I say this because of one important fact about the crisis. It's that banks' losses were so small. The ten biggest losses suffered by US financial institutions totalled $127.6bn. That compares to $4 trillion of losses incurred by investors in US equities in the first year of the crash in tech stocks in 2000. How, then, can such small losses have had a devastating effect when much bigger ones led to only the mildest of recessions?
The answer lies in network effects. Losses on credit derivatives were concentrated in a few highly interlinked (and highly geared) organizations. That meant that losses incurred by one bank led to counterparty risk for others – the danger they wouldn't get their money back on loans and trades - and a drying up of liquidity, as the Bank of England's Andy Haldane has described. By contrast, losses on tech stocks were spread more evenly across millions of (lower geared) and less inter-connected investors, and so were more easily absorbed.
This difference yields an obvious implication - that one way to prevent future crises is to ensure that banks are less interconnected. John Kay calls for "shorter, simpler, linear chains of intermediation" and "loose coupling that gives every part of the system loss absorption capacity."
But how feasible is this? Some recent research makes me fear not.
First, some parts of the economy are, by their very nature, necessarily more interlinked with other parts. If an oil refinery, for example, were to close it would have more serious effects upon an economy that if an equal-sized biscuit factory were to do so. The MIT's Daron Acemoglu has shown that such interlinkages are important in determining the size of recessions. And Charles Jones at the University of California Berkeley has shown that they can help explain differences in long-run growth too; one reason poor countries are poor is that they are dependent upon some weak links (such as a bad road) that hampers production generally.
And the thing is that banks, by their very nature, are densely connected to the rest of the economy. Even the dullest utility bank will get deposits from across the economy and lend to many industries and so be interlinked. Its failure would therefore be disruptive.
Secondly, William Gornall and Ilya Strebulaev of Stanford University have shown that banks' high borrowing arises from their role in the supply chain. Their argument is as follows. Interest payments are tax deductible, which gives firms an incentive to borrow. However, for various reasons, non-financial firms cannot take as much advantage of this as they'd like. But banks can do so, in part because their ability to diversify lending and the fact that they are senior creditors makes them less risky in normal times than non-financial firms. If they borrow heavily, they can pass on the tax advantages of high debt to their customers, by charging lower rates than they otherwise would.
(You might object that the crisis shows that banks aren't less risky than non-financial firms. You'd be wrong. Remember that around ten per cent of firms close each year; non-financial firms are therefore very risky.)
In this sense, banks' high gearing arises naturally from their position in the supply chain.
All this has a worrying implication. The high gearing, tight networks and interconnections that caused small bank losses to have horrible effects might not be easily fixable flaws, but rather inherent ineliminable features of the economy. Yes, it would be nice if banks were more loosely coupled. But this might not be easily achievable.
You might think that, rather than shifting to looser, more flexible networks it would be easier to manage or regulate banks well so they don't fail in the first place. However, it could be that the bounded knowledge and rationality that cause "bad" management are also inherent, ineliminable features of a complex environment - though that's another story.
Maybe some dangers don't have easy fixes.
