(Update: further refinement of bond haircut, and notes on the subject from The Baseline Scenario and Naked Capitalism blogs)
We often hear options about "temporarily nationalizing" or restructuring banks which lay out a choice between protecting creditors of banks (those who have loaned banks significant sums of money, also called bondholders) or wiping them out. Similar language is about whether to guarantee all bank debt, or only certain kinds. While it's widely said that bank restructuring of a failed bank should leave stock shareholders with nothing (since the bank itself is worth nothing on net), it's sometimes said that we can't wipe out the bondholders since this kind of bank debt is so widely held as an investment by pension funds and many other institutions.
Also, bank investors (creditors, bondholders) need some sense of what the rules are for the future, in order to have confidence enough to lend money to banks.
But just as in much of life, it isn't necessary to choose between 0% and 100% here.
One intermediate form of bondholder protection is to convert the bondholders of failed banks into shareholders in a taxpayer-restructured bank. But, this can reduce the upside for taxpayers, who sometimes will have put in far more money than the bondholders, unless the taxpayers get the great majority of the stock and bondholders relatively little. Even if such a scheme is followed, there remains the critical question of just exactly what percentage of the new stock the bondholders will get. Should they get exactly a proportion of the new stock that their original capital would represent as a percentage of the total money bondholders and taxpayers together put into the banks?
I have one little proposal. When bondholders of banks are given a return of their money, let the bondholders take a "haircut" (partial loss) exactly equal to a fixed multiple of the interest rate they would have gained if the bank hadn't failed.
In other words, if a bank bond has an interest rate of 6% and the multiple for bond haircuts in bank failures is set to 1.5, then when such a bank fails the bondholder takes a haircut of 9% (1.5 * 6%), receiving the remainder of his original capital (91% in our example just now) back during the federal restructuring of the bank.
In a case where debt is converted into new stock, the bondholder receives a share of stock in proportion to a reduced capital amount after a haircut and then in proportion within all capital sources for the bank, including taxpayer bailout money in all forms.
These two methods solve the problem of socializing too much of the losses onto taxpayers and the resulting reverse-robin hood effect of making bondholders perfectly whole at net middle-class taxpayer expense.
In the future, with the possibility of a real partial loss (yet also a limited, predefined risk), bank creditors/investors (bond buyers) would examine banks more carefully, and banks themselves in turn would be managed more prudently in order to attract capital.
Baseline Scenario points out declining confidence in bank bonds, and points out how the bond market is estimating the future of such banks. James Kwak also notes the Fed might try buying such bonds to increase confidence. This all begs the question of whether propping up such banks is a tenable solution. For another view on this see Edward Harrison's guest post on Naked Capitalism, in which he says we need to guarantee bondholders be made whole. But the need is to remove uncertainty, and this can be done by creating a defined haircut that is predictable and known, as I have suggested above.