LIBOR rates (the interest rates at which banks lend money to each other without posting collateral) and the comparable overnight index swap rates (OIS), i.e. the expected future policy rates (the Fed Funds rate in the U.S. and similar rates in the U.K and in the euro area) over the same horizon. The persistence of such a spread is surprising because banks should in principle be able to arbitrage it away–up to the cost of the insurance they need to buy if they want to protect themselves against future fluctuations in policy rates. To arbitrage they simply need to borrow overnight on the money market–for instance at the Federal Funds rate–roll over the funds for three months and use them to lend to another bank at Libor.
Up to the Summer of 2007 the spread between Libor and future expected policy rates was very small, about 10 basis points–which probably reflected the cost of insurance. Since then it has fluctuated between 50 and 100 basis points depending on when and in which market–U.S. dollar, British pound, euro–you consider. On July 15 2008 the spread is 75 basis points in the U.S., 65 in the Euro area. Spreads so large have been unusual even in midst of serious financial crisis, such as in the Fall of 1998. (see Figure 1).
There are two views about why such a spread persists and about the possibility of central banks to reduce it bringing Libor down to the level of the overnight rate. Europeans tend to think that the spread reflects credit risk. (Remember that Libor loans are not collateralized). Since there remains widespread uncertainty about the strength of banks’ balance sheets, Libor loans are risky and the spread simply reflects the market assessment of such risk. Assuming that this is the reason, European central bankers think that it would be inappropriate for them to try to eliminate a ”market price”. Thus we should live with it. It has been suggested that this explains why the ECB and the BoE have very reluctantly followed the Fed in announcing the swap lines created (in early May) among the three central banks to make it possible for Euro area and U.K. banks to borrow overnight dollars–and symmetrically for U.S banks to borrow pounds and euros. The Fed proposed these swaps–which are in effect credit lines–to try and bring the spread down; the Europeans thought that this step was inappropriate, or in any case useless
We can assess the ”European” view computing the market’s assessment of the probability that a loan may not be reimbursed, i.e. that the bank fails and forgoes completely its obligations –admittedly an extreme case, since a fraction of the loans is typically repaid. In the Euro area, with (annualized) overnight rates at 4%, a 50bp spread implies (assuming risk neutrality, which may not be right either) a default probability of about 5% over a three-month loan. When the spread rises to 80 bp (as in December ’07) the default probability is 7,5%. In the U.S., with the Fed Funds at 2%, a 50 bp spread implies a default probability of about 10%. In both cases not small numbers.
A problem with the ”European” view is that if Libor reflected the creditworthiness of banks, we should observe spreads to vary across banks depending on the perceived state of their balance sheets. This does not appear to the the case. Figure 2 (reproduced from the slides that accompany a speech given in early May by New York Fed’s economist William Dudley, available at www.newyorkfed.org/newsevents/speeches/2008/dud080515.html) shows that the range of Libor rates for 16 reporting banks is rather small: 10 basis points, hardly a reflection of a market characterized by widespread credit concerns.
The Fed seems to hold a different view, which starts from the presumption that after the Bear Stearns episode it is very unlikely that a U.S bank would be allowed to fail–and that even if it did the Fed would intervene protecting bondholders (including the banks which have lent to the failed institution at Libor) and shifting the loss entirely onto shareholders. The Fed suggests instead that what underlies the spread is not credit risk, but rather a ”shortage of bank capital”.
Consider a bank that has enough capital: it can borrow and make a new loan without with the capital it has, without going beyond its target level of leverage. In other words, the shadow price of its capital is zero. Such a bank will arbitrage between Libor (rLibor) and the expected cost of rolling over overnight funds and insuring against fluctuations in the overnight rate
Consider instead a bank that, in order to make a new loan, must raise new capital, or reduce the capital it has allocated to other activities. For such a bank the spread between the lending rate and the cost of borrowing must equal the shadow price of capital
For instance, with a capital requirement, under current Basel rules, of 1,6% and a shadow price of capital of 20% (what many banks are promising to attract new investors), the spread between Libor and the overnigh rate (net of the insurance premium) is 32 basis points. Before the crisis the insurance premium was abround 10 basis points: it may have risen considering the increase in volatility. This gives an overall spread of 42 basis and possibly more depending on the current level of the premium. It thus explains some of the divergence from historial levels, but is far from the peaks observed during the crisis.
