Measuring liquidity risk

The failures in liquidity risk management over the past year have pushed financial institutions and software vendors to reassess their models. What is on offer in terms of liquidity risk systems, how useful are they, and what steps are being taken to improve or develop liquidity risk technology? By Clive Davidson

Regulators have identified many shortcomings while investigating the causes of the credit crisis. Key among the findings is the realisation that many assumptions about liquidity were built on shaky foundations. Financial institutions presumed they would always be able to raise funding and woefully underestimated the amount of capital they might need to satisfy contingent obligations. Firms rarely considered the implications of investing in complex products and, as a result, were left holding on to investments they weren't able to shift.

The response from regulators has been rapid. In an attempt to improve liquidity risk management practices, the Basel Committee on Banking Supervision on June 17 published its revised Principles for Sound Liquidity Risk Management and Supervision. The consultation paper emphasises the need for a qualitative approach to liquidity risk management, with a heavy focus on appropriate processes, management and human judgement. It puts the onus on financial institutions to articulate a liquidity risk tolerance that is appropriate for their business strategy and role within the financial market, and states senior management should be ultimately responsible for maintaining a sufficient liquidity buffer (Risk August 2008, pages 28-31).

However, there is also recognition of the need for quantitative analysis, using appropriate tools and technology. But exactly what tools and technology are needed? As financial institutions wrestle to bring liquidity into their enterprise-wide risk frameworks, third-party vendors are trying to interpret the evolving requirements of liquidity risk management and are adapting their technology to fit.

"Naming and categorisation of risks is a first step to make the risks quantifiable, manageable, transferable and tradable," says Ahmet Inci, consultant at Zurich-based risk systems vendor Iris Integrated Risk Management. In the case of market risk and credit risk, the definitions are clear and the techniques and technologies to manage them are mature - although the recent crisis has led to renewed assertions they are inextricably linked, and that a failure to integrate their management, including the systems that support them, was a major contributor to bank losses.

Clarification

In the revised principles, the Basel Committee clarifies its definition of liquidity risk and identifies two key aspects - funding liquidity and market liquidity. Funding liquidity is defined as "the risk that the firm will not be able to meet efficiently both expected and unexpected current and future cashflow and collateral needs without affecting either daily operations or the financial condition of the firm". Market liquidity risk, meanwhile, is described as "the risk that a firm cannot easily offset or eliminate a position at the market price because of inadequate market depth or market disruption".

Having defined liquidity risk, the next step is to identify where existing market and credit risk tools might be used, and where new technology is needed to manage the idiosyncrasies of liquidity risk. An obvious place to start in identifying the overlaps is the source of the risk.

"Liquidity risk can be said to be a function of credit risk when counterparty cashflows are at stake, and a function of market risk when assets are at stake. But it encompasses many other variable factors, such as the timing and volume of cashflows," says Arun Pingaley, head of product management at Bangalore-based risk system vendor Reveleus.

The sources of credit and market risk are well known and can be modelled in an informed way, using historical data. While liquidity risk is a function of these risks, it becomes more complicated due to the uncertainty over the timing and volumes of the cashflows that are involved, Pingaley adds.

Donald van Deventer, chairman and chief executive of Honolulu-based risk system vendor Kamakura, agrees liquidity risk is inextricably bound with market and credit risk. As such, its management is heavily dependent on being able to properly model these variables. "Banks need to have default probabilities for counterparties that are driven up and down by macro factors - for instance, oil prices, house prices and the point in the business cycles. But they also have to recognise the supply of liabilities to the financial institution also depends on the default probability of the financial institution itself, a fact almost everyone ignores," says van Deventer.

It was this factor - specifically, an inability to raise funding after the wholesale markets ground to a halt and a subsequent withdrawal of depositors' cash - that ultimately undermined mortgage lender Northern Rock in the UK, van Deventer says. "If one can't stress test house prices and see the resulting move in default probabilities, asset values and liability supplies, then one cannot possibly identify house prices as the cause of the problem, nor can one identify a hedge that would mitigate the risk," he says.

For a bank to understand and quantify the risks of its default probability worsening, behavioural models are required that can predict customer responses to news, says Mario Onorato, director, risk solutions at Toronto-based Algorithmics.

"Behavioural models can be crucial, as instruments with no fixed maturity can have a major impact on a bank's liquidity position, depending on customers' reactions to market conditions. For instance, a loss in confidence in the bank can lead depositors to quickly withdraw their money, while a worsening in the economy can cause increased drawing on committed and non-committed rollover facilities granted to customers."

Michael Hall, head of business development at Brussels-based risk system vendor Fermat, believes making behavioural assumptions explicit is critical to properly modelling liquidity risk. With financial institutions increasingly interdependent, a liquidity risk issue seemingly concentrated at one firm can quickly have systemic implications. "Some people in one region of the world not paying their loans may have a huge impact on the entire financial market, even if there is no direct link," he explains. "These indirect links must be available within any liquidity risk management tool in order to model complex scenarios where a relatively small event can have huge systemic impacts. Compared with other risks, the scenario definition is a big part of liquidity risk analysis."

One of the main - and most challenging - characteristics of liquidity risk is its low-frequency, high-impact nature. If market and credit risks are like the wind and rain - always with us but with the potential for extremes in hurricanes and flooding - then liquidity risk is more like an earthquake: rare, sudden, difficult to predict, but likely to have devastating consequences. "As a result, the Basel Committee's new guidelines are strongly focused on the need to cope with the unexpected," notes Onorato. "This has implications both for management processes around liquidity risk, as well as technology."

