View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

llllproving tl1e
Monetary Aggregates

Report of the Advisory Committee on
Monetary Statistics

Board of Governors of the Federal Rcsc1 ve S,·stcrn
Federal Reserve Bank of St. Louis

Published in June 1976

Library of Congress Catalog Card Number 76- 14517

Copies of this report may be obtained from Publications Services, Division of Adm inistrative Services,
Board of Governors of the Federal Reserve System,
Washington , D.C. 20551. The price is 1.00 per copy;
in quantities of IO or more sent to one address, 85
cents ea h. Remittances should be made payable to
the Board of Governors of the Federal Reserve System
in a form collectible at par in U.S. currency. (Stamps
and coupons arc not accepted.)
Federal Reserve Bank of St. Louis

IIllproving the
Monetary Aggregates

Report of the Advisory Committee on
Monetary Statistics

Washington, D. C.
Federal Reserve Bank of St. Louis


In early 1974, the Board of Governors of
the Federal Reserve System asked a group of
promment economists to review the monetary
aggregates used by the Federal Reserve m
the formulation and implementation of
monetary policy The experts were asked
to focus their mvestigation on a techmcal
evaluation of the quality of the monetary
statistics m question
The Advisory Committee on Monetary
Statistics was chaired by Professor G L Bach
(Stanford Umversity), Professor Phillip D
Cagan (Columbia Umversity) served as Executive Secretary Other members of the Committee were Professor Milton Friedman
(Umversity of Clucago), Professor Clifford
G Hildreth (Umversity of Mmnesota), Professor Franco Modigham (Massachusetts Institute of Technology), and Dr Arthur Okun
(the Brookmgs Institution) Professor Paul
McCracken '(Umversity of Michigan) was a
member of the Committee origmally, but
withdrew because of the pressures of other
The Committee's report contams seven
prmcipal recommendations that relate to the
measurement, defimtion, adjustment for seasonal variation, and publication of the several
statistical series on the monetary aggregates
In selectmg the monetary aggregates to be
Federal Reserve Bank of St. Louis

exammed, the Committee was gmded, as its
i eport mdicates, by "received doctrine among
leadmg monetary economists and practitioners
m monetary policy "
The Board's research staff prepared a number of studies for the Committee A compamon
volume to this Report, contammg eight staff
papers drawmg on these origmal studies, will
be published by the Board at an early date
Board staff support of the work of the
Advisory Committee on Monetary Statistics
was supervised throughout most of the period
by James L Pierce, who at the time was
Associate Director of the Division of Research
and Statistics, and, subsequently, by Edward
C Ettm, Adviser m the Div1S1on of Research
and Statistics Members of the staff of the
Division of Research and Statistics and the
Division of International Fmance workmg
with the Committee were Darwm L Beck,
Helen T Farr, Arthur B Hersey, Darrel W
Parke, David Pierce, Richard D Porter,
Henry S Terrell, Thomas Thomson, and
Neva Van Peski In addition, Anton S Nissen,
Assistant Vice President at the New York
Federal Reserve Bank, contributed to the
work of the Committee
The Board of Governors greatly appreciates
the contribution made by the Committee and
will carefully conside1 its recommendations

Arthur F Burns, Chairman
Board of Governors
of the
Federal Reserve System








Summary of Recommendat10ns



Conceptual and Defin1t10nal Issues
Alternative aggregates _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


Recent financial developments _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


Problems of consohdat10n _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


How to define the "pubhc" What holders? _ _ _ _ _ _ _ _ _ _ _ __

Federal Reserve Bank of St. Louis




Measurement Issues and Recommendat10ns
How M1 1s constructed now _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


Very short-run (transitory) variations m
the monetary aggregates _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


Nonmember bank deposits _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


Consolidation of data from
different financial mst1tut10ns _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


Seasonal adjustment _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __


Ownership of demand deposits _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __



Summanes of Staff Papers
Federal Reserve Bank of St. Louis





The Board of Governors requested tlus Committee to review the basic monetary stat1st1cs
(especially the so-called monetary aggregates)
used by the Federal Reserve m formulatmg
and conductmg monetary policy, to evaluate
the1r adequacy, and to present suggest10ns for
the1r improvement It asked the Committee to
study and make recommendat10ns only on the
stat1st1cs m quest10n-not to evaluate monetary policy or to mvest1gate the s1grnficance of
the aggregates relative to mterest rates or credit
market md1cators
We have adhered ngorously to this d1rect1ve
It 1s 1mposs1ble, however, to select the monetary totals, or aggregates, to examme without 1udgmg to some extent the1r usefulness for
policy purposes (for example, decidmg that the
"money stock" as usually defined to mclude
ad1usted demand deposits plus currency m the
hands of the public-generally termed M 1may be important for policy purposes) In makmg these dec1s10ns, we have been gmded as far
as possible by received doctrme among leadmg
monetary economists and pract1t1oners m
monetary policy, rather than 1mposmg our
views as to the optimal theoretical approach
to policy issues Indeed, the views of members
of the Committee differ substantially on these
The present basic monetary stat1st1cs of the
Federal Reserve are the product of many
years of mtens1ve work-by mdependent research workers as well as by Federal Reserve
staff members The stat1st1cs have been
steadily improved over the years by repeated
Federal Reserve Bank of St. Louis

rev1S1ons We have been impressed by the
care and qualt1y of work devoted to collectmg
and combmmg the data that comprise these
senes Yet conceptual ddiiculties have led to
contmumg debates over some of the senes,
and techrncal problems of data collect10n and
processmg have prevented compilat10n of
senes m full accord with the conceptual
foundat10ns In view of the substantial weight
given monetary aggregates m recent years, it
is important that the data used be the best
that 1t is possible to obtam
As we emphasize m the sect10n on conceptual issues, no one monetary aggregate
1s clearly preferable to all others on all scores,
each has its theoretical and practical strengths
and weaknesses as a gmde to, or mtermediate
target for, monetary policy operat10ns, and as
a measure of the effectiveness of such operat10ns Given our terms of reference and the
limitat10ns imposed by the time available, we
have concentrated mamly on the reserve base,
or "high-powered" money, and on the mam
deposit-based senes (M 1 , M 2 , M 3 , M 4 , and M 5 ),
suggestmg changes that we believe are feasible
at a reasonable cost and that could substantially improve the conceptual validity and
measurement accuracy of the aggregates mvolved We provide first a summary of our
ma1or recommendations, with very bnef explanat10ns of each, then a more complete analysis of the conceptual and defirnt10nal issues
mvolved, and finally a detailed rat10nale for
the specific statistical recommendat10ns made
by tlus Committee


Improvmg the Monetary Aggregates Committee Report

While, withm the broad framework of our
assignment, we have had complete mdependence as to topics to consider, approaches to
issues mvolved, and recommendat10ns, we wish
to acknowledge the extensive and mvaluable
assistance provided by members of the Board's
economic and statistical staff They have been
fully cooperative and have produced several
dozen special studies at our request, many of
them of very substantial magmtude Without
this staff work we would have faced a vastly
Federal Reserve Bank of St. Louis

longer task of contractmg out such studies to
others, mdeed no outsider could have accomplished a number of the detailed tasks we assigned m exammmg both present statistical
procedures and alternatives we wished to consider We deeply appreciate this assistance
The staff has combmed these studies mto
eight Staff Papers, which are published as a
separate volume The Appendix at the end of
this report lists and briefly summarizes the
Staff Papers



Summary of Recommendations


Alternative concepts of money

Since no one monetary aggregate is clearly
preferable to all others on all scores, we recommend that the Federal Reserve publish regularly the "reserve base," and the ma1or monetary aggregates currently designated M 1, M 2 ,
M 3, M 4 , and M 5 , although with substantial
modifications as indicated below


Nonmember bank deposits

To reduce large errors zn preliminary estimates of deposits at nonmember banks, we recommend prompt establishment of a weekly reporting sample of large and small nonmember
banks and collection of weekly-average-ofdazly-deposits data from nonmember banks
four times annually in connection with call
Recent experiments with a weekly reportmg
sample of large and small nonmember banks
convmce us that regular collection of such add1t10nal mformat10n could dramatically reduce
the large errors now often made m est1matmg
nonmember bank deposits between call report
dates, and that such data could be obtamed
and processed without unreasonable cost to
either reportmg banks or to the Federal Reserve In add1t10n, the present hm1tat10n of
call report data to a smgle day 1s a substantial
add1t10nal source of error that could be eliminated by collectmg weekly-average data on call
reports In combmat1on, these two reforms
could substantially ehmmate the errors now
faced m estimatmg nonmember bank deposits 1
As this report 1s bemg completed, we are mformed
that the Federal Deposit Insurance Corporation
(FDIC) will begm to collect from nonmember banks
7 days of deposit data for the week surroundmg each
call report date, begmnmg m March 1976
Federal Reserve Bank of St. Louis