This view has an additional problem. It requires that all banks are capital constrained. This is unlikely to be true, and just a few unconstrained banks could arbitrage away the spread. For this to happen, however, the unconstrained banks should be large relative to the market. Otherwise, as they lend at Libor, they will also eventually hit a capital constraint.
The bottom line is that both the European and the Fed’s view have problems. An interesting alternative explanation has been suggested by MIT’s Ricardo Caballero. Banks could be engaging in ”predatory behavior”. Banks that have ”free” capital might be tempted to behave strategically and refrain from lending to banks which instead need the funds to overcome a liquidity crisis. Here is how the argument goes.
Since we cannot assume that the Fed will bail out all banks in trouble, it is possible that a liquidity crisis might result in a bank failing. The experience of Bear Stearns then suggests that faced with a possible failure the Fed protects bondholders, but wipes out shareholders. This means–as in the Bear Stearns– JPMorgan case—that the bank with ”free” capital can acquire a competitor to which it has denied a loan at a price close to zero. Predatory behavior can explain the persistence of the spread even in the presence of a few large banks that are not capital constrained. (Caballero has recently expanded this view in an article in the Financial Times)
Finally let’s come to central banks and what they can do to reduce the spread. The simplest option would consist in eliminating the need for banks to borrow at Libor from other banks and provide them the funds they need directly through the Term Auction Facility (TAF). The facility (created earlier this year) allows a bank to borrow reserves from the Fed posting assets (of any quality) as collateral. Currently the TAF is a 28-days facility. To pursue this route and make it equivalent to borrowing at Libor, the Fed might need to extend the horizon of the TAF from 28 to 90 days. (The ECB and the Bank of England have similar facilities.) One remaining difference is the need to post collateral, which is a requirent to access the TAF, while no collateral is needed in the case of interbank borrowing. But since the Fed accepts almost any asset as collateral, this would not be a serious constraint.
There remains an underlying problem. How credible is the Fed’s committment to supply funds to banks through the TAF–even in the case of a liquidity crisis that might bring a bank down–without affecting the monetary base? This question has often been raised in recent months as markets worried about what might happen when the Fed ”runs out of TBills”.
There are three ways in which the Fed could expand its balance sheet– without affecting the outstanding stock of base money: (i) if it were allowed to issue its own Bills (as some other central banks do, e.g. the People’s Bank of China), (ii) if it could induce banks to hold more reserves for any given level of the moneary base, (iii) if the Treasury were to issue more Bills than it needs to finance the deficit, and the Fed bought them back from the market. Option (iii) is straightforward: however it is not unlimited, unless Congress raises the yearly limit of Treasury issues. Options (i) and (ii) are related but only (i) would put the Fed in the position of issuing an unlimited amount of Bills to buy banks’ paper. To compare options (i) and (ii) it is useful to consider their implications on the Fed’s balance sheet
It is unclear whether today the Fed has the authority to issue its own Bills, and the Fed hopes that the law currently discussed in Congress that would allow it to pay (starting in October 2011) interest on reserves might contain a line which explicitly authorizes the Fed to issue its own Bills.
Options (i) and (ii) are obviously very different. If the Fed could issue its own paper the ability to expand its balance sheet would be unlimited–and its commitment to bail out any bank without affecting M would become fully credible. Inducing banks to increase their reserves is much less powerful. If we compare the balance sheet of the Fed with that of the ECB (which pays interest on reserves) we see that paying interest on reserves could add some U.S.$ 100 bl to the liabilities side: not small, but not unlimited either. Moreover, banks can certainly be induced to hold a higher volume of reserves if these are remunerated, but in order to do so—at an unchanged level of base money–banks must liquidate other assets. This would happen, but only over time.
Thus the pressure that the Chairman of the Fed is putting on Congress to bring back the date in which it could start paying interest on reserves, might be motivated by the hope that the law will contain an explicit authorization to issue ”Fed Bills”, more than by the effect it will have on the stock of reserves. Summing up. The two competing views about why a Libor-OIS spread persists–credit risk or a shortage of bank capital–both have problems. The possibility that banks with still abundant capital might be engaging in predatory behavior seems a better explanation for the persistence of the spread. Central banks, and the Fed in particular, would in principle have the power to get rid of the spread: they could simply eliminate the need for banks to borrow at Libor from other banks and provide them the funds they need directly, e.g. through the Term Auction Facility (TAF). This leaves us with a final basic question: can central banks credibly committ to provide unlimited high quality paper to banks without affecting the monetary base? In the U.S. case such a credible committment would require Congress to explicitly authorize the Fed to issue its own bills.