in a foreign currency, with another bank that routinely trades that currency. From the perspective of the account-owner, this is a Nostro (from the Italian for “ours”) or Due-From account. We maintain a record of “our” money maintained on the books of the other bank. The other bank, the account-holder, views the same balance as a Vostro (Italian for “yours”) or Due-To account.

Payments are then cleared by changing the balances of these correspondent accounts. There isn’t settlement in the strict sense of the term, but the debtor bank has to fund the creditor bank through purchase and transfer of the relevant currency. Depending on the credit arrangements between the banks, the account holder may need to fund ahead of the need for a balance to clear payments, or may be allowed to “overdraw” for a short time to allow for payment timing uncertainties.

Several challenges exist:

1. Need to maintain many bilateral relationships in order to make payments in even the most common currencies. There is no global settlement mechanism similar to in-country Real Time Gross Settlement systems or similar. In fact, this is the biggest (although far from only) reason that correspondent banking exists.
2. Reconciliation of Vostro account balances by the account owner can be complex because of payment delays, disputes, fees, and non-payment transactions. It is generally based upon receipt of Nostro account statements or individual transaction confirmations from the account holder, often in the form of SWIFT 940- or 950-series messages.
3. Both banks have regulatory responsibilities, including sanctions screening, and AML transaction monitoring, along with complex associated holding and reporting requirements. Identification of parties to transactions can be particularly challenging, and banks are now expected to identify Ultimate Beneficial Owners (the key human persons controlling complex legal entity structures).
4. Funding of Nostro accounts is expensive, partly because of very high regulatory capital and liquidity requirements under Basel III. As a result, banks are loth to maintain significant balances from which payments can be drawn. This leads to …
5. Counterparty Risk when the account holder is prepared to extend intraday or overnight credit to the account owner in order to facilitate payments in advance of account funding. This is exacerbated by timing mismatches between payment elements of complex transactions. Generally, only the most trusted and well-known banks will be extended credit in this way, but risk of default still exists and must be managed, and reflected in regulatory capital, reporting and stress testing. Note that counterparty risk also exists in reverse – if a bank maintains a balance at a bank that fails, its credit balance may be at risk.
6. FX Risk results from payment delays, particularly in volatile currency situations. Examples would be currencies of countries experiencing unpredictable inflation or political instability. However, as the recent Sterling response to the Brexit decision shows, any currency can be subject to sudden shocks.
7. The cost of payments sent through correspondent banking relationships has a higher proportional impact on lower value and/or less urgent payments. The use of correspondent relationships for individual remittances, small trade payments and the like is a last resort in the absence of anything less expensive. In addition, with the “de-risking” of correspondent banking, pricing is subject to lower levels of competition. Payment costs to the originator and/or beneficiary include wire fees from several parties, SWIFT transaction charges, and FX rate spreads. Typically a wire transaction costs between $10 and $30 depending on the typical payment volumes from the originator. Also FX spreads will be greater for infrequent originators, or for less commonly traded currencies. (The reasons for FX spreads are somewhat complex and outside the scope of this article).

How Well Does Ripple Address These Issues?

In assessing the promise of DLT for global payments, I thought I’d attempt a subjective and personal scoring of the Ripple global settlement network. Scores are given out of a possible 5, with a 5 meaning that the issue is entirely resolved.

Bilateral Relationships – 1/5

In its current iteration, Ripple does not address the complexity of handling many bilateral correspondent relationships at all. However, it does offer the potential for implementation of some kind of cross-border netting mechanism. This would still require some kind of bank-owned central player to act as the clearing house – the DLT will not in itself address this issue. It would, however, greatly ease record-keeping, security, fraud-avoidance, etc.