Consolidation of accounts at
different financial mst1tut10ns

We tentatively recommend a new, simpler
process of handling interbank deposits and
cash items in process of collection when consolidating data from different financial institutions, in order to eliminate certain biases and
to obtain a more accurate measure of M 1 and
other aggregates.
There 1s general agreement that cash items
m process of collect10n (mamly checks) should
be deducted from demand deposits on banks'
books, m order to avoid the double countmg
of deposits already credited to accounts of
rec1p1ents but not yet deducted from accounts
of payers However, serious problems arise m
makmg tlus ad3ustment, because some banks
that clear checks through correspondents
show checks m process of collect10n on their
balance sheets as "due from banks" mstead
of "cash items m process of collect10n
(CIPC)", because cash items mclude checks
not drawn on private accounts (for example,
checks on U S Treasury, mterbank, and some
foreign accounts not mcluded m Mi, as well
as money orders, redeemed Government bond
coupons, and food stamps), and for other
reasons to be detailed later
To ehmmate this apparent bias (overstatement), we tentatively propose an alternative
means of consohdatmg the accounts of the
banks mvolved-bas1cally by deductmg directly from gross demand deposits (wluch mclude "due to banks") both "due from banks"
and "cash items," m lieu of the present more
elaborate set of ad3ustments made to obtam
ad3usted demand deposits Prehmmary calculatwns made at our request by the Board's
staff suggest that tlus change would reduce the


lmprovmg the Monetary Aggregates Committee Report

level of M 1 by some $8 billion from presently
reported figures and would reduce week-toweek and month-to-month variat10ns m the reported money stock, which may now reflect
spurious fluctuations m the volume of domestic
and foreign checks m process of collect10n
Smee the reasons for the large reduct10n m the
reported level of M 1 under this new method as
compared with the method now used are not
entirely clear, however, we recommend adoption of the proposed change only tentatively,
sub1ect to further mvest1gat10n by the Board
along the Imes currently bemg undertaken


Foreign deposits in the Umted States
and U .S dollars held abroad

To obtain the most useful aggregates for
U S policy decisions and actions, we recommend elimination from the U S monetary
aggregates of deposits held in the United States
by foreign commercial and central banks and
other official institutions, and continued exclusion of U.S. dollars (Euro-dollars) held
In an open economy hke that of the Umted
States, mteractions between domestic and mternat10nal transactions on trade and capital
accounts make 1t 1mposs1ble for the monetary
authorities to consider only domestic consequences of their act10ns-and by the same
token make any purely domestic measure of
the money stock to a degree unsatisfactory as
an mtermediate target variable As there 1s no
one ideal concept of money for domestic monetary control purposes, so there 1s no one ideal
concept for an open economy or for the world
economy, the existence of mternat10nal transact10ns that mteract with domestic transact10ns
m the Umted States makes the defimt10n problem more difficult than for a purely domestic
economy However, given the theoretical difficulty of prescribmg any ideal amount of foreign or mternat10nal money to be mcluded m
the U S money stock, the practical difficulties
m obtammg the desired data even 1£ they could
be conceptually specified, and the relatively
Federal Reserve Bank of St. Louis

modest role played by mternat1onal transact10ns m the U S economy-we recommend, as
a practical matter, use of a concept of money
focused primarily on the domestic economy
At present, all depas1ts of foreign md1v1duals and busmesses, foreign commercial banks,
and foreign central banks and other official
mstitut1ons at banks m the Umted States are
mcluded m the U S money stock, and no
US dollar deposits abroad (for example,
Euro-dollars) are mcluded, no matter who,
owns them We recommend mclud10g foreigners' deposits m the Umted States where
these are likely to be used primarily for purchases of US goods, services, and securities
and exclud10g all U S dollar deposits abroad
-mamly because there 1s no practical way of
mcorporatmg these data mto current US
money stock series even though some such
balances may be held pnmanly with a view
to purchases m the Umted States Applymg
these criteria, we recommend that depas1ts of
foreign commercial banks and foreign central
banks and other official mstitut10ns 10 the
Umted States be excluded from the US
money stock, smce these are apparently held
primanly for clearing Euro-dollar transactions, for financmg foreign exchange transactions, and as mternat10nal monetary reserves, but that deposits of foreign md1v1duals
and busmesses contmue to be mcluded
The Federal Reserve should, however, contmue to publish, as memorandum Items, data
on deposits of foreign commercial and central
banks and other official mstitutions m the
Umted States and US dollar deposits abroad,
so that those wishmg to mclude them 10 the
U S monetary aggregates, or to use them for
other purposes, will be able to do so


Seasonal adjustment of
monetary aggregates

We recommend that the Federal Reserve
authorities publish periodically the seasonal
ad1ustment factors they propose to use in arriving at the desired money stock throughout
the year ahead (the "policy" seasonal), so that


Summary of Recommendations

the Fed's attempts to eliminate seasonal variations will not be confused with more basic
determination of the desired money stock or
other monetary aggregates We further recommend that, in estimating seasonal ad1ustment
factors for the money stock, looking backward
(the "descriptive" seasonal), the Fed substitute
for the so-called Census Bureau X-11 seasonal
adjustment method a modified method that
more effectively uses the daily data available
The Federal Reserve authont1es and most
other users of monetary statistics work pnmar1ly with seasonally adjusted senes Because the
Federal Reserve itself to a substantial extent
controls the amount of money, to isolate any
"natural" seasonal m the money stock-mdependent of Federal Reserve policy act10ns-is
very difficult To a considerable degree the
Fed produces the seasonal vanat10ns that
exist m observed Mi, partly m order to reduce or elimmate seasonal vanat10ns in interest rates Thus, when the Fed publishes
h1stoncal money stock senes, seasonally adJUsted by usmg a "descriptive" seasonal reflectmg seasonal patterns m the money stock
after Federal Reserve policy act10ns, users
should recogmze that such seasonally adjusted
data are not necessanly those that were used
by Federal Reserve authont1es m makmg their
policy dec1S1ons The Fed should also contmue to publish seasonally unadjusted data
for the monetary aggregates


Short-run (transitory) variations m
the monetary aggregates

To highlight the dangers of overemphasizing short-run variations in the monetary aggregates, we recommend that the Fed publish
further information on the short-run, nonsystematic or transitory, variability of the monetary aggregates
Apart from seasonal and basic longer-term
movements, the monetary aggregates are sub11ect to a variety of shoi t-term day-to-day and
week-to-week vanat10ns that anse from fluctuatmg payments among the Treasury, the public, and the banks, items m process of collec-
Federal Reserve Bank of St. Louis

t10n, reportmg and tabulatmg errors, and the
like Our analysis suggests that such day-to-day
transitory vanat10ns alone can mtroduce a substantial, nonsystematic variability, or error, m
reported growth rates From month to month
the transitory component m the annualized
growth rate of M 1 is likely to exceed 2½ percentage pomts one-third of the time, from
quarter to quarter, to exceed ½ percentage
pomt one-third of the time The comparable
transitory component, or error, m M 2 will be
about half as large Users should be aware of
the dangers of placmg too much emphasis on
reported short-term vanat10ns m the monetary
aggregates, especially on less than quarterly


Recent financial developments and
the monetary aggregates

Recent financial developments suggest the
possibility of radical changes in the Nation's
payment order or withdrawal accounts and
we do not recommend changes in the definition of M 1 or other monetary aggregates now,
we do recommend that the Federal Reserve
begin to collect and publish systematically data
on new close substitutes for demand deposits
( such as negotiable order of withdrawal and
payment order of withdrawal accounts and
overdraft facilities if possible), and that it
develop experimental aggregates that combine demand deposits with those savings accounts that are readily convertible to a
demand basis
Fmancial mnovat10n and regulatory changes
have been rapid m recent years Combmed with
the prolubit10n of payment of explicit mterest
on demand deposits and other regulatory
changes, high mterest rates have stimulated the
development of vanous close substitutes for
demand deposits These substitutes are still relatively small m dollar amounts, but they may
be begmnmg to have substantial effects on the
rate at which the currently defined money stock
turns over If these developments contmue,
they may change substantially the lustoncal


Improvmg the Monetary Aggregates Committee Report

relat10nships between the present monetary
aggregates and aggregate demand for goods and
services Thus, the Federal Reserve and other
supervisory agencies should begm now to col-
Federal Reserve Bank of St. Louis

lect and analyze the data needed to understand
these new relationships as they develop, mcludmg the possible introduction of new aggregates to take new developments mto account



