As Brian noted, I will summarize our current approach. As indicated on page 7, I will discuss depository institutions’ demand for central bank balances, how the Desk manages the supply of balances, and outcomes in the federal funds market. There are three components of demand. Some depository institutions (DIs) hold balances to satisfy the reserve requirements shown on page 8. Many more hold contractual clearing balances, and some routinely hold excess reserves.
Page 9 notes that DIs can meet their reserve requirements by holding currency, by holding balances at a Reserve Bank, and by holding deposits at a correspondent that holds an equal amount at a Reserve Bank. These assets earn no interest, so DIs have an incentive to reduce their required reserves to the level of vault cash and reserve balances they would choose to hold if there were no requirements. In practice, DIs cut required reserves by using sweep programs that shift deposits out of transaction accounts into linked nonreservable accounts. This stratagem works because the Fed applies reserve requirements to end-of-day, post-sweep, deposits. About half of the DIs have no reserve requirement. A large majority of the remainder meet their requirements entirely with vault cash. Only 1,500 DIs hold balances to meet reserve requirements, so last year required reserve balances averaged less than $7 billion, an amount equal to 0.1 percent of total deposits in the U.S. banking system. Economists claim that central banks impose reserve requirements to ensure a sizable demand for central bank balances. Imposing a high required reserve ratio on a narrow deposit base and then allowing DIs to run sweep programs and use vault cash to meet reserve requirements does not achieve that goal.
More than 7,000 DIs have accounts at Reserve Banks. On an average day, they use their accounts to make and receive more than 0.5 million interbank payments with a value of roughly $2.5 trillion. As noted on page 10, many of those DIs need a larger balance to avoid frequent overnight overdrafts than to satisfy reserve requirements. About 5,700 choose to hold a contractual clearing balance. The incentive to maintain a contractual balance is that such balances earn implicit interest that offsets the fees a DI incurs when it uses our priced services. But these “earnings credits” cannot be paid in cash, so DIs that do not use our services have no incentive to hold a contractual balance. Page 11 shows that DIs hold a bit less than $7 billion of contractual clearing balances. The rough equality with required reserve balances is coincidence.
Required and contractual balances, though small, facilitate the implementation of U.S. monetary policy in two ways, as indicated on page 12. First, required and contractual balances are set before the start of each reserve maintenance period, so they establish a predictable lower bound on the period-average demand for balances. Second, DIs are allowed to meet reserve requirements and contractual balance commitments on average over multi-day reserve maintenance periods. This averaging feature helps make the demand for balances interest elastic. DIs reduce the opportunity cost of meeting requirements by holding smaller balances on days when the fed funds rate is above target and larger balances on days when the funds rate is below target. Carryover provisions and clearing bands, which give DIs some flexibility to hold too many or too few balances in one maintenance period and the opposite in the next period, also help make the demand for balances interest elastic.
Page 13 summarizes the third component of demand for balances: desired excess reserves. Large DIs usually aim to hold zero excess reserves on average. Small DIs as a group generally hold positive excess reserves. The left half of the graph on page 14 illustrates that the daily sum of required, contractual, and excess reserve balances averaged about $15 billion from January through July of last year but varied between $10 billion and $25 billion. The day-to-day variation largely reflects big banks’ behavior: Big banks want much larger balances on days when an unusually large volume of payments flows through their Federal Reserve accounts than on other days. This strategy lowers their risk of incurring overnight overdrafts on high-payment- flow days while reducing their opportunity cost of holding non-interest-bearing balances on other days. The right half of the graph shows that demand for balances has oscillated even more widely since August. Even at $25 billion or more, DIs’ balances are not big enough to clear $2.5 trillion of Fedwire payments per day without incurring overdrafts. Indeed, DIs make heavy use of daylight credit. At least in theory, ready availability of inexpensive daylight credit reduces DIs’ demand for central bank balances. A proposed reduction in the cost of collateralized daylight credit—the proposal is now out for public comment—may further reduce demand for balances.
We turn next to the supply of balances, beginning on page 16. You’ve told the Desk to keep the federal funds rate close to target on average. The Desk does so by trying to make each day’s supply of balances equal to the quantity that DIs would demand that day if the funds rate were at target. The Desk’s strategy is to use outright purchases or sales of Treasury securities, plus 14-day and 28-day repurchase agreements, to supply a level of balances somewhat lower than the minimum amount the banking system is likely to demand going forward. The Desk then uses one- to seven-day repo to supply the remainder. The Desk structures its operations so that maturing repo almost always leave the supply of balances short of the quantity demanded and then undertakes a new repo to fill in the gap. But as indicated on page 17, the Desk does not completely control the supply. Unanticipated changes in autonomous factors can make the supply of balances larger or smaller than projected. The staffs at the Board and in New York do a very good job, but not a perfect job, of predicting changes in these factors.
