Thu Mar 7th, 2013 at 03:53:58 AM EST
I've been thinking about regulatory environments lately -finance, insurance, but also medical or IT.
Actually the last two will be relevant to the idea only as illustration. The proposal is about the financial world.
So, as you may or may not know, norms like Basel (2 or 3 at any rate) or Solvency 2 will require a lesser quantity of capital set aside if you have an internal process to evaluate counterparty risk. This process must be pretty thoroughly documented internally (internally is an operating word here) and will be subjected to audits.
As you also know, audit firms are paid by the company that they are auditing. Thus creating the mother of all conflicts of interest.
front-paged by afew
But how systematically relevant (aka too big to fail) institutions evaluate their risk should be of concern to all. Just like we need to make public (Ben Goldacre in Bad Pharma explains it better and more thoroughly than I could try to do, but the idea should be pretty clear anyway) all about how a pharmaceutical company conducted a trial in order that possible mistakes are detected quickly, and indeed regulation demands that this be the case.
Yes, I know, he also demonstrates vividly that the application of such regulation is appalling, and that's something that needs to be addressed, but the principle stands.
You can probably see where I'm going. Why not demand that institutions that adopt the internal process (in order to reduce their capital requirements) disclose it fully?
They might kick and scream, but at least the enforcement would be easy: if you don't do it, your capital requirements are instantly doubled. Plus, it may actually reduce their costs in the long run -things will not have to be reinvented everywhere. And it would give academics the possibility to conduct large meta-studies that would help us have a much better idea of the likelihood of rare risks.
When trotting out the idea I was told that they (in that case they meant insurers -I was talking to an actuary) spend a lot in that and want to keep it a differentiating factor. Well, you may want to allow for some patenting of truly innovative practices if that's needed. Again, enforcement would be trivial since companies would have to fully disclose what they are doing (nobody, and I mean nobody, would have an internal process and not say it, since it would be a pure waste of money: you wouldn't get the reduction in required capital).
So, what do you think? It's systematically relevant to everyone. Should this data not be more like open-source code than the secret formula they currently are?
(and while you're at it, please enforce the full disclosure of pharmaceutical and agribusiness data).