«Research Division Federal Reserve Bank of St. Louis Working Paper Series Understanding the Accumulation of Bank and Thrift Reserves during the U.S. ...»
(2011) use bank-level data to understand whether the doubling of reserves requirements imposed by the Federal Reserve in 1936-37 increased reserves demand and induced a credit contraction, contributing to the deep recession of 1937-38. They ﬁnd, on the contrary, that reserves requirements were not binding on bank reserves demand in 1936 and 1937 and therefore had little impact on credit availability. They thus argue that increases in reserves demand between 1935 and 1937 reﬂected changes in the fundamental determinants of reserve demand. Similarly, Calomiris and Wilson (2004) argue that increasing reserves demand during periods of ﬁnancial upheaval may be due to the liquidity signaling eﬀect that high levels of reserves provide to depositors and creditors. Van Horn (2009), examining the years prior to the Depression era, found that Federal Reserve System non-member banks, which had no access to the lender of last resort, increased their ratio of ER to assets after the ﬁrst banking panic much more than member banks, which could access emergency lending through the Fed. Ennis and Wolman (2012) ﬁnd a similar result for uninsured foreign banks during the 2007-09 crisis.
Japan’s Lost Decade also provides an important modern example of ER accumulation. Japanese banks began a sustained increase in ER accumulation in mid-2001, which peaked in 2003:Q3 when the ratio of actual to required reserves reached 5.9.
Fig. 2 compares the sharp increase in the ratio of actual to required reserves within the Japanese banking system between 2001 and 2006 and the increase in ER during the Great Depression and during the recent U.S. Great Financial Crisis. The recent reserves accumulations of Japan and the United States were much more pronounced, clearly dwarﬁng the accumulation of reserves (in terms of the ratio of excess to required reserves) during the U.S. Great Depression. Ogawa (2007) studies the determinants of bank-level reserves accumulation in Japan during the 1998-2002 period. His results suggest that a strong precautionary motive induced banks with large numbers of bad loans to accumulate relatively more reserves.5 He attributed Japanese banks’ precautionary behavior to the general instability in the Japanese banking system and poor balance sheet health.
We use Fig. 3, 4, and 5 to compare the three historical episodes. In the ﬁgures the excess-to-required reserves ratio (ERR) is plotted against a short-term rate that represents potential alternative investment opportunities to holding cash. Along with these two series, the graphs plots the 24-month rolling correlations between the two, which is predominantly negative, indicating episodes of increasing ERR are those during which this rough measure of the opportunity cost of holding reserves decreases.
Our paper is organized as follows. Section 2 discusses historical reserves accumulation in the United States. In Section 3 we develop the testable model, and Section 4 presents the data and the estimation. Section 5 studies the relationship between the CPP and cash and reserves accumulation. Section 6 concludes.
2. Reserves accumulation: The U.S. experience (2007-10)
2.1 Institutional details In this subsection, we discuss the institutional structure of the U.S. banking system, which provides a basis for our model in Section 3 and the estimation in Section 4.2.
In the United States, depository institutions must hold an amount of funds in reserve (reserves requirement) against speciﬁed deposit liabilities in the form of vault cash or deposits with the regional Federal Reserve Banks. The Federal Reserve Board’s Regulation D speciﬁes the dollar amount of a DI’s reserves requirement through a reserves ratio applied to reservable liabilities (Table 1). Although reservable liabilities consist of net transaction accounts, non-personal time deposits, and eurocurrency liabilities, since December 27, 1990, only net transaction accounts carry a nonzero reserves requirement.6 5 Other work (see, e.g., Uesugi, 2002; Hamilton, 1997; Thornton, 2001) considers the liquidity eﬀect: the proposition that monetary expansion lowers short-term nominal interest rates.
6 The Board of Governors has sole authority over changes in reserves requirements within limits speciﬁed by law. See http://www.federalreserve.gov/monetarypolicy/reservereq.htm.
The reserve ratio depends on the amount of net transaction accounts at the DI.
