Friday, May 01, 2009

Networks, finance and ecology

A couple of days ago, the FT reported on a speech by Andy Haldane, the Bank of England's head of financial stability:

Financial risk management and regulation should cast aside many elements of traditional finance theory and learn lessons from ecology, the spread of diseases, biology and engineering, according to a senior Bank of England official...

[Haldane] likened financial mathematical models which "pointed to the stabilising effects of financial network completeness" to the now-discredited 1970s-style ecological orthodoxy that asserted the complex networks of enemies and parasites in rainforests would always guarantee their survival.

What I think Haldane's talking about is the belief that more complex food webs were more stable, because diversity is, like, good. This was challenged by Bob May's famous 1973 finding that if you add links to a food web at random it actually becomes less stable the more complex it gets. SInce then, to quote a 2007 paper "the complexity–stability debate has been a central issue in ecology: does network complexity increase or decrease food-web persistence?"

And yet, to quote me, nature is full of large groups of species interacting in complex ways, and field and lab studies suggest that more complex, diverse ecosystems, in fact, show smaller fluctuations in their population sizes. So how's that work?

What ecology has shown us is that complexity per se isn't a disaster, so long as it's the right sort of complexity. Real food webs aren't random, and highly interconnected webs are in fact more robust to the removal of species - i.e. if one species/company goes extinct/bust, there isn't a cascade of other extinctions/bankruptcies. (This is Jennifer Dunne's work.)

The piece also says Haldane said that "the biggest and most interconnected banks should be subject to tougher regulations than smaller firms because they were most likely to be a super-spreader of financial risk. He likened them to heroin users or promiscuous homosexuals who were most likely to spread the HIV virus."

This is an allusion to the finding that small-world networks, such as the Internet, are robust to the removal of a random link, but very sensitive to the targeted removal of the best-connected links. This might be a bad thing in terms of cybercrime, but if you're trying to, say, stop disease spread, it can be turned to your advantage, because by targetting the best connected/most promiscuous individuals you can have a big impact on spread.

AIG, Citigroup etc., in other words, aren't too big to fail - they're too interconnected to fail.

Food webs are a bit different. In a study of the food webs of fish and plankton species living in lakes in the Adirondacks, Dunne and colleagues found that the species most vulnerable to extinction are the ones that result in the fewest secondary extinctions. This suggests, at least in the absence of human manipulation, that the structure of ecosystems maximizes biodiversity persistence. Researchers are now trying to deduce what shapes ecological networks into these robust configurations-forces such as natural selection or thermodynamic constraints on energy flows within the food web might each be at work.

So complexity per se isn't a bad thing. It's what you do with it that counts. Financially, what we need is to understand what makes a robust network, and then try and build that into the system - although that might not be compatible with everyone getting as rich as possible as quickly as possible. Then, although we won't see the shocks coming, we will be able to reduce their impact.

That's not to say the science is all neatly worked out. Haldane said that financial theory was a generation behind ecology. Which made me want to put my money under the mattress, because, although ecologists are getting some ideas about the general properties of ecological networks, food-web theory is still full of contradictory and disputed ideas. Still, maybe this will see a wave of ecologists following George Sugihara into finance, echoing the wave of maths and physics PhDs who became quants a decade or two ago. Remember how well that turned out?

Haldane's compete speech is here. (This post is based largely on the SFI Bulletin piece linked to above, 'Risk in Financial Markets - Learning from Nature' what I wrote (sound of own trumpet) 18 months ago.)

1 comment:

Unknown said...

thanks for the valuable information