Conceptual and Definitional Issues

In conductmg monetary policy, the Federal
Reserve should use as an mtermediate target
that monetary total (aggregate), or those
totals, through which it can most reliably affect
the behav10r of its ultimate ob1ectives-the
price level, employment, output, and the like
Wluch total or totals best satisfy that requirement depends m turn on (I) how accurately
the total can be measured, (2) how precisely,
and at what costs mcludmg unwanted side
effects, the Fed can control the total, and (3)
how closely and reliably changes m the total
are related to the ultimate policy ob1ectives
If a total cannot be measured, it cannot be
used effectively as an mtermediate target for
policy purposes At the same time, for the purposes of Federal Reserve policy there is little
pomt to measurmg precisely some total that
the Fed cannot control or that has no mfluence
on ultimate policy objectives Accordmgly,
there is no way to tackle the problem of measurement without implicitly or explicitly reachmg conclus10ns about the feas1b1lity of control
and the closeness of mfluence We have done so
mostly by relymg on our prior general mformat10n and the preva1lmg views of experts on
monetary policy rather than by special studies,
though we have made some special calculat10ns
m trymg to decide how to handle borderlme
To avoid havmg to go more deeply mto the
problems of control and mfluence, we have not
tried to select a smgle aggregate but rather
have dealt with a number of alternative monetary aggregates and exammed how the measurement of those aggregates could be improved
Considerat10ns of feasibility have narrowed
our task by ruhng out some totals that economic analysis suggests would be superior to
Federal Reserve Bank of St. Louis

the totals that can actually be measured Here
the mam issue is between measuring monetary
totals (1) as they are recorded on the books of
the ultimate owners or holders of money (md1v1duals m their capacity as ultimate wealth
holders, busmess enterprises, governmental
bodies other than the monetary authorities),
or (2) as they are recorded on the books of
financial mstitut10ns
On analytical grounds, we would prefer to
measure the total amount of currency and various categories of deposits held by the "public"
m whatever form or mst1tut10n or place If possible, of course, 1t would be desirable to subdivide monetary totals among various groups
and locat10ns of holders However, 1t 1s currently impractical to measure such totals
directly The basic available data are reported
by the issuers of US currency (the US Treasury and the Federal Reserve) and by the financial mstitut1ons whose liabilities are
generally labeled "deposits" (commercial
banks, mutual savmgs banks, savmgs and loan
associat10ns, and credit umons), rather than by
the owners of the currency and deposits
As a result, we have been forced to restrict
ourselves to totals that can be constructed from
the books of financial mst1tut10ns That 1s, we
have been forced to accept data correspondmg
to the characteristics and nat10nal locat10n of
the issuers of currency and of the financial mst1tu t10ns, or correspondmg to the character of
their hab1ht1es or assets, rather than, as we
should prefer, data directly from money holders on their monetary assets However, the data
from the issuers do permit some distmct10ns
among holders, so one important question 1s
what holders of money to mclude m the


lmprovmg the Monetary Aggregates Committee Report

In considermg conceptual and definitional
issues, four topics deserve particular attention
(1) alternative aggregates, (2) effects of recent
financial developments on these totals, 1 (3)
problems of consolidation of accounts, and (4)
how to define the public

Alternative aggregates
Three distmct1ve bases for definmg monetary totals have played an important role m
monetary literature In addition, there has
been much mterest m credit, as d1stmgmshed
from money
Measurmg credit totals mvolves problems
that differ from those m measurmg monetary
totals For banks, mformat1on on credit totals
comes from the asset side rather than from the
liability side of the balance sheet Further,
"bank credit" can be viewed as part of much
broader totals, which could also mclude commercial paper, Treasury bills, longer-term Government securities, corporate bonds, mortgages,
and so on m great diversity Some elements of
these totals are held by mstitutions whose liabilities are mcluded m one or another monetary total, other elements are not In our Judgment, there 1s substantial mformational value
m credit totals and components, but we have
not been able to consider their measurement
adequately Hence, we make no specific recommendations concernmg them
With respect to monetary aggregates, one
basis for definmg such a total 1s to regard
money as correspondmg to assets that are
generally used to discharge obhgat1ons and
that are not the explicit liability of nongovernmental entities m the soCiety Traditionally
such assets have corresponded to specie In the
Umted States today they correspond pnmarily
to the non-mterest-bearing fiat issues of the
ultimate monetary authority The terms "lughpowe1 ed money" and "monetary base" have
been used to refer to tlus total We shall refer
to 1t as "the base "
For the Umted States today the base mcludes
all currency outside the Federal Reserve and
the Treasury plus all bank deposits at Federal
Federal Reserve Bank of St. Louis

Reserve Banks It 1s the total among those considered here that can probably be most accurately measured a1~d most p1ec1sely controlled
by the Fed There are a few (but relatively
mmor) amb1gmt1es about its measurement
Moreover, Federal Reserve control 1s not complete and immediate Commercial banks may,
at their option, borrow some additional reserves at the Fed The laggmg of actual reserve
reqmrements behmd the deposits to wluch
they are related means that the Fed m effect
must provide additional reserves when banks
run short, smce 1t 1s impossible for the banks
to alter the deposits they had held I or 2
weeks prev10usly But the maJor defect of the
base arises from the widespread belief, based
on both empirical studies and theoretical
analyses, that this total m the Umted States 1s
less closely and reliably lmked to the ultimate
obJecuves of policy (employment, prices, and
so on) than are some other aggregates However, there are some bits of evidence m the
opposite direction, and this total does have
the great advantage of bemg less subJect to
mfluence by financial mnovations than are
broader totals Hence, we recommend that the
Fed regularly publish figures on the base, as
defined above-probably on a weekly basis
along with the current money stock data
A second basis 1s to regard money as correspondmg to assets generally used to discharge
debts, that 1s, those assets that are used as
"media of exchange" Tlus criterion 1s sometimes hard to apply (for example, 1s a $10,000
bill to be regaided as a medmm of exchange,
are postal money orders,), but these problems
of classification are, m practice, mmor There
1s little dispute that-at the present time m
the Umted States-currency, commercial bank
demand deposits, and t1aveler's checks are the
only maJor items generally used as media of
exchange and that the bulk of these items can
be so used However, as we note m the next
sect10n, recent developments-such as "checkless" computerized payments and, m effect, permittmg checks to be written on savmgs accounts at commercial banks, savmgs and loan
mstitutions, and mutual savmgs banks-may


Conceptual and Definitional Issues

be changing tlus s1tuat10n so substantially as to
reqmre a revised definit10n of the ma1or monetary aggregates
The symbol M 1 rs generally used to refer to
tlus concept of money as a medmm of exchange P1oblems of measuring M 1 , as 1t has
ordina1 ily been defined, anse pnmanly from
the necessity of estimating the amounts held
from bank records, these d1fficult1es will be
considered late1 Problems of controlling M 1
anse from changes in the rat10 of currency held
by the pubhc to its demand deposits and from
changes in the rat10 of demand deposits to
bank 1eserves, the latter in turn reflect shifts of
funds among different categories of bank deposits, changes in excess reserves held by banks,
and the lag in reserve reqmrements Such
changes alter what 1s termed the "money mult1pher"-that 1s, the rat10 of M 1 to the base
They also ause from inadequate data on nonmember banks, and on cash items and related
transact10ns Most students regard M 1 as more
closely and more reliably related to ultimate
ob1ect1ves than rs the base But there 1s some
concern that financial innovat10n 1s changing,
and probably loosening, the relat10nsh1p between M 1 , as 1t 1s ordinanly defined, and such
ultimate pohcy goals as employment, output,
and pnces
A tlurd basis 1s to regard money as assets that
serve as a "temporary abode of purchasing
power"-in wluch sellers of goods, services, or
financial assets hold the proceeds in the intenm
between sale and subsequent purchase of other
goods, services, or assets-and that are, or are
readily convertible into, media of exchange
Tlus hqmd asset concept 1s regarded by some
scholars as coming closest to capturing the essential feature of money and as more closely
and 1ehably ielated to ultimate ob1ect1ves than
the other concepts
Unfortunately, tlus concept has the most
ambiguous em pineal content of the three measures It can correspond to M 1 plus time and
savings deposits of commercial banks other
than large certificates of deposit (CD's) (now
defined as M 2), or M 2 plus deposits at mutual
savings banks, savings and loan associat10ns,
Federal Reserve Bank of St. Louis

and credit un10ns (now defined as M 3), or
either M 2 or M 3 plus large CD's (now defined
as M 4 and M 5) Each of these totals, in turn,
could be constructed differently by, for example, d1stingu1shing time from savings deposits, as suggested in the following sect10ns
And still broader aggregates could be constructed by including such items as Treasury
bills, Senes E Government bonds, cash surrender value of hfe insurance pohc1es, and so
on In general, though by no means umformly,
the broader the concept, the greater the problems of measurement and cont10l