The Garn-St. Germain Depository Institutions Act of 1982 imposed a zero percent reserve requirement on the ﬁrst $2 million of reservable liabilities. The amount of net transaction accounts subject to a reserve requirement ratio of 3 percent was set at $25 million under the Monetary Control Act of 1980.7 Net transaction accounts over the low-reserves tranche are subject to a 10 percent reserve (see Table 1 for current requirements).
To ensure that DIs can meet their funding needs, eligible DIs can borrow under the primary credit program of the discount window. For example, if a DI experiences operational diﬃculties with its funds management systems, it is at risk of an overnight overdraft, for which it can receive funds through the federal reserve discount window or the interbank market. Funding needs at an individual institution can also arise from circumstances in which aggregate reserves in the banking system are signiﬁcantly lower than what the New York Fed Open Market Desk was anticipating in its management of the federal funds rate target. During the recent ﬁnancial crisis, the signiﬁcant strains in interbank funding markets prompted changes in the terms of discount window borrowing: (i) On August 17, 2007, the Fed extended the maximum term for borrowing to 30 days, renewable at the request of the borrower, and reduced the spread on the federal funds rate target from 100 to 50 basis points. (ii) On March 16, 2008, the Fed further extended the term for borrowing to 90 days and reduced the spread on the federal funds rate target to 25 basis points (see Gilbert et al., 2012).
2.2 The IOR after October 2008
DIs prefer to minimize the amount of ER they hold because neither vault cash nor reserves at the Federal Reserve normally yield interest income. However, on October 9, 2008, Federal Reserve Banks started paying interests on required reserve balances and excess balances. ER jumped, likely in response to this policy change, the intensiﬁcation of the crisis, and the fact that the Fed stopped sterilizing its open market purchases (see Fig. 1). The darkest green section in Fig. 1 shows the increase in bank deposits, representing required and ER, in the Federal Reserve System. The large increase in reserve holdings began in 2008:Q4, reached its peak in 2009, and then remained at a new higher level through 2010:Q4.
In the ﬁrst three months of nonnegative IOR payments, a distinction arose between 7 The exemption amount is adjusted each year according to a formula speciﬁed by the act. The low-reserves tranche
is also adjusted each year.
balances held to fulﬁll reserves requirements (“required reserves balances”) and balances held in excess of required reserves balances and contractual clearing balances (“excess reserves balances”). The rate paid on required reserve balances was 10 basis points below the average federal funds rate target, while the rate paid on excess balances was 75 basis points below the lowest target. The reference window was the maintenance-period federal funds rate. The spreads were subsequently reduced twice before the end of 2008.8 However, these intraquarter changes do not aﬀect our discussion as we study quarterly data.
2.3 IOR and monetary policy
Ceteris paribus, whether banks have an incentive to lend reserves depends on the relationship between the return on alternative investments and the ﬂoor rate. Keister et al. (2008) explain how the payment of IOR can generate a ﬂoor that “divorces” money from monetary policy, as the supply of reserves is then not necessarily tied to the target interest rate, allowing central banks to increase the supply of reserves without driving market rates below target. The use of the IOR as a ﬂoor rate for the relevant policy rate removes the opportunity cost of holding reserves at the central bank. The addition of the IOR as another monetary policy tool has been advocated by Woodford (2000), Goodfriend (2002), and others and has been adopted by many central banks (Bowman et al., 2010) such as the Bank of Canada, the Reserve Bank of New Zealand, and the European Central Bank. The Fed was granted explicit authorization to pay IOR by the Financial Services RegulatoryAct of 2006. The implementation date was originally established as October 1, 2011, but was changed to October 2008 to provide the Fed with an additional monetary policy tool during the U.S. ﬁnancial crisis.
