A Black Swan In The Money Market

08/01/2008
Featured in print Digest

The authors find no statistical relationship between the LIBOR-OIS interest rate spread and the utilization of the Term Auction Facility.

Following the unexpected leap in interest rates on the overnight loans between banks in early August 2007 and the subsequent turmoil in world money markets - a so-called "black swan" event, something that market participants had not seen before and therefore assumed could not exist - the Federal Reserve took several steps to stabilize the situation. In A Black Swan in the Money Market (NBER Working Paper No. 13943), John Taylor and John Williams demonstrate that increased counterparty risk between banks contributed to the rise in money market interest rate spreads. They find no evidence that the Fed's liquidity measures reduced these spreads.

The Fed made several attempts to improve conditions in money markets and thereby reduce the spread between term inter-bank lending rates, such as the three-month Libor (London Inter-Bank Offer Rates), and the overnight rate. An initial step, lowering the penalty on borrowing at the discount window, and thus bringing the discount rate below the prevailing Libor, failed to encourage banks to borrow from the discount window. Four months after the crisis had begun, the Fed introduced the Term Auction Facility (TAF), which allowed banks to borrow from the Fed without using the discount window.

Because the spread narrowed between December 2007 and the end of February 2008, it appeared that the TAF was working. Soon, however, the spread began widening again, and the renewed stress in the markets necessitated a host of new Fed actions and lending facilities. Since the financial turmoil continues to this day, Taylor and Williams analyze the initial phase of the crisis, specifically the 161 trading days between August 9, 2007 and March 30, 2008. They focus on three key money market interest rates: the target for the federal funds interest rate as set by the federal Open Market Committee, the daily effective overnight federal funds rate in the market, and the interest rate on the three-month Libor.

The "black swan" is perhaps most notable by the fact that from January through August 8, 2007, the standard deviation of difference between effective funds rate and the target was only 3 basis points, but from August 9, 2007 to March 30, 2008 the standard deviation was 20 basis points. Taylor and Williams suggest that the spreads between the three-month Libor and the Fed's overnight federal funds rate target increased dramatically starting in August and fluctuated erratically thereafter. During the year prior to August 9, 2007, the three-month Libor spread above the target federal funds averaged only 11 basis points with a standard deviation of merely a single basis point. Similar non-volatile changes in spreads between term rates and overnight rates were apparent for other Libor maturities and for several other countries.

Among the many explanations put forward for the marked increase in the Libor spread are "counterparty risk" (bearing in mind that inter-bank lending in the Libor market or term, Fed funds is unsecured); "liquidity risk," in which traders at one bank are reluctant to expose their institution's funds during a period in which those funds might be needed to cover the bank's own shortfalls; the notion that banks needed liquidity for purposes of their balance sheets in end-of-year financial reports; and expectations of future interest rate changes, especially declining rates.

To test these risk, liquidity, and expectation theories, Taylor and Williams developed a model that examined the Libor rate, the Overnight Index Swap (OIS) rate, and the accepted bid on the Term Action Facility (TAF). Comparing the behavior of these three activities and plotting their spreads allowed the researchers to separate out patterns of risk, liquidity, and expectations. For comparison and for supporting evidence, they also applied the same adjustment method to calculate the Libor spreads in the Euro, sterling, yen, and other currencies.

The researchers applied simple regressions to daily data to test how risk measures and liquidity measures like the Federal Reserve's Term Auction Facility affected the Libor-OIS spread. Five sets of regressions corresponding to different risk measures all emerge with highly significant results in both one-month and three-month maturity regressions. But, in stark contrast, the authors find no relationship between the TAF and the interest rate spread.

Nor should this be especially unexpected, say Taylor and Williams. They remind us that in assessing the effects of the TAF, it is important to note that it does not increase the amount of total liquidity in the money markets. Any increase in liquidity that comes from banks borrowing from the Fed using the TAF will be offset by open market sales of securities by the Fed to keep the total supply of reserves from plummeting. The actions are essentially automatic in the sense that the Fed must sell securities to keep the federal funds rate on target. For these reasons, Taylor and Williams conclude that counterparty risk is a key factor in explaining the spread between the Libor rate and the OIS rate, and the TAF should not have an effect on the spread.

-- Matt Nesvisky