St. Louis Adjusted Monetary Base Series
November 18, 1996
The new adjusted monetary base, adjusted total reserves and adjusted nonborrowed reserves time series are chain indexes created in segments. Major changes in the structure of reserve requirements demarcate the beginning and end of segments, and are the reason why these are calculated as chain indexes. The reserve adjustment magnitude, or RAM, "adjusts" the monetary base for changes in the demand for base money due to changes in statutory reserve requirement ratios within a given structure of reserve requirements (where the structure defines the types of deposits that are reservable, perhaps by class or type of depository institution), conditional on an assumed model of depository institutions' demand for base money; see Burger and Rasche (1977) and Anderson and Rasche (1996a). When there is a major change in the structure of reserve requirements - such as the extension of reserve requirements to nonmember banks and thrifts under the Monetary Control Act - carrying the same RAM across the break is inappropriate. Rather, the old RAM should end and a new RAM start. During periods both before and after the break, for example, the AMB equals the sum of the monetary source base and a RAM - but not the same RAM before and after the break.
Adjusted Monetary Base
The adjusted monetary base equals the sum of the monetary source base and an appropriate RAM adjustment. The new AMB is constructed in four segments: January 1936 - December 1972, December 1972 - January 1975, January 1975 - October 1980, and October 1980 to date. The overall AMB for 1936 - date is created by splicing the individual segments at the overlapping months: December 1972, January 1975, and October 1980.
The new adjusted monetary base includes four RAMs: RAM1 (1936-72), RAM2 (1972-75), RAM3 (1975-80) and RAM4 (1980-date). Each RAM has a different base period, corresponding to a particular structure of reserve requirements. The systems of statutory reserve requirements that correspond to RAM1, RAM2 and RAM3 are discussed in Tatom (1980). RAM4 is based on the set of requirements in effect during the reserve maintenance period ending January 7, 1991; see Anderson and Rasche (1996a).
Adjacent pairs of RAM are measurable both before and after each splice date - in December 1972, January 1975, and October 1980 -- permitting chaining the overall AMB index. For the most recent subinterval, from November 1980 - date, the AMB equals the sum of the monetary base and RAM4 (1980-96). For the period January 1975 - October 1980, the AMB equals the sum of the monetary base and RAM3 (1975-80), multiplied by the ratio of the adjusted monetary base in October 1980 including RAM4 to the adjusted monetary base in October 1980 including RAM3. For example, the AMB for July 1980 is:
AMB (July 1980)
= [Monetary Source Base (July 1980) + RAM3 (July 1980)]
The first term is the AMB for July 1980 as it would be calculated in July 1980; the second term "splices" values of the AMB for periods before October 1980 to later values. For 1972-75, the AMB is multiplied by two ratios. The AMB for July 1974, for example, equals:
AMB (July 1974)
= [Monetary Source Base (July 1974) + RAM2 (July 1974)]
The first ratio chains the 1972-75 data to the 1975-80 data, and the second chains the 1975-80 data to the 1980-96 data.
For dates through October 1980, RAM1 (1936-72), RAM2 (1972-75), and RAM3 (1975-80) are due to Tatom (1980); see his Appendix 1, Table 1. We use only the RAM adjustment, not the published adjusted monetary base or adjusted reserves data shown in Tatom (1980); a different monetary source base, discussed in the next section, is used to construct the new adjusted monetary base. Tatom's three RAM adjustments are based, respectively, on the structures of reserve requirements in effect during 1935, 1972 and 1975. The new RAM4 (1980-96) for dates beginning November 1980 is calculated by the method described in Anderson and Rasche (1996a) from deposit and reserves data submitted weekly since 1980 to the Federal Reserve by about 12,000 depository institutions. For each week (through January 1984), or biweekly reserve maintenance period (beginning February 1984), the calculation proceeds in three steps:
For January 1936 - December 1958, the new AMB includes the currently-published (on FRED®) St. Louis monetary source base. Beginning January 1959, the new AMB includes a revised St. Louis monetary source base equal to the sum of three variables: currency in circulation outside Federal Reserve Banks and the Treasury; deposits of domestic depository institutions at Federal Reserve Banks; and float-pricing related as-of adjustments. This measure of the monetary source base corresponds to line 8 of Table 1 in Anderson and Rasche (1996a, 1996b). All data are obtained from the Division of Monetary Affairs at the Federal Reserve Board of Governors, and are published on the Board's weekly H.4.1 statistical release. The second item, Federal Reserve deposits, equals the sum of two items published on the H.4.1: reserve balances and required clearing balance contracts, the latter shown in a footnote on the first page of the release. The third item, float-pricing related adjustments, is a small item mandated by the Monetary Control Act's requirement that the Federal Reserve recover from depository institutions the value of float generated in check processing; it is included in "service-related adjustments" in a footnote on the first page of the H.4.1 release. (Note that the aggregate amount of reserve balances shown on the H.4.1 is defined by Board staff as equal to: the aggregate amount of depository institutions' Federal Reserve deposits, minus the aggregate amount of required clearing balance contracts and service-related adjustments, minus other small unpublished accounting adjustments.)
