LSE: "Banking crises are unavoidable"

There is only one way to avoid a banking crisis, according to the co-director of a new London School of Economics research centre - and that's not to have a banking system. By Laurie Carver

The narrow room is a typical academic’s office – a wall of textbooks, a felt-tip scrawled whiteboard, an espresso machine and a leather chaise longue – but it belongs to a man with atypical views. Jon Danielsson is one of two directors of the new Systemic Risk Centre (SRC) at the London School of Economics, launched in January with a £3.76 million grant from the UK government and a mandate to look for risks that could cause the next financial crisis.

That does not mean Danielsson believes crises are avoidable. “Look at the kind of countries that do not suffer banking crises – North Korea, Cuba and India, for example. So there’s really only one way to avoid a crisis – don’t have a banking system. That’s not really an option,” he says.

His conversation is littered with historical examples of the system’s stubborn vulnerability: the freeze of 1866, when not even gilts could be sold; the collapse of wholesale bank Overend & Guerney that caused it, which shows even a partnership model can bring down a financial system; and the liquidity squeeze that followed the outbreak of World War I, and helped give birth to the Swiss financial system.

As a result, Danielsson argues, politicians and regulators should not be aiming to make crises impossible, but to ensure the costs are manageable and borne correctly. To help with that, the SRC is abandoning traditional regulatory wisdom, which tries to ensure the strength of individual banks, and is instead focusing on what Danielsson calls endogenous risk – problems the system’s structure brings on itself.

This involves techniques such as network analysis – looking at exposures within a system and spotting incentives that may drive self-reinforcing behaviour, which is hard enough to do in theory, but calibration involves privileged datasets that regulators and industry are reluctant to share. This may be changing, however.

There’s really only one way to avoid a crisis – don’t have a banking system. That’s not really an option

“Authorities are realising it’s worth having this kind of research, and are beginning to engage. It’s a different kind of model to that used by the Basel Committee on Banking Supervision, looking at the whole system. We try hard to find automatic stabilisers, structures that are resilient,” he says.

Danielsson has long had misgivings about the Basel Committee’s approach to regulation. In 2001, as regulators were drawing up the Basel II framework, which gave banks more freedom to model their own regulatory capital requirements, Danielsson wrote an article called The emperor has no clothes: limits to risk modelling, arguing that the new approach could increase both idiosyncratic and systemic risk, potentially destabilising the financial system.

“I was pilloried,” he says. But that has not changed his view any more than it has the Basel Committee’s – as Danielsson sees it, anyway. “Basel II was shown to be deficient in the crisis, and the committee said: ‘Okay, we got it wrong. Now we’re going to do the same thing, but more intensively.’ It’s a shame they tried to just reapply the same technology, rather than start to think seriously about the system,” he says.

Part of the problem is that real financial risk measurement is impossible, in his view. Models are not useless, but the numbers they provide need to be treated with scepticism, rather than used in what he calls a mechanical way. “I teach this stuff, and I’m not anti-models. But I think the people who really use them understand their limitations, while regulators have a tendency to treat them as a black box. Actually, when models are telling you what you want to see, that’s likely a contrarian signal. Real and perceived risks are negatively correlated,” he says.

Regulatory scepticism towards modelling has increased, of course, resulting in the imposition of standardised approaches, floors and multipliers. But Danielsson says standardisation could be a source of danger, too. “A certain amount of variation is good. What you want in a systemic crisis is for firms not to be positioned the same way – if Barclays is selling, then HSBC should be buying. But if models and capital are all standardised, banks see the world the same way. We want to avoid homogeneity,” he says.