As a graduate student in the early 1990’s I developed a mathematical model to explore the question of what the optimal level of bank deposits and deposit insurance premiums should be. For those whose enthusiasm for differential equations is limited, a key conclusion was that deposit insurance rates should be set counter-cyclically, meaning when the probability of a downturn is high – near the top of an economic boom – rates should be increased, and vice versa. The intuition here is that one does not want to pull capital out of the banking sector during a bust, when it’s most crucial to revive lending. Rather, it’s optimal to build reserves against bank failures when the system is flush and the chance of a bust is most elevated. Note that risk is actually lower post-collapse: Humpty Dumpty may have had a great fall, but afterward, assuming the king’s men know how to make an omelet, it’s all upside (or, in that case, sunny-side up).
At the time, my thesis advisor and I applied this model to the European Union, illustrating difficulties in harmonizing policy if member states differed in risk aversion and economic conditions. (if we failed to convince policymakers of this I suspect the late unpleasantness has done so). But it now appears those who need a lesson in when to raise deposit insurance rates reside not in Brussels but Washington.
Unfortunately, in 1996 Congress capped the FDIC insurance fund at 1.25% of deposits, constraining the FDIC’s ability to raise premiums during economic expansion. With banks able to finance lending through means other than deposit base growth (e.g. bonds, securitization etc.) tying insurance premiums to deposits in a fixed ratio meant the risk level of a bank’s assets could grow faster then either deposits or deposit insurance premiums, a recipe for an undercapitalized FDIC fund.
Now, the insurance fund is down to less than 11 billion and more bank failures loom as the “other shoe” of commercial mortgages gone bad threatens to drop. With newly expanded authority, the FDIC levied a $5.6 billion special assessment in the second quarter and has authorized itself the continued imposition of such fees moving forward. These acts effectively raise deposit insurance premiums during a period of economic weakness, which is precisely the wrong time to do so. Healthy banks are drained of capital needed to revive lending, and, paradoxically, are left more vulnerable to losses. Simultaneously, those supported by TARP, TALF etc. now effectively have those funds prematurely withdrawn, at cross-purposes with the Federal programs’ mission of restoring bank balance sheets.
Bloomberg reports that, to his credit, Rep. Barney Frank, Chair of the House Financial Services Committee said yesterday “This is not the time to raise assessments on the banks,” and FDIC Chairman Sheila Bair is said to be considering having banks pre-pay 2010 premiums in lieu of a surcharge. This is a step in the right direction if the prepayments are counted as assets on bank balance sheets, thus avoiding an immediate reduction in lending capital. But those assets would vanish and capital fall next year, just as bad commercial loans come a cropper and the fund is further strained.
Though it might be politically unpalatable, taxpayer-funded Treasury loans to the FDIC, coupled with borrowing from stronger banks would be less counter-productive alternatives. They’d also be more honest, since troubled banks paying premiums with money lent through federal programs amounts to taxpayer support for bank insurance anyway. Borrowing from the Fed is also an alternative, and while I like many others am loath to grow the central bank’s balance sheet further, so long as the FRB and the FDIC are responsible in seeing to it premiums are raised in recovery years so that the loans are repaid, this expansion need not be inflationary.
If the king’s men want to put Humpty back together again, they need to realize you can’t have your egg and eat it, too.
Tags: bank bailout, counter-cyclical, deposit insurance premium, FDIC, invisble hand, risk aversion, Trester