Variations in borrowing from the primary dealer credit facility also affect the supply of balances. For example, Citigroup Global Markets borrowed on 20 of the first 25 business days that the PDCF was in operation, in amounts that varied from $0.5 billion to $2.7 billion. Barclays Capital borrowed on 21 days, in amounts that ranged from $1 billion to $7 billion. Four other primary dealers were frequent borrowers; another six borrowed less often. Changes in PDCF credit are not always captured in the staff’s daily projections, so they can, and do, cause the supply of balances to deviate from the intended level. In any case, if the day’s projected supply is not close to the forecast of quantity demanded, the Desk conducts an open market operation to make the two roughly equal. As noted on page 18, the Desk was in the market almost every business day—indeed, all but six business days—from January 2006 through July 2007, replacing maturing repo with larger or smaller repo as projections showed a need to add or to drain balances.
How well does this approach to implementing monetary policy work? If the key criterion is keeping the funds rate close to target, our current approach works well in normal times and not so well when interbank markets are under stress. But our current approach imposes substantial and unnecessary burdens. As the graph on page 20 indicates, the effective fed funds rate is almost always within a few basis points of target during normal times, as it was from January through July of last year. But there have been many larger-than-usual deviations from target since last August, for two primary reasons. First, daily variations in demand for balances have become larger and more difficult to forecast. Second, demand apparently has become less responsive to temporary deviations of the funds rate from target—that is, demand has become less elastic. Even so, the average of daily deviations from target has been close to zero since mid-September. Pages 21 and 22 summarize equilibrium in the federal funds market and the reasons for large deviations from target. Fed funds typically trade near the target early in the morning because buyers and sellers usually expect the Desk to supply enough balances to get the funds rate to target in the afternoon. But the funds rate sometimes is firm or soft in early trading, signaling a likely shortage or surplus of balances and leading the Desk to aim for a somewhat larger or smaller supply than otherwise.
After the Desk conducts the day’s operation, three things happen concurrently: The supply of balances responds to changes in autonomous factors and PDCF credit; the demand for balances is realized as DIs make and receive payments; and DIs trade federal funds, determining the day’s average or “effective” funds rate. If the supply of balances is close to the actual quantity demanded and if the payment system and the federal funds market work normally, the funds rate will be close to target. Any excess demand or supply generally does not become apparent until late in the day, when the Desk is unable to adjust the supply of balances because primary dealers no longer have uncommitted collateral. DIs that end up with larger balances than they want late in the afternoon seek to sell fed funds. When the banking system as a whole has sizable excess balances, DIs that try to sell fed funds late in the day find few buyers. Because balances earn no interest, the funds rate can fall to zero in that situation. On the other hand, an excess demand for balances makes the funds rate rise relative to target. If the funds rate climbs sufficiently above the primary credit rate, some DIs overcome their reluctance to borrow, raising the supply of balances and helping limit the increase in the funds rate.
Last August provides an interesting case study. The demand for balances rose as DIs sought greater liquidity. The Desk increased the supply. Even so, the funds rate traded firm relative to target almost every morning as European banks bid aggressively for fed funds to lock in dollar funding before the end of their business day in Europe. The firm morning rate suggested a shortage of balances. But late in the day in New York, the funds rate often fell well below target as domestic banks that held larger-than-normal balances during the day tried to sell fed funds rather than hold big non-interest-bearing balances overnight.
While our current regime keeps the funds rate close to target on average, it imposes sizable and unnecessary burdens. As noted on page 23, DIs use real resources to run sweep programs and to carefully manage each day’s balance in their Federal Reserve accounts. While those efforts generate private gains, they are a waste from a social perspective. Even with sweep programs, the opportunity cost of holding unremunerated reserve balances averaged about $360 million per year during the past two years. In addition, the banking system and the Federal Reserve spend many millions of dollars each year to ensure and monitor compliance with complex reserve requirement rules.
As indicated on page 24, our current approach to implementing monetary policy has strengths: It usually keeps the funds rate close to target, and it supports an active interbank market. But our current approach also has shortcomings: It allows occasional large deviations of the funds rate from target even in normal times and more-frequent large deviations when interbank markets are disrupted. Our approach is less than transparent; even well-informed market participants sometimes are surprised by the Desk’s daily operations, and there was widespread misunderstanding of the Desk’s actions last August. Finally, our current approach imposes burdens that simply are not necessary to enable the Desk to keep the funds rate close to target. Theory and foreign experience suggest that it is possible to reduce the shortcomings of the current U.S. approach to implementing monetary policy without sacrificing its strengths. Jim Clouse and Spence Hilton will now discuss a range of options for improving the U.S. approach.