Within our Group, we are looking carefully at ways of reducing the liquidity facilities needed to support our securities settlement activity. This is not due to lack of capacity but more to the potential cost of liquidity for clients.
This is extremely relevant to the T2S environment, where we move from same day batch settlement in commercial bank money for most markets, to overnight batch in central bank money for the bulk of transactions. It is clear that there is regulatory attention around the risks, or rather reliability, of unsecured at bank option facilities. This could give rise to capital cost either in the form of an attribution of a risk asset weighting or by a requirement to gain part, or all, one’s liquidity needs by way of committed facilities.
There are many ways of reducing demand for liquidity. At one level firms can internalise. It is evident that using a single cash account will reduce demand for liquidity. It is evident that any ability to pool or offset across different currency accounts will also reduce demand for liquidity. Reducing intraday limits at client level and accepting slower settlement will do likewise. Improving settlement velocity either by changed algorithms or cycles within settlement systems can have a meaningful impact. In house liquidity provision, especially for prime asset rich entities capable of generating on demand liquidity through intraday repos and other transactions, is an obvious option.
But the process is not a one way benefit. Many of these advantages bring adverse factors. In several, there is a need for segregated account set ups and these eliminate the internalisation benefits of omnibus accounts, within an agent, embracing both sides of a transaction. They may also increase CSD costs if segregation eliminates the scale advantages of omnibus holdings. And, in some markets, especially under T2S, it is not clear how liquidity lines would apply in the batch overnight processes where omnibus structures are used.
In recent meetings, we focused on two core issues. What constitutes liquidity needs? How can they be reduced?
In a settlement life cycle, settlement normally occurs when the settlement counterparties have the appropriate cash and stock. But settlement commitment can occur in some markets when a trade is matched. In a batch omnibus arrangement, the agent bank supplies liquidity at portfolio level having assessed need starting with a worse case, sum of the debits, for credit exposure to the underlying clients. In a designated environment, the bank can establish a credit limit at client specific level. If the bank is merely acting as settlement bank, in T2S as an example, they can again operate a credit limit at client level.
But the credit limit is not the utilisation limit. And, if we are to measure and apply constraints on liquidity, we need to understand the difference and reduce the gap between the two to the maximum extent possible. The credit limit is the maximum exposure that the bank could occur. Liquidity is the maximum exposure that may occur. In theory, one would hope that the level of credit availability and the usage of liquidity are comparable; in practise they are usually far apart. Thus it is important to understand what needs to be assessed. Is it the in house credit limits or actual liquidity exposures?
Let us look at the omnibus account. We can take Austria as an example and base our analysis on known facts within the current environment. We know that the bulk of trades are advised and matched by T+1 and thus the move to T+2 is not an issue. We know that around 80-90% of all Austrian trades currently settle in the first morning run of settlement by the OeKB. If we extrapolate this to T2S, we can expect a similar 80-90% to settle in the batch. UniCredit Bank Austria has an ample pool of liquidity and can make this available for securities settlement as it will substantially be freed up again at start of business for Target payments which occur after the final T2S batch run.
What is harder to assess is what liquidity will be needed for each client. That would require an analysis of flows of each client within the omnibus account and, even then, would have to adjust the result for an impossible judgement. It may be that all settlement occurred in the first batch due to the quantum of liquidity provided by UniCredit Bank Austria. But could it have provided less liquidity and still settled in the overnight, albeit in a later batch? It appears illogical for clients to be deemed to be using liquidity due to actions of UniCredit Bank Austria over which they have no control. In omnibus accounts, liquidity demand should logically crystallise only after the batch run and be assessed on the basis of the start of day open position.
At the segregated or settlement bank account level, the process is much simpler as the limit is applied at client level and is usually known to the client, or can be estimated by them, as can utilisation as long as they focus on the end of batch position. Within the batch the position is opaque. So the question here is whether to apply liquidity on limits, utilisation or end of batch position.
Once we move into RTGS intraday trade by trade processing, the position is clear with liquidity usage being the debit in the cash account. But again, here, and elsewhere, there are other criteria to accommodate. How does auto collateralisation impact the picture? If a transaction is self-funding, is it deemed to be a user of liquidity? I would suggest that the treatment should be identical to repos transactions. For auto-collateralisation is a synthetic repo and reverse repo. There also needs to be clarity in whose book of records such a transaction would fall, especially if it is through a segregated or omnibus securities account.
The market will also have to adapt to any changes in the treatment of liquidity. We need to avoid unintended consequences of changes to the intraday efficiencies of the current process. But we need to ensure that the intraday risks of the process are managed. It is imperative that markets assess how to reduce overall liquidity demand of their settlement processes. At one level, this may require rule changes ranging from the obligations arising from locked in trades to the definition of finality in batch processes. From experience, the latter issue must reflect the ability of a market to reverse the batch as well as its risk preferences. At another level, it has to reflect the need for efficient settlement.
I have always advocated settlement efficiency targets to ensure that individual firms do not manage their flows by settling purchases from sales proceeds only, at start of day, before injecting liquidity for any net residual settlement. A careful analysis of settlement discipline rules is needed if we are to place cost or capital behind settlement liquidity usage. But efficient settlement at lower liquidity can also be achieved by increasing the number of batches run or by the adoption of the circles processing structure used, as an example, in the CREST system. And we could be well advised to look at the agency netting potential of CCPs, both to reduce custodian demand for liquidity by reducing settlement flows and supporting broker needs by enabling client side transactions to be included in CCP runs.
The danger for the market is that cost of liquidity could drive inefficiencies or even costly settlement discipline rules that increase demand for liquidity. The debate needs to become more open and imaginative!
John Gubert
Chairman
Global Securities Services
Executive Committee