Recent financial developments

This IS a particularly difficult time at wluch
to determine definitively the precise empmcal
counterparts to the alternative aggregate concepts hsted in the preceding sect10n Financial
innovat10n and regulatory changes affecting
the payments mechanism have been particularly rapid in iecent yeais, stimulated by the
very high rates of interest Given the p1olub1t10n on the payment of exphot interest on demand deposits and the inab1hty of savings and
loan associat10ns, mutual savings banks, and
credit unions to hold demand deposits, these
lugh interest rates have stimulated the development of substitutes fo1 demand deposits
Combined with the differential ceilings on interest rates that may be paid on vanous categories of time and savings deposits and the
changes in these ceilings over time, the lugh
interest rates have stimulated d1ffe1ent1at10n of
deposit categories and alterat10ns in their cha1actenst1cs
The base 1s the only total about whose empineal counterpart these developments raise
no problems, though, of course, these developments may have affected the demand for the
base However, future developments may affect
tlus concept as well For example, some economists have proposed that the Federal Reserve
pay interest on deposits at Federal Reserve
Banks If tlus were to occur, the quest10n would
anse as to whether the base should be narrowed to correspond solely to non-interest-


Improvmg the Monetary Aggregates Committee Report

bearmg currency or broadened to mclude
other mterest-bearmg obligations of the Government

Certificates of deposit
The large negotiable CD's mtroduced m the
1960's seem very different from the earlier time
and savmgs deposits at commercial banks and
more like open market commercial paper Accordmgly, some scholars have elimmated them
from the monetary totals that they have used
m their studies, and the Fed has done the
same by excludmg marketable CD's m denommations larger than $100,000 from the
time and savmgs deposits that 1t adds to M 1
to ob tam M 2 However, the formal d1stmction
between negotiable and nonnegotiable seems
largely techmcal, smce banks generally permit
large purchasers to convert from one to the
other at will Moreover, the $100,000 d1vmon
1s clearly arbitrary and has a different s1gmficance at different pnce levels Hence, the present procedure must be regarded as makeshift
until enough ev1dence.1s accumulated to permit a more satisfactory resolution
We have made a number of tests to determme whether M 2 or M 4 1s more closely related
to nommal mcome and to the Fed's ultimate
policy obJect1ves On the whole, the evidence
favors M 2 But the evidence 1s weak, and it
may be that the results reflect simply the trans1t10nal effect of the mtroduction and rapid
growth of CD's Negotiable CD's have been
important for too few years to provide an adequate test Hence we recommend emphas1zmg
M 2 and M 3 , but contmumg to mom tor the performance of M 4 and M 5

NOW, POW, and similar accounts
There has recently been a proliferation of
expenments designed to provide the eqmvalent
of checkmg services to holders of what are
techmcally classed as savmgs deposits at commercial banks, mutual savmgs banks, savmgs
and loan associations, and credit umons, or of
accounts at so-called money market funds The
Federal Reserve Bank of St. Louis

NOW (negotiable order of withdrawal)
accounts permitted m Massachusetts and New
Hampslure are a dramatic example In addition, some savmgs and loan associations have
made arrangements to make telegraphic transfers of Federal funds at the order of holders of
larger accounts, and banks have been permitted
by the Federal Reserve and the FDIC to transfer savmgs to demand accounts on telephomc
order smce Apnl 1975 Many of the rapidly
expandmg money market funds, which belong
to a group of mstitut10ns heretofore entirely
outside the scope of the usual monetary totals,
have arranged with cooperatmg banks to permit the transfer of funds by check or its eqmvalent Moreover, smce November 1975 profitmakmg busmess corporat10ns have been permitted by the Federal Reserve and the FDIC
to hold savmgs accounts of up to $150,000 per
account, and they apparently shift funds frequently between demand and savmgs accounts
To date, NOW accounts and closely related
substitutes for demand deposits apparently
total only about 1 per cent of M 1 , but they are
mcreasmg rapidly m importance Estimates of
such accounts are now mcluded m M 2 and M 3
Some observers believe they should be mcluded
m M 1 In our Judgment it is too soon to make a
defimte decision on how to treat these accounts,
especially smce there may be substantial
changes m their character or sigmficance m the
next few years, mcludmg possible changes m
regulat10ns applymg to them and related bank
deposits Because many such substitutes are
claims on mstitutions not subject to regulat10n
by the Federal Reserve, collection of data on
them will reqmre the active cooperation and
assistance of other Government agencies that
now obtam data from the relevant mstitutions
We urge such agencies to cooperate with the
Fed for tlus purpose
If present trends contmue, we suspect that
w1thm not more than 2 or 3 years it will be
desirable for the Federal Reserve to re-examme
the treatment of such accounts and possibly to
alter the defimt10ns of some of the present
aggregates to take them more specifically mto

Conceptual and Definitional Issues


Distinction between time and
savings accounts

cons1derat10n of the treatment of savmgs and
time deposits

The development of CD's on the one hand
and of substitutes for demand deposits on the
other, plus the differential ceilmgs on mterest
rates that may be paid on savmgs and time deposits, have mcreased the practical importance
of the distmct10n between time and savmgs
accounts Tlus distmct10n was for many years
purely formal, and such accounts are combmed m such aggregates as M 2 and M 3 But
more recently, the distmct10n has become
more sigmficant Smee July 1973, banks have
been requrred to impose relatively large mterest penalties on the withdrawal of time
deposits before maturity, and as a result
holders of such deposits now have reason to
take the maturity more seriously In effect,
savmgs deposits have become much more
similar to demand and checkmg deposits, as
noted above, and time deposits more similar
to securities
Thus, 1t may well be that a better empmcal
counterpart than M 2 or M 3 to the concept of a
temporary abode of purchasmg power will
be totals that add to M 1 only savmgs deposits
at commercial banks as an alternate to M 2 , and
at mutual savmgs banks and savmgs and loan
associat10ns as an alternate to M 3, and that
exclude time deposits at both Or, 1t may be
that such a demand-plus-savmgs-depos1t total
will be a replacement for M 1, correspondmg to
the "media of circulation" concept, and that a
broader total, which mcludes not only time deposits but also CD's, money market funds, and
perhaps other items, will be a replacement for
M 4 and M 5 to correspond to the concept of a
"temporary abode of purchasmg power"
Here agam, 1t 1s too soon to recommend a
change But we commend the Board's staff for
some prehmmary studies that explore new
totals constructed along these Imes We recommend strongly that such studies be contmued,
and that the recons1derat10n of monetary aggregates some 2 or 3 years from now as we suggested above be accompamed by a parallel re-

Nonbank traveler's checks
Federal Reserve Bank of St. Louis

Liability for traveler's checks issued by banks
1s now mcluded m M 1 as part of demand deposits In prmciple, hab1hty for traveler's
checks issued by nonbanks should also be mcluded, but 1t 1s not, simply for lack of data
This 1s not a new problem, but 1t 1s mcluded
here because 1t 1s the same kmd of problem as
those raised by recent financial mnovat10ns
However, its importance 1s changmg, partly
because of w1demng compet1t10n m the issuance of traveler's checks Unfortunately, we
see no practical way to remedy this defect now

Credit cards and the ((checkless society"
An mcreasmg volume of purchases 1s bemg
made on credit cards, and direct credits of
wages and salanes to bank accounts and debits
to purchasers' bank accounts by sellers through
computer networks will probably spread m the
years ahead, perhaps movmg the economy toward an mcreasmgly "checkless society" m the
foreseeable future Insofar as credit-card purchases and the ehmmat10n of physical bank
checks merely provide more convement and
efficrent means of transferring demand deposits,
they do not call for any redefimt10n of the
money stock-although they may lead to a
lugher velocity of crrculat10n Insofar as they
actually mvolve creat10n of new transferable
money by sellers who temporarily mcrease
the spendmg power of buyers, they certamly
mcrease the volume of credit, although they
do not mcrease M 1 as now defined or as 1t
might be defined m response to the financial
developments so far considered
If credit cards and a checkless socrety largely
supplant present methods of payment, 1t will
become desirable to redefine M 1 and the other
deposit totals based on 1t m a more fundamental way For the time bemg, however, we
recommend no change m the defimt10n of the
totals based on the growmg use of credit cards
and direct cred1tmg and deb1tmg of deposit


lmprovmg the Monetary Aggregates Committee Report

accounts through computerized systems that
elimmate or postpone use of physical checks
Increasmg use of these payments methods may
raise measurement problems because they may
mcrease such items as bank float (considered
later) and mterbank deposits relative to net
holder balances, but they do not yet raise
ma3or defimt10nal problems