The new monthly AMB is seasonally adjusted with a sliding window X11-ARIMA procedure. First, an ARIMA model is used to forecast the not seasonally adjusted AMB two years beyond the end of last full year of data (at the time of this writing, 1995), through December 1997. Then, beginning in 1950 (fluctuations in earlier data seem too unstable to reasonably estimate a seasonal component), the standard Bureau of the Census X11 filter is applied sequentially to a window of eight years of data, the final window spanning January 1990 through December 1997. This method permits more time variation in the estimated seasonal factors than would be obtained by applying X11 directly to longer spans of data. Such flexibility seems desirable for the monetary base because a time-series plot of its monthly growth rates suggests a sharp decrease in its seasonal amplitude after 1990, perhaps due to heavy exports of currency.
Seasonal adjustment factors for biweekly (reserve maintenance period) data are obtained by a
ratio-of-moving-average procedure. In this method, a set of initial estimates of biweekly
seasonally-adjusted levels of the adjusted base is obtained via polynomial interpolation between
observations on seasonally-adjusted monthly levels. An initial set of seasonal adjustment factors
are obtained by dividing actual not-seasonally-adjusted biweekly levels by these initial estimated
seasonally-adjusted levels. This process is iterated to convergence, subject to the restriction
that the final seasonally-adjusted biweekly levels average to the given seasonally-adjusted
monthly levels.
Issues in Defining "Total Reserves" of Depository Institutions
There are two major, alternative economic measurements of "total reserves."
A. A Definition Motivated by Statutory Reserve Requirements
This narrow definition is "eligible assets of depository institutions subject to Federal Reserve reserve requirements." "Eligible" here refers to assets that may be used to satisfy statutory reserve requirements against deposits, not be confused with the concept of assets eligible to be used as collateral for discount window loans.
This definition of total reserves focuses on satisfying statutory reserve requirements and a depository's business needs for vault cash and Federal Reserve deposits are not explicitly considered. Rather, it is assumed that the amounts of vault cash and reserve balances held to satisfy statutory requirements are sufficient to satisfy the depository's payment needs.
Under this definition, total reserves is measures as:
These measures of total reserves are constructed by staff of the Division of Monetary Affairs
at the Federal Reserve Board and published on the Board's weekly H.3 statistical release. Details
of their construction and adjustment for changes in reserve requirements are available from the
Division of Monetary Affairs. Published sources for these data include:
|
|
Total Reserves, adjusted for changes in reserve requirements and seasonally adjusted |
|
Total Reserves, adjusted for changes in reserve requirements, not seasonally adjusted |
|
Total Reserves, not adjusted for changes in reserve requirements, not seasonally adjusted |
|
Nonborrowed reserves, adjusted for changes in reserve requirements, seasonally adjusted |
|
Excess reserves, not adjusted for changes in reserve requirements, not seasonally adjusted |
|
Free reserves, adjusted for changes in reserve requirements, not seasonally adjusted |
|
Total discount window borrowing, not seasonally adjusted |
|
Extended credit discount window borrowing, not seasonally adjusted |
|
Adjustment plus seasonal discount window borrowing, not seasonally adjusted |
|
Data Sources: Federal Reserve Board, Aggregate Reserves of Depository Institutions and the Monetary Base (H.3), weekly statistical release, and Table 1.20 in the Federal Reserve Bulletin, monthly. Annual historical data are released by the Board's Division of Monetary Affairs as Reserves of Depository Institutions (H.3 historical package), mimeo. |
B. A Definition Motivated by Depository Institutions as Financial Intermediaries and Sellers of Payments Services
One implication of the results reported by Anderson and Rasche is that broad measures of reserves may be important to modeling the role of depository institutions in the economy. In addition to satisfying statutory reserve requirements, depository institutions must hold sufficient vault cash and Federal Reserve deposits to convert retail customer deposits into currency and to make interbank payments, on request. The measure of total reserves published by the Federal Reserve Bank of St. Louis equals the monetary source base minus currency held by the nonbank public. Measured in this way, total reserves includes all base money held by depository institutions, including the vault cash held by nonmember banks and thrifts prior to the Monetary Control Act and, more recently, the Federal Reserve deposits held by depositories to satisfy required clearing balance contracts.
Although reserve requirements have varied significantly since the founding of the Federal Reserve, it seems inappropriate to analyze the behavior of depository institutions solely from the narrow viewpoint of cash assets eligible to satisfy statutory reserve requirements. Consider, for example, the exclusion of vault cash at nonmember depository institutions from the narrow measure of total reserves discussed in section A (above) for dates prior to the Monetary Control Act. The monetary aggregates (M1, M2 and M3) were extended in 1980 to include deposits at nonmember institutions, but the narrow definition of total reserves excludes the vault cash held by these institutions to service their deposits. This argument may be extended to the inclusion of required clearing balance contracts in total reserves: Federal Reserve deposits are held to service the customer deposits included in the monetary aggregates. A desire to maintain logical consistency among measures of money and measures of total reserves suggests that narrow measures of reserves may be incomplete.