Historical data shortage

Financial institutions would normally look to develop stress scenarios to test their resilience to a liquidity crisis. However, the rarity of liquidity risk events means there is a shortage of historical data banks can use. As a result, firms are better off asking their most experienced risk managers to dream up imaginative but plausible scenarios, says Philippe Carrel, executive vice-president for business development in the trade and risk management division of Thomson Reuters. At the same time, firms need to consider what data is required in order to monitor the likelihood of these liquidity risk scenarios arising. "The people in the bank have to ask what data they need if they want to truly understand the risk and to have a chance to react and prevent it having a serious impact," Carrel adds.

Having access to comprehensive, complete, correct, timely and up-to-date data is a prerequisite to successful liquidity risk management. However, the data required for liquidity risk management needs to be highly granular - more so than the data used to manage other risks. Carrel gives the example of a bank that invests in a fund of funds, which in turn invests in hedge funds, which invest in collateralised debt obligations, which are based on subprime mortgages in California. "In that situation, the bank's real exposure is to the underlying of the underlying of the underlying," says Carrel. Liquidity risk management requires this level of data granularity.

This has significant technology implications, and means sophisticated and powerful data management tools are fundamental to liquidity risk supervision. The firm must be able to connect to all relevant external sources of data, and amalgamate this information with internal sources in a central data warehouse. Fermat's Hall says it is essential for financial institutions to calculate liquidity risk at the group, as well as individual bank level. Therefore, the data warehouse must be able to cope with group level analysis, as well as the granularity of individual contracts. "One of the most important aspects from a systems perspective is to have a good data mart and up-to-date information about the different exposures generated at any level of the organisation," Hall adds.

In addition to data management tools, a key technological requirement is an engine to generate cashflows for all deals and instruments. "To implement a robust liquidity risk management framework according to Basel II, it is necessary to calculate the expected inflows and outflows of existing and future contracts at the transaction and event level to measure the impact of the actual business and risk management strategy," says Inci of Iris.

Because of the nature of liquidity risk - specifically, its sudden onset and high impact - it is important to be able to perform dynamic simulations of future cashflows under different scenarios. "Dynamic simulations of future business are of particular importance for liquidity risk in comparison with market risk," says Algorithmics' Onorato. "They are crucial for the ability to comprehensively understand the conditions and constraints for funding stability in the future, including an assessment of a bank's resilience under stress conditions."

On top of all that, it is essential to be able to simulate a bank's total cash needs as the macroeconomic environment changes, and to use appropriate analytics to understand the risks that appear, says Kamakura's van Deventer. "Market risk is typically measured with a single period value-at-risk, which is useless from a liquidity risk point of view because it is single period analysis - the ending balance sheet is assumed to be the same as the beginning balance sheet, macro factors are ignored, and probabilities of default are assumed to be zero."

Traditional credit analytics, such as credit-adjusted VAR, typically ignore the liability side of the balance sheet and focus on credit losses, which, as demonstrated by Northern Rock (where funding dried up before losses were realised), can dramatically understate cash needs. "An accurate answer requires a full multi-period simulation where macro factors drive probability of defaults, spreads, values and liability supplies," adds van Deventer.

In other words, liquidity risk management stretches current data management, cashflow generation and scenario simulation technologies to the limit. It is also more demanding in terms of stress testing. Although stress-testing applications usually enable banks to shock conventional risk factors, this is not adequate for liquidity risk. "The recent crisis has taught that liquidity problems can arise even with little or no connection with the risk factors normally considered in stress-testing exercises," says Onorato. "Therefore, it is important to allow users to manage stress testing by individually inputting their own stress assumptions in a variety of ways, including stressing risk factors, behavioural models or new business assumptions."

A new technology that could be of use in the management of liquidity risk is the news reading services that market information providers such as Thomson Reuters and Dow Jones have introduced (Risk February 2008, pages 73-751). These services tag news so machines can read and interpret information. If a bank programs its automated news reader to look out for liquidity risk indicators and sends alerts to its risk systems, it could potentially react quickly enough to take defensive action, says Carrel of Thomson Reuters. "If there was a credit event coming over the news network, the alert could trigger a block on trading limits, so no-one could dump credit on the bank."

More use of data visualisation tools - such as heat maps, where risk information is represented as a colour-coded map with the red end of the spectrum indicating areas of higher risk - might also be useful in spotting an impending liquidity crisis, suggest some vendors. But overall, vendors agree liquidity risk management is less about the development of new tools and technology, and more about the adaptation of systems already in use for market and credit risk and asset-liability management (ALM). However, these systems would need to be able to handle the required data and have the power and flexibility to perform the necessary cashflow projections, scenario simulations and stress tests without restrictions.

None of the risk system vendors interviewed for this article offer a specific liquidity risk system. As Carrel points out, it is not a stand-alone risk. And, anyway, if a vendor did offer a stand-alone liquidity risk system now, it would probably have been in development before the recent crisis, and so would be unlikely to have incorporated the lessons learned from it. Kamakura's van Deventer sums up the way in which liquidity risk is intertwined with other risks: "Our clients would say they are measuring total risk, of which liquidity risk is a symptom of a risk problem in some other area, such as credit, market or ALM."

Liquidity risk management may require some new or upgraded capabilities in risk systems, but it also requires that systems get the basics of market and credit risk and ALM right.