Tuesday, October 21, 2008

From transparency to intelligibility in regulating finance

As I've been reading coverage of the rolling financial crisis, it occurred to me again that one of the tools that I’ve recommended for managing complex regulatory systems - transparency - might be relevant, though with a tweak.

More than lack of transparency, a root cause of the melt-down seems to have been a lack of intelligibility. The constant refrain is that nobody understood the ramifications of the financial instruments they were creating, managing or (not) regulating.

I’ve explored this topic in the context of software development on this blog (the hard intangibles thread), and I'm now convinced that the cognitive constraints that lead to problems on large software projects apply in finance, too. As I blogged last August:
“The sub-prime mortgage debacle is a problem of cognitive complexity. A lack of understanding of the risks entailed by deeply nested loan relationships is leading to a lack of trust in the markets, and this uncertainty is leading to a sell-off.”
At that point I hesitated to draw the corollary: that limits should be imposed on the complexity of the intangible structures we create.

The current approach is that complex novel approaches are left unregulated, on the assumption that only “informed investors”, those who are supposedly smart enough to understand the risks, will be exposed to losses. We’ve learned that this is not the case: the informed investors are pretty dumb, and the rest of us pay for their ignorance.

I now think we should invert the regulatory presumption: the more complicated an instrument, the more firmly it should be supervised.

The hard question is how to measure the intelligibility of financial instruments in order to decide if they deserve additional scrutiny. The Mom Test for user interface design - "would your mom be able to figure this out?" – seems reasonable, but it’s hard to see how the SEC would use it in practice. A more commonly used equivalent, the Politician Test, doesn’t help either since the comprehension of politicians is a function of campaign contributions.

We’re left with algorithmic complexity: the length of the program required to specify the object. Financial wizards will surely plead commercial confidentiality in order to avoid disclosing their algorithms; but a private assessment by an impartial regulator need not lead to a leakage of competitive advantage.

1 comment:

Anonymous said...

I have been reading Terry Pratchett's "Making Money" for comic relief:

"'I read somewhere that the coins represent a promise to hand over a dollar's worth of gold...'
'...In theory, yes' he said after a few moments. 'I would prefer to say that it is a tacit understanding that we *will* honor our promise to exchange it for a dollar's worth of gold, provided we are not, in point of fact, asked to.'
'So it's really not a promise?'
'It certainly is, sir, in financial circles. It is, you see, about trust.'
'You mean, trust us, we've got a big expensive building...'"