Adjusted Total Reserves
Similar to the AMB, and for the same dates, adjusted total reserves is constructed in four segments. Within each segment, adjusted total reserves equals total reserves plus RAM. The overall adjusted total reserves series is built by chaining together the four segments at the same dates and in the same manner as the AMB.
Despite common practice, a correct time series for adjusted total reserves cannot be obtained by subtracting the currency component of M1 from the AMB. Rather, adjusted reserves must be calculated from unchained data on the adjusted reserves component of the AMB, and then chained as is done to measure the AMB.
The RAM adjustment is the same RAM used for the adjusted monetary base. Monthly and biweekly data are seasonally adjusted by the same methods used for the adjusted monetary base.
Adjusted Nonborrowed Reserves
Similar to total reserves, adjusted nonborrowed reserves is constructed in four segments. Within each segment, adjusted total reserves equals total reserves plus RAM. The overall adjusted nonborrowed reserves series is built by chaining together the four segments at the same dates and in the same manner as the AMB. This procedure is necessary because a correct time series for adjusted nonborrowed reserves cannot be obtained by subtracting borrowings from chained adjusted total reserve. Rather, adjusted reserves must be calculated from unchained data on the adjusted reserves component of the AMB, and then chained as is done to measure the AMB and adjusted total reserves.
The RAM adjustment is the same RAM used for the adjusted monetary base. Monthly and biweekly
data are seasonally adjusted by the same methods used for the adjusted monetary base.
Published Data Sources:
See Table A-2.
Table A-2: Sources of Data on the New Adjusted Monetary Base and Reserves
|
FRED® Abbreviation
|
Frequency of Data
|
FRB St. Louis Publication:
|
New Monetary Base and Total Reserves, Available on FRED® | |||
Revised St. Louis Adjusted Monetary Base, not seasonally adjusted |
ambns_r
|
monthly
|
none
|
Revised St. Louis Adjusted Monetary Base, seasonally adjusted |
ambsl_r
|
monthly
|
Monetary Trends
|
Revised St. Louis Adjusted Total Reserves, not seasonally adjusted |
none
|
monthly
|
none
|
Revised St. Louis Adjusted Total Reserves, seasonally adjusted |
aressl_r
|
monthly
|
Monetary Trends
|
Revised Monetary (source) Base, not seasonally adjusted |
sbase_r
|
monthly
|
none
|
Revised RAM, not seasonally adjusted |
ram_r
|
monthly
|
none
|
Revised St. Louis Total Reserves, not adjusted for changes in reserve requirements, not seasonally adjusted |
none
|
monthly
|
none
|
Revised St. Louis Adjusted Monetary Base, seasonally adjusted |
base_r
|
bi-weekly
|
U.S. Financial Data
|
Revised St. Louis Adjusted Total Reserves, seasonally adjusted |
adjres_r
|
biweekly
|
US Financial Data
|
Data Series on FRED® that will be discontinued at the end of 1996 | |||
Adjusted St. Louis Monetary Base (old), not seasonally adjusted |
ambns
|
monthly
|
none
|
Adjusted St. Louis Monetary Base (old), seasonally adjusted |
ambsl
|
monthly
|
Monetary Trends
|
Adjusted St. Louis Total Reserves (old), not seasonally adjusted |
adjresns
|
monthly
|
none
|
Adjusted St. Louis Total Reserves (old), seasonally adjusted |
adjressl
|
monthly
|
Monetary Trends
|
Monetary (source) base (old), not seasonally adjusted |
sbasens
|
monthly
|
none
|
RAM (old), not seasonally adjusted |
ram
|
monthly
|
none
|
Data Series on FRED® that are discontinued as of October 10, 1996 (data will not be updated or extended) | |||
Adjusted Fed Credit, nsa |
afcns
|
monthly
|
none
|
Adjusted Fed Credit, sa |
afcsl
|
monthly
|
none
|
Adjusted Fed Credit, sa |
afedcr
|
weekly
|
none
|
Richard G. Anderson and Robert H. Rasche. "A Revised Measure of the St. Louis Adjusted Monetary Base." (PDF [an error occurred while processing this directive] ) This Review, March/April 1996, pp. 3-13.
Richard G. Anderson and Robert H. Rasche. "Measuring the Adjusted Monetary Base in an Era of Financial Change." (PDF [an error occurred while processing this directive] ) This Review, November/December 1996, 3-37.
Charts for the paper are available in gif format.
Albert E. Burger and Robert H. Rasche. "Revision of the Monetary Base," Federal Reserve Bank of
St. Louis Review (July 1977), pp. 13-27.
John A. Tatom, "Issues in Measuring An Adjusted Monetary Base," this Review (December
1980), pp.11-29.
Comments and questions may be addressed to Richard Anderson at anderson@stls.frb.org or (314) 444-8574.