Dollars and Jens
Wednesday, February 18, 2009
bank solvency, mark-to-market, and capital requirements
All the cool kids are linking to John Hempton's extended discourse on bank solvency. A point a number of people, including me but most notably Arnold Kling, is that
The whole point of regulatory capital is to ensure buffers in case of a really bad downturn. When the really bad downturn happens the buffers will be (naturally) be used. It’s perfectly normal (and in my view acceptable) to have inadequate regulatory midway through a nasty downturn.I think that's better said than it has been said before.
His third definition of solvency is what I was trying to give in my last blog post of this title, and I'd like to see it apply during the crisis, at least to the extent that a bank shouldn't be forcibly shut down unless there's a reasonable likelihood — not necessarily high, but at least 25% or so — that it fails that test. It's hard to measure, though, which is why it's not a good measure for use for regulatory purposes in ordinary times. (The 25% rule wouldn't be bad under normal times; I guess what I mean to say is that the primary bank adequacy measures should be more easily measured, and banks should be put on notice to raise capital on that basis rather than a less-precise standpoint that a private investor would be expected to take.) It also applies, even more strongly, to GE, which I recently saw lumped in with banks because of its financial services involvement; GE has very little tangible equity, and is more highly leveraged on that basis than any of the financials. GE, however, has many operating divisions that are clearly worth more alive than dead; the "tangible" in "tangible equity" imagines that you're forced to sell off not just divisions, but dead capital. If GE is a bank, it is one that would be much better able to sell off operating parts for more than the value of their assets, to a much greater degree than is the case for "other" banks.
This definition of solvency requires that assets be fairly valued. One of the things Hempton notes is that the value of a cash stream depends on your cost of funds, and the market value of an asset may be lower than its value to a bank if the participants in the market have a higher cost of funds than the bank does. There's some value to this point, but it would nice, in his hypothetical, if a bank could at least be asked to mark down from 90 cents to 75 cents. A bank that can't do this and remain solvent under the operating value definition of solvency is playing with creditors' money.*
It's worth noting that Hempton talks repeatedly of the solvency of "the system", as though it were a single, large corporation potentially on the brink of failure. Given all the talk about "nationalizing banks", I rather think that's what the FDIC and various banking regulators do all the time, and I very much hope there's no plan to nationalize all of the banks — at which point I'm not sure what the discussion is about. (Sweden, in the nineties, really only nationalized one bank, though that one was 25% of Sweden's banking — larger than Citigroup and Bank of America combined. I don't think we need a more socialist solution than Sweden.)
Also — I mention this mostly because I linked to the post from four months ago — I'm no longer confident that Citigroup and perhaps even Bank of America are solvent, even under this definition.
*In re TARP as originally billed and now as Geithner seems to plan, note that a bank unwilling to mark down its assets below n% of par value is not going to be willing to sell its assets for less than n% of par value, as that would amount to the same thing. The government has lower cost of funds than hedge funds do, and a public-private pool might be able profitably to pay closer to the 75 cents in the aforementioned example than the market for these assets is currently supporting, but if you're hoping to mop up bad assets that way, you're either going to have to make the banks willing to do it, or you're going to have to shut down the ones that won't. A good solvency measure is thus required for the Troubled Asset plan to work.
Labels: capital requirements