The advantages of IOR have been detailed in the literature (for example, Goodfriend, 2002; Keister et al., 2008, and the references therein). In theory, the short-term interest rate target should be larger than the IOR to allow the central bank to alter the supply of reserves without moving the eﬀective short-term interest rate away from its target. In practice, discrepancies may occur in both normal and crisis times. For example, during most of the U.S. ﬁnancial crisis and continuing into the next few years, the IOR and the eﬀective federal funds rate diﬀered. Bech and Klee (2011) use a marThe ﬁrst reduction brought the spread to zero for required balances and to 35 basis points for excess balances;
both were reduced to zero by the maintenance period ending on November 19, 2008. However, after the December 2008 FOMC meeting, the interest rates on required reserve balances and excess balances were both set at 25 basis points, the upper bound of the newly established target range for the federal funds rate of 0 to 25 basis points. See Bech and Klee (2011) for an excellent discussion.
ket micro-structure approach to explain the conditions under which such a discrepancy may emerge—for example, when some traders (e.g., Fannie Mae and Freddy Mac) that cannot be paid IOR by law are willing to trade federal funds below the federal funds target rate.
The literature on the conduct of monetary policy with an IOR policy is diverse and growing. Martin et al. (2013) derive a simple model to show that aggregate bank lending and aggregate reserves are disconnected when interest is paid on reserves. Hornstein (2010) develops a stylized monetary model and ﬁnds that, although the responses of inﬂation and output to innovations in the target interest rate with an IOR policy are slightly diﬀerent from models in which reserves yield zero interest, such diﬀerences are small. Ashcraft et al. (2011) use data on intraday account balances held by banks at the Fed combined with Fedwire interbank transactions to identify precautionary hoarding of reserves and reluctance to lend during the ﬁrst phases of the U.S. ﬁnancial crisis.
They then use these results to develop a model with credit and liquidity frictions in the interbank market consistent with their evidence on precautionary motives. Below we develop a more stylized model that focuses on the bank’s reserve allocation decision at a lower frequency, consistently with the quarterly data we use.
3. A simple model of excess reserves accumulation
In this section, we develop a simple model of reserve determination that allows us to focus on the factors impacting reserves accumulation that can also be identiﬁed empirically using available bank-level data.
Consider a bank i that faces the problem of allocating a given level of deposits Di between an interest-bearing asset and cash or reserves at the Federal Reserve.9 When the interest rate paid on reserves holdings, rIOR, is zero, any positive differential between the returns on the asset (rA, for example, the yield on 3-month or 1-year Treasury bonds on the secondary market) and the zero-yield reserves induces a proﬁt-maximizing bank to maintain reserves (Ri ) at the minimum required level (δDi, a share δ of deposits). When rIOR is positive, banks may have an incentive to hold ER depending on the relationship between the return on the interest-bearing asset and the IOR (among other factors).
Since deposits can be withdrawn at any time, the bank also faces the risk of large unanticipated withdrawals and, in some cases, of a bank run, if the funds for such 9 We
from the eﬀect of information acquisition on deposit behavior; see Baltensperger and Milde (1976) for
an analysis including this feature.
withdrawals are not available. Although in classical models such as Diamond and Dybvig (1983) bank runs are the result of customers’ withdrawals, the same logic applies in the shadow banking market system and the interbank market. Should this occur, the bank can obtain funds only by paying a penalty rate of rp rA.
A bank facing these scenarios is eﬀectively maximizing expected interest income
subject to a resource constraint based on required reserves:
where δ is the reserve requirement and Li is the (stochastic) deposit withdrawal rate, or reserve losses. The last term of equation (1) is a convex function of Ri and is diﬀerentiable if the random variable and Li have a continuous density f (x). Since the objective function is concave, using the ﬁrst-order condition, the optimal amount of
reserves is determined by the following equation:
rp Pr [Li ≥ Ri ] = (rA − rIOR ) − λ, (3) where λ ≥ 0 is the Lagrange multiplier associated with the required reserves constraint.10 The key trade-oﬀ for a bank is therefore between the expected cost of a liquidity shortage on the left-hand side of equation (3) versus the opportunity cost of holding reserves on the right-hand side.