Liquidity stress testing is the process by which a bank or other financial institution estimates what would happen to its cash inflows, outflows and stock of liquid assets under adverse conditions. It asks whether, and for how long, the institution could continue to meet its payment obligations if depositors withdrew funds, lenders refused to roll over borrowing and assets became harder to sell. The results determine how large a liquidity buffer the institution needs.
It is a central part of liquidity risk management and a regulatory requirement for banks in all major jurisdictions. The failures of 2007 and 2008 showed that institutions meeting capital requirements could still collapse through loss of funding, and the Basel Committee's 2008 principles for sound liquidity risk management made regular stress testing a core expectation. The liquidity coverage ratio is in effect a standardised stress test set by regulators; internal testing goes beyond it.
Scenarios
Institutions are expected to run several types of scenario.
- Idiosyncratic stress. An event specific to the firm, such as a large loss, a rating downgrade, a fraud or adverse publicity, causes customers and counterparties to withdraw.
- Market-wide stress. Funding markets seize up, asset prices fall and haircuts rise for all participants, while the firm's own standing is unchanged.
- Combined stress. Both occur together, which is normally the most severe case and the one against which buffers are sized.
Scenarios are run over several horizons, commonly from overnight to a year, and at different levels of severity. Reverse stress testing starts from the outcome, the point at which the firm could no longer pay, and works back to identify what combination of events would cause it.
Key assumptions
A stress test is a set of assumptions applied to the cash flow projection used in liquidity gap analysis.
- Deposit outflows. A deposit run-off rate is assigned to each category, with insured retail balances treated as most stable and large uninsured deposits and deposits from financial firms as least stable.
- Wholesale funding. Unsecured wholesale funding is assumed not to roll over, and secured funding to roll only against the best collateral.
- Contingent demands. Customers draw on committed credit lines; derivative positions require more collateral; a downgrade triggers contractual payments.
- Asset liquidity. Securities in the buffer are sold or pledged at stressed haircuts, and less liquid assets are assumed to generate no cash within the horizon.
- Inflows. Loan repayments are reduced, since the firm may need to keep lending to protect its franchise.
- Transferability. Limits on moving funds between entities, countries and currencies are recognised.
The output is a projection of the net cash position over time, compared with available liquid assets. The survival horizon is the point at which the buffer is exhausted.
Use of results
Boards set a risk appetite for liquidity in terms of stress results, for example survival for a minimum number of days under the combined scenario. Results drive limits, the size and composition of the buffer, the pricing of liquidity charged to business lines and the design of the contingency funding plan, which sets out the actions to be taken if stress indicators are breached. They also inform recovery planning. Intraday needs are addressed through intraday liquidity monitoring, since a firm can fail to make a payment during the day even with adequate end-of-day balances.
Regulatory requirements
In the European Union and the United Kingdom, banks document their stress testing in an internal liquidity adequacy assessment, which supervisors examine in the supervisory review and evaluation process and which may lead to additional liquidity requirements. In the United States, large banking organisations must run internal liquidity stress tests over set horizons, including overnight, 30 days, 90 days and one year, and hold a buffer sufficient for the 30-day result. Supervisors and central banks also run their own system-wide exercises.
Lessons and limitations
Stress tests rely on historical behaviour, and recent events have shown how quickly that can change. In March 2023 customers of Silicon Valley Bank sought to withdraw about 42 billion US dollars in a single day, a speed far beyond standard regulatory outflow assumptions, made possible by digital banking, social media and a concentrated depositor base. Authorities have since reviewed run-off assumptions, the treatment of uninsured and concentrated deposits, and banks' readiness to borrow from central bank facilities.
Other weaknesses are common. Assumptions about selling or pledging securities may ignore unrealised losses that make sale costly, or operational obstacles to using collateral. Scenarios may be too mild, or too similar to the regulatory ratio to add information. Results depend on data from many systems, which can be slow to assemble in a real crisis. Liquidity and solvency interact, since doubts about capital trigger withdrawals, yet they are often tested separately. For these reasons supervisors treat stress testing as one input among several, alongside limits on funding concentration and qualitative review of governance.