Problems of consolidation
A ma3or techmcal problem m constructmg
estimates of M 1 through M 5 1s the correct consolidat10n of the accounts of mdividual financial mst1tut10ns and their depositors The most
obv10us example 1s mterbank deposits The
sum of all deposit liabilities on the books of
banks will exceed the correspondmg assets of
the nonbank public by the deposit liabilities
of some banks to other banks Consolidatmg,
1ather than simply combmmg, the books of
the banks reqmres the ehmmat10n of such
mterbank deposits This general prmc1ple 1s
clear, but its applicat10n raises difficult problems, wluch we will consider after first dealmg with bank float

Bank float
A less obv10us example is cash items m the
process of collect10n When X deposits to his
account m Bank A a check drawn by Y on
Bank B, he 1s given immediate credit by Bank
A, wluch matches its mcreased liability by an
mcrease m the asset category, CIPC Bank A
forwa1ds the check to a Federal Reserve Bank
for collect10n The Federal Reserve Bank credits Bank A's account at the Federal Reserve
Bank (Bank A m turn transfers the correspondmg sum "Items m process of collect10n" to
"deposits at Federal Reserve Bank"), debits
Bank B's account, and forwards the check to
Bank B, wluch subsequently debits Y's account
During any time mterval that elapses between cred1tmg the check to X's account and
debitmg it to Y's account, total deposit liabilities m the system are lugher m tlus account
by the amount of the check In the usual
Federal Reserve Bank of St. Louis

termmology, that sum is double counted by
the amount of the bank float This assumes
that X views the check as added to his cash
balance as soon as he deposits 1t and that Y
deducts the check from lus balance when he
estimates that X will have deposited it This
1s by no means the only possible assumpt10n
Another 1s that Y deducts the check when he
writes it-wluch adds mail float to bank float
as double countmg The other extreme 1s
that Y does not deduct the check until he
estimates it has been debited agamst lus account-m wluch case check-kitmg offsets bank
The1e 1s no completely convmcmg evidence
as to how transactors view the t1mmg of debits
and credits to their accounts, but various
earlier studies have suggested that deductmg
bank float (but not mail float, wluch m any
case cannot be estimated sat1sfactonly) produces an M 1 total that 1s more closely related
to nommal nat10nal mcome (an mtermediate
ob3ect1ve of Federal Reserve policy) than 1s
the total obtamed without the deduct10n
Special calculat10ns made at the request of
the Committee gave the same result Hence
we have accepted this concept m analyzmg
the problem of bank float
The est1mat10n and deduct10n of bank float
1 a1se measurement problems because of the d1£fe1 ent treatment of similar items by different
banks (for example, entermg a check en route
to a correspondent bank as "due from banks"
rather than CIPC) and because of the possible
mclus10n m CIPC of items that have no
counterpart m the deposit total and hence cannot be regarded as double countmg (for e~ample, food stamps or checks on other banks
whose deposits are already deducted when mterbank deposits are subtracted) These measurement items are discussed further m Sect10n 4

"Bank" versus "public"
The various aggregates 1mphc1tly mvolve
drawmg different Imes between banks and the
public, which m turn call m principle for different consolidat10ns of accounts The most

Conceptual and Definitional Issues

obv10us example is the difference between M 4
and M 5 M 4 treats commercial banks as banks
but mutual savmgs banks, savmgs and loan
associat10ns, and credit umons as part of the
public Hence, deposits of the thnft mst1tut10ns at commeroal banks are not treated as
mterbank deposits and are not excluded m
estimatmg M 4 They do not duplicate any of
the commercial bank liabilities to the public
mcluded m M 4 For M 5 , on the other hand, the
deposits of the public (now defined to exclude
the thnft mstitut10ns) at the thnft mstitut10ns
are mcluded But it would be double countmg
to mclude both these deposits and the deposits
of thnft mstitut10ns at the commercial banks
These should be regarded as mterbank deposits-a treatment that is not now followed
This problem anses for every total, and we
shall consider them one at a time, after which
we shall give our recommendat10ns for all
For the base, only the Federal Reserve and
the Treasury are banks, everythmg else is
the public Hence, the base correctly elimmates
only Treasury deposits at Federal Reserve
Banks and cmrency held by the Treasury and
the Federal Reserve It correctly mcludes all
cash m the vaults of commercial banks and
thnft mstitut10ns and all non-Treasury deposits at Federal Reserve Banks because these
do not duplicate any other element m the


not be counted twice Smularly, deposits held
by commercial banks at other commeicial
banks or at the Federal Reserve on account of
demand deposits should be treated as mterbank deposits smce they do not correspond to
liabilities to the public
On the other hand, vault cash or deposits at
other banks held by commercial banks on account of time deposits a1e, for the M 1 total, m
the same category as cash or deposits at other
banks held by mutual savmgs banks or savmgs
and loan mst1tut10ns They do not duplicate
any other item mcluded m M 1 and hence
should not be subtracted
The conceptual issue is clear Howeve1, it is
not easy to carry out the conect treatment m
practice because there is no way to connect
particular asset items with particular liability
items How can we determme what fract10n of
commercial banks' vault cash and deposits at
other banks 1s held on account of demand deposits and what fract10n on account of other
liabilities? Because of tlus difficulty, the d1stmct10n 1s not made now All vault cash and
deposits of commercial banks at othe1 banks
are subtracted m calculatmg M 1 The result 1s,
on tlus account, an underestimate of the conceptually valid total
The part of bank vault cash and of deposits
at other banks held by commercial banks on
account of large negotiable CD's should not be,
but 1s, subtracted m calrulatmg M 2 The error
1s m the same dnect10n as for M 1 but of course
1t 1s much smaller m magmtude


In some ways M 1 raises the most troublesome problem For this total, banks mclude, m
add1t10n to the Federal Reserve and Treasury,
only that part of commeroal banks that corresponds to their demand deposit liabilities
Any vault cash held by commercial banks on
account of demand deposit liabilities should
be deducted m computmg the currency holdmgs of the public smce the public holds this
cash mdirectly through its demand deposit
holdmgs at commercial banks, and it should
Federal Reserve Bank of St. Louis


The cash held by thuft mst1tut10ns and then
deposits at commercial banks other than any
large CD's held should be subtracted m computmg Ms because these items duplicate their
liab1lit1es to the public Smee tlus 1s not done
now, Ms 1s accordmgly overstated
Tlus total 1s the only one other than the base
that 1s now conceptually correct For tlus total,


Improvmg the Monetary Aggregates Committee Report

all vault cash and deposits at other banks held
by commercial banks should be and are subtracted

Like Ma, the total 1s currently overstated because vault cash and deposits of the thnft mstitut10ns at other banks are not subtracted
The accounts of the thnft mstitut10ns are combmed, rather than consolidated, with the accounts of the commercial banks
We recommend that m computmg Ma and
M 5 the Federal Reserve consolidate rather than
combme the accounts of commernal banks and
thnft mstitut10ns This means that currency
holdmgs of thnft mstitut10ns plus their deposits at commercial banks should be subtracted from Ma as currently calculated, and
that this sum plus large CD's held by thnft
mstitut10ns should be subtracted from M 5 as
currently calculated
In order for this consolidation to be feasible,
1t will be necessary that the agencies now collectmg data from thnft mst1tut1ons reqmre
them to report a more detailed breakdown of
the category "liqmd assets" than they now report At a mmimum, mformat10n will be reqmred separately on currency plus demand and
time deposits at commercial banks, large CD's,
and other Iiqmd assets
While we recogmze that M 1 and M 2 as currently calculated are not conceptually precise,
we do not propose any change m present procedures because of (l) the arbitranness of any
div1Sion of commercial bank holdmgs of currency and deposits at other banks mto the
parts held on account of demand deposits, large
CD's, and other time and savmgs deposits, and
(2) the belief that the error mvolved m the
present procedure is reasonably stable over
time and hence does not affect seriously estimates of changes over time 2
•Dissenting footnote by Milton Friedman, m which
Ph1ll1p Cagan concurs I believe 1t would be desirable
to attempt to correct the error m M 1 and M 2, despite
Federal Reserve Bank of St. Louis