In addition, until there is broad acceptance of something like the Ripple network among international banks worldwide, banks will find themselves dealing with two kinds of correspondent relationship – those founded on traditional methods (including SWIFT messaging), and those carried out through the Ripple network. SWIFT has over 9,000 member banks, all of whom provide correspondent banking and particularly payments services, so it will take many years to reach critical mass. This could be considered the worst of both worlds, and potentially could make this score temporarily negative. So I think I’ve been kind.

Note that for Ripple to become truly a global settlement network, balances on the distributed ledger would need to be publicly tradeable (perhaps analogously to Bitcoin). This is not currently proposed (at least on the Ripple website).

Reconciliation – 4/5

Since correspondent account balances are maintained in real time, and on a consent-based ledger, reconciliation is in principle extremely easy, reducing the burden on correspondent banking operations groups. However, this only fully works if everything that impacts a Nostro balance is transacted in the DLT. This would have to include fees, interest credits, overdraft fees, funding events, payments that are transacted through different paths (e.g. foreign draft clearing), and non-payment transaction activity. This will take time.

Regulatory Responsibilities – 1/5

The Ripple protocol does nothing directly to help banks address their regulatory requirements around KYC, KYCC, CDD, sanctions screening or AML monitoring. Having said that, there are other DLT-based initiatives that seek to address in particular identity-related challenges. For example, opportunities exist for secure digital identities at all levels, risk-driven AML monitoring processes, and value-added information sharing services. Adoption of some of these approaches as a part of the Ripple ecosystem could be very beneficial.

Funding – 3/5

Payment timing mismatches will be significantly reduced if the majority of cross-border payments become essentially instantaneous and final. However, there will still be some capital and transaction costs associated with funding correspondent accounts (regardless of what ledger is used to track balances), so long as friction remains in FX and capital markets.

Counter-party risk – 2/5

Ripple claims to eliminate counter-party risk. However, in the “global settlement network”, a correspondent banking relationship still exists between the banks of the remitter and the beneficiary. Outstanding balances between the banks can be maintained on the Ripple ledger, providing considerable assistance in reconciliation of positions and non-repudiability of transfers. However, banks will still at times hold debit or credit positions in relation to their counter-parties, and failure of a counter-party could still result in losses. This means that credit limits still need to be maintained, and credit risk analysis is still an essential part of correspondent banking.

FX Risk – 4/5

Most FX risk results from delays in execution of payments: a payment initiated today in one exchange rate environment may not be completed until tomorrow or later, when the exchange rates for a particular pair of currencies may have changed significantly due to an external economic or political shock. In the Ripple model, the day or more delay has been reduced to a few seconds. There is still a small window of risk (which is why I didn’t give this a 5/5) but it will be rare for FX losses to occur.

Cost of Payments – 3/5

The increased efficiencies resulting from DLT based payments will drive down costs, and start to move the world toward affordability of low value cross-border payments such as remittances. This will be especially valuable for developing countries, whose economies and households are very dependent upon payments from emigrants. However, the cost of maintaining correspondent relationships in general will remain, along with the considerable cost of compliance and need to price FX (though the risk portion of this should be significantly lower because of near real-time transaction execution).

Opportunities do exist to reduce FX spread cost, by provision of market transparency on available rates. However, the bilateral nature of settlement will still create friction and associated imperfect competition.

Conclusion

There is tremendous scope for modernizing correspondent banking. Opportunities exist beyond the scope of this article, which has focused purely on international payments.

Distributed Ledger Technology will be a significant enabler over the coming several years. It won’t be as quick as most of us would like, but it will come.

Ripple’s approach is as good as any out there right now, and I look forward to seeing how some of the challenges get worked out, and how additional players will provide extensions to the model that start to push up some of the scores I’ve given above.

Disclaimer: even though I have more than 30 years of experience in wholesale banking, much of which has included correspondent banking and international payments, this is a complex area and I don’t know it all. Some people will have different perspectives, and for sure I’ll have missed some implications. I may also have misunderstood aspects of the Ripple “global settlement network”. Please feel free to add to the value of this article with your comments!