Sir Jon Cunliffe, Deputy Governor for Financial Stability of the Bank of England, argues that the leverage ratio — which ignores risk weighting when calculating the ratio of bank assets to tier 1 capital — is a vital safeguard against banks’ inability to accurately model risk:
….. while the risk-weighted approach has been through wholesale reform, it still depends on mathematical models — and for the largest firms, their own models to determine riskiness. So the risk-weighted approach is itself subject to what in the trade is called “model risk”.
This may sound like some arcane technical curiosity. It is not. It is a fundamental weakness of the risk based approach.
Mathematical modelling is a hugely useful tool. Models are probably the best way we have of forecasting what will happen. But in the end, a model — as the Bank of England economic forecasters will tell you with a wry smile — is only a crude and simplified representation of the real world. Models have to be built and calibrated on past experience.
When events occur that have no clear historical precedent — such as large falls in house prices across US states — models based on past data will struggle to accurately predict what may follow.
In the early days of the crisis, an investment bank CFO is reported to have said, following hitherto unprecedented moves in market prices: “We were seeing things that were 25 standard deviation moves, several days in a row”.
Well, a 25 standard deviation event would not be expected to occur more than once in the history of the universe let alone several days in a row — the lesson was that the models that the bank was using were simply wrong.
And even if it is possible to model credit risk for, say, a bank’s mortgage book, it is much more difficult to model the complex and often obscure relationships between parts of the financial sector — the interconnectedness — that give rise to risk in periods of stress.
Moreover, allowing banks to use their own models to calculate the riskiness of their portfolio for regulatory capital requirements opens the door to the risk of gaming. Deliberately or otherwise, banks opt for less conservative modelling assumptions that lead to less onerous capital requirements. Though the supervisory model review process provides some protection against this risk, in practice, it can be difficult to keep track of what can amount to, for a large international bank, thousands of internal risk models.
The underlying principle of the Basel 3 risk-weighted capital standards — that a bank’s capital should take account of the riskiness of its assets — remains valid. But it is not enough. Concerns about the vulnerability of risk-weights to “model risk” call for an alternative, simpler lens for measuring bank capital adequacy — one that is not reliant on large numbers of models.
This is the rationale behind the so-called “leverage ratio” – a simple unweighted ratio of bank’s equity to a measure of their total un-risk-weighted exposures.
By itself, of course, such a measure would mean banks’ capital was insensitive to risk. For any given level of capital, it would encourage banks to load up on risky assets. But alongside the risk-based approach, as an alternative way of measuring capital adequacy, it guards against model risk. This in turn makes the overall capital adequacy framework more robust.
The leverage ratio is often described as a “backstop” to the “frontstop” of the more complex risk-weighted approach. I have to say that I think this is an unhelpful description. The leverage ratio is not a “safety net” that one hopes or assumes will never be used.
Rather, bank capital adequacy is subject to different types of risks. It needs to be seen through a variety of lenses. Measuring bank capital in relation to the riskiness of assets guards against banks not taking sufficient account of asset risk. Using a leverage ratio guards against the inescapable weaknesses in banks’ ability to model risk.