How to define the public What holders?
As currently defined, M 1 , M 2 , Ma, M 4 , and
M 5 all attempt to measure holdmgs of the relevant categones of assets by the nonbank pubhc-mdividuals, partnerships, corporat10ns
(IPC), States, counties, mumcipahties, and
Government agencies other than the Federal
Reserve and the Treasury-with some except10ns Though holdmgs of the U S Treasury
and the Federal Reserve are excluded as holdmgs of banks rather than of the public, mmor
amounts of currency held by U S Government agencies are mcluded The totals mclude
currency held abroad and deposits held m the
Umted States at commercial banks and Federal Reserve Banks by the foreign public, foreign banks, and foreign governments, central
banks, and mternat10nal mstitut10ns They
exclude dollar deposits held by the US
public at banks (mcludmg U S banks and
branches) located m foreign countnes or m
US terntones and possess10ns
Some of the deviat10ns of current practice
from the ideal concept are necessitated by lack
the two valid pomts made m this paragraph The error
1s substantial m absolute size and unless an attempt 1s
made to ellmmate 1t, we shall not know when and 1f 1t
vanes over time
The procedure I recommend as a mmimum is to allocate to deposits other than demand deposits, m computmg M 1 , an amount of cash and deposits at other
banks equal to reqmred reserves on deposits other than
demand deposits, and m computmg M 2 , to allocate to
large CD's reqmred reserves on such CD s The 1mpllc1t
assumpt10n that banks hold zero precaut10nary or free
reserves on account of deposits other than demand deposits 1s no doubt extreme, so not all error would be
ellmmated, but It seems to me far more reasonable than
the current imphc1t assumpt10n that they hold zero reqmred reserves agamst these deposits A more senous
error 1s to regard no demand deposits at other commercial banks, and no items m process of collect1on, as allocable to deposits other than demand deposits
The amounts mvolved are substantial For example,
reqmred reserves on ume and savmgs deposits m June
1975 were more than $10 billion when M 1 as currently
estimated totaled a bit under $300 b1lhon Hence, the
understatement of M 1 on this account 1s a m1mmum of
3 per cent
True, this understatement is not likely to change
much over time, but also the cost of mtroducmg this
correct10n mto the estimates, 1f done as I suggest, 1s
tnvial The reqmred mformat1on is all now available,
so only extra computation 1s reqmred

Conceptual and Definitional Issues

of data This is clearly the case for currency
held abroad, which cannot be estimated separately from that held m the United States With
respect to the other items, current practice
raises two mam issues that we wish to explore
(I) treatment of US Treasury and other Federal Government agencies, and (2) treatment
of deposits of foreigners m the United States
and of dollar deposits of US residents outside the United States

Federal Government
Insofar as the U S Treasury 1s part of the
ultimate monetary authority, its accounts
should be consolidated with those of the Fed
and its holdmgs excluded as part of bank
assets and of M 1 through M 5 The real quest10n arises m regard to the U S Treasury as
an operatmg entity and about other Federal
Government operatmg agencies On the
whole, we recommend the contmued exclus10n, so far as possible, of all U S Government holdmgs of currency and deposits on
two grounds First, it 1s somewhat arbitrary
to separate the holdmgs of the Treasury m its
capacity as an ultimate monetary authority
from its holdmgs as an operatmg agency or
as the fiscal agent for other operatmg agencies Second, even as an operatmg entity, the
relat10n between money balances of the Federal Government and how 1t spends the money
cannot be regarded as homogeneous with this
relat10n for the rest of the public, given the
Federal Government's unique power to create

Foreign deposits in the United States and
dollar-denominated deposits abroad
US currency and deposit liabilities of banks
m the Umted States are held by a variety of
domestic and foreign holders Furthermore, m
recent years there has been a vigorous growth
of what may be labeled "offshore" dollar deposits, that 1s, dollar-denommated deposits
that are hab1lit1es of foreign branches of U S
banks or foreign banks located abroad These
deposits are also held by a variety of domestic
and foreign holders Table 1 portrays systematically these various categories of dollar-
Federal Reserve Bank of St. Louis


denommated deposits as of December 31, 1974,
msofar as the mformat10n 1s available The
table does not mclude US currency, smce no
reliable mformat1on 1s available on the amount
and d1stnbut10n of such currency held domestically and abroad
Components presently mcluded
m the aggregates
As presently defined, M 1 mcludes, first, all
outstandmg US currency outside the Treasury, Federal Reserve Banks, and U S commernal banks Thus, M 1 mcludes currency
held abroad by the foreign public and foreign banks, there bemg no mformat10n for
excludmg this component even 1f 1t were
deemed desirable to do so M 1 mcludes,
second, all demand deposit hab1hties of banks
m the Umted States (mcludmg not only US
banks but also branches and agennes of foreign banks) except those due to the U S Government and to other banks m the United
States (so-called mterbank deposits) It also
excludes nondepos1t liabilities of U S banks
to their related head offices or branches abroad
(column 2 of Table I) that may serve funct10ns s1m1lar to mterbank demand deposits
Thus, as presently defined, M 1 mcludes the
deposits at U S banks of all nonresidents, mcludmg foreign governments and (unrelated)
foreign banks, even though deposits of the
US Government and US banks are excluded
The broader aggregates, too, are defined as
mcludmg the deposits at US commercial
banks of foreign nonbanks, official mst1tut10ns,
and (unrelated) commercial banks and (for M 3
and M 5 ) such savmgs accounts as these orgamzat10ns may have at US savmgs mst1tut1ons
No dollar-denommated deposits at bankmg
offices outside the Umted States, covered m
the nght-hand port10n of Table I on page 17,
are now mcluded m any of the standard US
monetary aggregates
Critena for recommendat10ns
We have taken a fresh look at the current
defimt10n of the aggregates with two questions


Improvmg the Monetary Aggregates Committee Report

m mmd Whether 1t might be appropriate to
change tlus defimt10n with regard to the types
of deposit holders and/or types of issuers to be
mcluded m one or more of the aggregates, and
whether 1t would be desirable to collect add1t10nal mformat10n, on a contmumg basis, for
any types of dollar-denommated deposits
In reachmg our recommendat10n as to which
of the entries m Table I should be mcluded
m the aggregates at this time, we have been
gmded by two mam cntena
1 Would mclus10n of the category make
the U S monetary aggregates more useful to
monetary policymakers concerned with settmg,
1mplementmg, or momtonng policies aimed at
acluevmg final pohcy ob1ect1ves, such as pnces,
output, and employment~
In an open economy (one that has trade and
financial relat10nsh1ps with other economies),
no purely domestic defimt10n of the money
stock can logically be completely satisfactory,
JUSt as the monetary authont1es m such an
economy cannot completely disregard foreign
trade and capital movements m makmg their
stab1hzat10n dec1s10ns How much weight monetary authont1es may wish to put on such foreign transact10ns will presumably depend on
how important these transact10ns are relative
to domestic economic act1v1ty m the country
concerned The weight may also vary accordmg to currently accepted practices regardmg
the stability or vanabihty of exchange rates
The more weight monetary authont1es place
on mternat10nal cons1derat10ns, the less satisfactory any domestic defimt10n of the money
stock will be as an mtermediate target for
monetary act10n aimed at mfluencmg final
ob1ect1ves such as employment and pnces,
smce m an open economy there 1s no way of
neatly separatmg the mteractmg domestic and
mternat10nal effects of monetary policies
Unfortunately, no precise tests are feasible
to determme how useful different concepts of
money are as mtermediate target variables for
monetary policy act10ns m the Umted States
or m other open economies Given the theoretical difficulty of prescnbmg the "ideal" mclus10n of foreign-or mternat10nal-money
m the U S money stock, the practical d1ffi
Federal Reserve Bank of St. Louis

cult1es m obtammg the desired data even 1f
they could be conceptually defined, and the
relatively small role played by mternat10nal
transact10ns m the US economy, as a practical matter we recommend use of a concept of
money focused pnmanly on the domestic economy For the American economy, the followmg operat10nal cntena seem reasonably consistent with a broad spectrum of views about
the nature of the transm1ss10n mechamsm between money and various final economic ob1ect1ve variables, mcludmg the views that
money affects mcome directly and that 1t works
through the !mks of market mterest rates and
the ava1lab1hty of credit
(a) If the demand for dollar deposits by any
group of holders is controlled pnmanly by
forces other than those that control the demand
for money by the U S pubhc (basically the
volume of domestic transact10ns, wealth, and
mterest rates), there rs a pnma fac1e case for
the exclus10n of that group of holders and
holdmgs For example, 1f dollar deposits of
foreign central banks m the Umted States are
controlled by fluctuat10ns m reserves and by
balance of payments cons1derat10ns, rather
than by their associat10n with payments m
U S goods and secunt1es markets, then these
deposits should be excluded from the US
monetary aggregates (b) A closely related entenon is whether some broad measure of U S
mcome rs more closely associated with M 1 defined to mclude or exclude the given component Generally, cntena l(a) and l(b) will yield
the same conclus10n, although tlus may not
always be the case (c) Beyond these tests, as a
practical matter, 1f a category represents at all
times a small and reasonably stable proportion
of the aggregate, its treatment will not have
an important effect on the usefulness of the
resultmg aggregate
2 Costs of data collect10n are relevant, and
such costs should be considered m relation to
the expected usefulness of data collected
Relymg pnmanly on these cntena, the Committee makes the followmg recommendat10ns


Conceptual and 1Jefinitwnal Issues

TABLE I Dollar Liab1ht1es of Banks and Banlung Offices, December 31, 1974
In bdhons of dollars
In the UmtedStates 1

Outside the Umted States

Gross t1me and
savings deposits

habihues to
related bank
offices abroad




c1al banks

To U S restdents
US Govt
To Non US residents s




Commercial banks
Foreign branches of U S banks

' Incl udmg U S



Umted States


branches and agencies of foreign banks

(1 hetr habihties to head offices and other related offices abroad

are 1n column 2

thetr deposit hab1ht1es to vanous holders 1n
columns I and 3 )
2 Includmg US bank branches ID those countries Data exclude 1Dterbanl habihties of one bank to another withm the
same country
3 Data include mterbank habihties of a branch to other banks
withm the same country The breakdown of habiht,es to US
residents bet\\ een banks and nonbanks



' Less than $1 bilhon
'Data sholln m columns I through 3 are from (monthly)
reports to US Treasury for balance of payments statistics Datly
data are also avatlable for demand depos,t habihties to the three
categories of US residents, to foreign official 1nst1tut1ons, and to

Foreign deposits at US domestic banks
As m the case of domestic transactors, one
can d1stmgmsh three types of foreign depositors (1) private nonbanks (IPC), (2)
nat10nal governments and official mstitut10ns,
and (3) commercial banks
Deposits of foreign individuals, partnerships, and corporations We recommend contmuat10n of the present practice of mcludmg
IPC deposits m M 1 and the other aggregates
This recommendat10n rests on a combmat10n
of cntena I and 2 Empirical tests for cntenon
l(a) failed to provide clear support for either
mclus10n or exclus10n of this component
Tests for cntenon l(b)-comparmg the correlat10n of changes m mcome with distributed
lags of changes m M 1 mcludmg and excludmg
foreign IPC deposits as estimated from balance
of payments data-were also not very conclusive lncludmg foreign IPC deposits raises
the correlat10n between M 1 and mcome-suggestmg that these deposits be mcludedthough the difference is small Moreover,
Federal Reserve Bank of St. Louis


Foreign branches of
US banks 3



to BIS'


8 Euro
pean coun






















Foreign banks with agencies

and branches






Official mst1tut1ons


(1Dclud1Dg CD's)
and similar






unrelated foreign banks, but not for demand dcpoS!ts of foreign
SOURCE -Seasonally unadiusted s1Dgle date figures from the
call report for domestic U S banks and U S branches of foreign
banks, from Federal Reserve data (forms FR 886 (a) and (b))
for U S agencies of foreign banks, and Edge Act corporations
engaged m bankmg, from Treasury balance of payments data
for habihttes to related bank offices and for CD's issued to
foreign holders, from daLl of Nat10nal Associat10n of Mutual
Savmgs Banks, Federal Home Loan Bank Board, and National
Credit Umon Admmistrat10n, from BIS Annual Report and
unpubhshed breakdown for Euro dollars held m Umted States
and from rederal Reserve data (form FR 502) for habihues of
foreign branches of US banks

foreign IPC deposits are small (cnterion l(c) ),
m recent years they have accounted for less
than I per cent of M 1 or other relevant aggregates, so handlmg them m the cheapest
way seems appropriate And the cost of ehmmatmg them from domestic IPC deposits on a
daily-average basis would, accordmg to the
staff, be substantial
Deposits of foreign central banks and other
official znstitutwns We recommend that these
deposits of foreign central banks and other
official mst1tut10ns hereafter be omitted from
M 1 and the other monetary aggregates They
are small compared with the relevant monetary
aggregates-such demand deposits are currently about I per cent of Mi, while total deposits (mcludn~g CD's) are about 2 per cent of
M 4 , havmg mcreased sharply m the last 2
years Thus, their treatment is not an issue of
ma1or importance Nonetheless, we feel they
should be excluded, though this recommendat10n does not rest on a simple analogy with the
treatment of the deposits of the U S Government It rests, mstead, pnmanly on cntenon I


lmprovmg the Monetary Aggregates. Committee Report

The term "foreign official" covers a diverse
array of transactors, mcludmg central banks,
governments and their d1plomat1c and consular
establishments, purchasmg m1ss10ns, and mternat10nal orgamzat10ns such as the Umted Nat10ns The deposits of foreign central banks
and exchange stab1hzat10n funds are not used
to any substantial extent for payments m
goods, services, and private capital markets
m the Umted States, though they may be
related to US mterest rates, at times s12able
shifts apparently unrelated to this country's
gross nat10nal product (GNP) occur between
these deposits and foreign central banks' holdmgs of Treasury bills and other securities
For the other foreign official mst1tut10ns,
spendmg behav10r analogous to that of the
U S pubhc is conceivable However, our
empirical tests mdicate that only about threefourths of the movement of foreign official
deposits (as measured by the variance) 1s
accounted for by GNP, personal mcome, and
mterest rates, and even the latter effect 1s not
clear cut, by contrast, those variables explam
99 per cent of the variance for M 1 as now defined Thus, cntenon l(a) pomts margmally
to the om1ss10n of tlus component
Empmcal tests usmg cntenon l(b) pomt
slightly m the opposite direction Inclus10n
of foreign official mst1tut10ns m M 1 mcreased
the correlat10n with mcome, although the
mcrease was very small On balance, then, the
empirical evidence pomts margmally toward
the ehmmat10n of foreign official deposits, and
this conclus10n 1s strengthened by the potential for erratic movements of deposits of
foreign official mst1tut10ns
Foreign commercial banks' deposits m the
United States These claims of foreign commercial banks on banks m the Umted States,
which amounted to $28 billion at the end of
1974, consist of two d1stmct components The
largest-not now mcluded m the monetary
aggregates-mcludes nondeposit habihties of
U S banks to their foreign branches, and habil1t1es of branches or agencies of foreign banks
m the Umted States to their head offices or
other offices abroad (bottom entries m column
2 of Table I) We recommend contmued
Federal Reserve Bank of St. Louis

exclus10n of these habihties on the general
prmciple of excludmg from the aggregates
claims and habihties between branches of the
same banks
The other component consists of deposit habihties of banks m the Umted States to unrelated foreign banks, shown as the last items m
columns I and 3 of Table I Such demand deposits, which are now mcluded m M 1, have
grown rapidly m recent years and represent by
far the largest portion of foreign demand deposits m the Umted States (about 3 per cent of
M 1) Empmcal tests for cntena l(a) and l(b)
failed to produce readily mterpretable results,
partly because of the extraordmary growth
trend of this aggregate These tests did, however, show that the size of these demand balances vanes negatively with short-term U S mterest rates Time deposits of unrelated foreign
banks m the Umted States are small
In an effort to obtam more direct mformat10n on the purposes for which these balances
are typically held and on the factors controllmg
their s12e and movement, we requested the staff
to mterview offinals of ma1or US banks that
hold such deposits and also officials of foreign
banks The respondents were consistent m the
view that foreign commercial banks mamtam
demand deposits at U S banks pnmanly to
clear their Euro-dollar transact10ns, and secondarily to settle foreign exchange transact10ns (of which only a small percentage
directly mvolves the foreign commerce of the
Umted States) Respondents also agreed that
the demand deposits mamtamed m the Umted
States by foreign commercial banks are mamly
compensatmg balances, that is, they are mamtamed at the level reqmred to compensate the
This general prmciple is obviously applicable m the
case of two domestic offices of the same bank However,
smce banking offices abroad can be regarded as belongmg to the national banking systems of the countries m
which they are respectively located-and also since an
mcrease or decrease m an mternatlonal claim implies
an mternauonal capital flow-it can be argued that this
general principle 1s not automatically extendable to the
mternational field Alternatively, one can start from a
presumption that defimt10ns of national money stocks
should be consistent with each other and with the defimtion of a world money stock and logically argue for
exclusion of all mternational mterbank claims-those on
unrelated as well as on related bankmg offices

Conceptual and Definitional Issues

U S bank for services rendered to that account Some (notably Japanese) banks were
reported to hold compensatmg balances
agamst Imes of credit, with the deposits play!
mg essentially the role of commitment fees
This mformat10n provides a plausible explanat10n for the rapid growth of foreign
commercial banks' demand deposits m the
Umted States durmg the past decade It also
is consistent with the responsiveness of such
deposits to mterest rates, noted earlier
On the basis of this limited evidence, we conclude that the usefulness of M 1 and other aggregates would, on the whole, be improved 1f
foreign commercial banks' demand deposits m
the Umted States were deleted But we recommend that data on such deposits be published
monthly to permit analysts who wish to do so
to mclude them m the money stock
We recommend a similar exclus10n from M 2
and the broader aggregates (and similar separate publicat10n) of foreign commercial banks'
time deposits m the Umted States This treatment appears desirable for consistency with the
exclusion (as recommended above) of the
claims of related banks abroad, a large part of
which are well known to be mterest-bearmg
placements of funds analogous to mterbank
time deposits

Dollar deposits at banks zn
foreign countries
One of the most strikmg developments m
the mternat10nal capital markets m the past
decade has been the spectacular growth m dollar-denommated depasits at banks outside the
Umted States, mcludmg the foreign branches
of U S banks These deposits are frequently
referred to as Euro-dollar deposits because they
were first issued by European banks, though
the termmology is mISleadmg because such deposits are now accepted by banks at a variety of other locat10ns
Unfortunately, existmg mformat10n about
these deposits is limited The right-hand
port10n of Table 1 summarizes the available
mformat10n on offshore dollar liabilities-at
bankmg offices located m eight European coun
Federal Reserve Bank of St. Louis


tries and at U S bank branches abroad, mcludmg Europe (overlappmg with the other group)
and other areas The Euro-dollar liabilities of
the eight countries have grown more than 16fold m the past decade, and at the end of 1974
amounted to about $165 bill10n Tlus total mcludes interbank dollar deposits by commercial
banks outside the country of a reportmg bank
and all dollar deposits of nonbanks and central banks (Interbank deposits m dollars
within the London market are thus omitted)
The total amount of offshore dollar deposits,
mcludmg the London mterbank liabilities and
also deposits m Canada, Japan, the Bahamas,
Panama, and Smgapore, was near $300 billion
at the end of 1974, accordmg to some estimates
All Euro-dollar deposits are mterest bearing
and they resemble negotiable CD's issued m
the Umted States m that transact10ns occur
only m large amounts The bulk of the liabilities are to other banks Of the $165 billion
Euro-dollar total, for example, four-fifths was
mterbank More than nme-tenths of the $165
billion was due to asset holders outside the
Umted States
Informat10n on the maturity distnbut10n of
Euro-currency liabilities is published from time
to time by the Bank of England for the liabilities of banks m the Umted Kmgdom denominated m dollars and other non-sterlmg currencies (Dollar liabilities make up about fourfifths of these) Data for February 19, 1975, are
summarized m Table 2 As column 4 shows,
more than half of the London Euro-currency
habihties to nonbanks outside Britam have
remammg maturities of 1 month or longer
(The proport10n of outstandmg deposits havmg original maturities of 1 month or longer
would of course be greater)
Among the various types of offshore dollar
deposits, those that deserve closest consideration as candidates for mclus10n m some U S
aggregate such as M 4 or M 6 are the deposits
of U S residents As Table 1 shows, at the end
of 1974 there were $6 billion of such deposits
outstandmg at foreign branches of U S banks
and $1 bill10n at other European banks There
was also some unspecified quantity at foreign
banks outside Europe, of wluch the largest part


Improvmg the Monetary Aggregates Committee Report

TABLE 2 Matunty D1stnbutJ.on of UK Banks' Non
sterling, February 19, 1975






B1lhons of dollars
(and dollar eqmvalents)

39 2

97 3


19 2

Percentage d1stnbut1on

Less than 8 days
8 days to
less than 1 month
I month to
less than 3 months
3 months to
less than 6 months
6 months to
less than 1 year
1 year to
less than 3 years
3 years and over

15 4

22 4

50 2

19 7

17 2

19 4

19 0

32 2

28 9

14 5

24 6

20 3

18 2


14 5





100 0

100 0

100 0

100 0

27 6

-Bank of England Quarter/" Bulletin, June 1975

is undoubtedly the U S dollar deposits m
Canada of U S residents other than banks,
which were $2 billion at the end of 1974, accordmg to published Canadian statistics These
amounts are not negligible, but m total they
still represent only about 1 per cent of M 4
It is possible that some of these deposits are
related to domestic operat10ns but are held
abroad because of the higher yield m cash or
services thus obtamable There are, however,
reasons for doubtmg this hypothesis-at least
with respect to the port10n held at foreign
branches of US banks, which is the ma1or part
of the total Smee 1969 it has been the stated
pohcy of the Federal Reserve Board of Governors that U S banks should not accept deposits from U S residents at overseas offices
unless ~uch deposits are kept abroad for a defimte mternat10nal purpose It is known that
proceeds of Euro-bond issues by U S corporat10ns to finance operat10ns abroad have at
times been held m Euro-dollar deposits pendmg disbursement When the Federal Reserve
staff a few years ago mterviewed finance officers
of several large corporat10ns with extensive
mternat10nal operat10ns, the respondents reported almost unammously that they did not
Federal Reserve Bank of St. Louis

keep abroad balances that were to be used for
transact10ns m the Umted States
On the basis of these considerat10ns, we recommend, for the present, contmuat10n of the
current practice of not mcludmg m US monetary aggregates the dollar deposits held at
banks abroad by nonbank US residents We
do recommend, however, that data on such
deposits be collected and published monthly
and that a breakdown of such deposits by
maturities be obtamed and published
As to IPC dollar deposits of foreign businesses and mvestors held at foreign banks, we
have no satisfactory evidence on owners or the
purposes for wluch such deposits are held Undoubtedly the holders mclude some foreign
affiliates of US corporat10ns, but lt is likely
that these and other holdmgs are not sigmfican tl y related to transactions m the Umted
States Thus, we recommend that they not be
mcluded m the US monetary aggregates
Some recent research suggests that Eurodollar deposits, as components of a world monetary aggregate, play some role m determmmg
world output and pnces Moreover, as we emphasized above, m any open economy the hne
between money held for domestic and mternat10nal transact10ns is not a clear one Thus, we
recommend that Euro-dollar deposits be momtored closely, that the summary Euro-dollar
data presently collected on a quarterly basis
by the Bank for Internat10nal Settlements for
banks m Europe, Canada, and Japan be supplemented periodically by data for banks m
other areas, that the quarterly data be broken
down by classes of holders (nonbanks, commercial banks, and central banks), and that
further mformation be gathered on the maturity composit10n of the deposits of nonbanks
This recommendat10n will permit further
study of the data and their relat10nship to
both US and mternat10nal economic activity
Fmally, on the general prmciple that monetary aggregates should measure the stock held
by the nonbank public, we see no reason to
mclude any mterbank offshore dollar deposits
m the US aggregates, or, for that matter, m
world monetary aggregates



Measurement Issues and Recommendations

How M 1 1s constructed now 4
M 1 , the measure of narrow money stock, 1s
defined as currency m circulat10n plus private
demand deposits ad1usted at all commercial
banks The components of this measure are
shown m Table 3

The currency component of M 1 1s relatively
easy to construct It 1s defined as all currency
and com outside the Treasury, Federal Reserve Banks, and commercial banks Dally
data on currency m orculat10n outside the
Treasury and the Federal Reserve System are
available from daily statements issued by the
Treasury Department Smee member banks'
vault cash can be used to meet reserve reqmrements, member bank holdmgs are mcluded
on member banks' reports subm1tted to the
Federal Reserve for determmat10n of member
bank reqmred reserves Consequently, the
amount of vault cash held at member banks
1s available on a daily basis, but vault cash at
nonmember banks must be estimated
Estimates of vault cash held at nonmember
banks are based on the ratio of vault cash of
nonmember banks to vault cash of memoer
banks on call report dates Currently these
benchmark relat10nsh1ps are available four
times each year for smgle days Pnor to
March 1973, call report benchmark data were
available only twice a year-on June 30 and
December 31
Estimates of the vault cash rat10 for each
week between call report dates are based on
a straight-lme mterpolat10n Weekly estimates

Th1s section 1s based directly on a statement pro
v1ded by the Board's staff
Federal Reserve Bank of St. Louis

of nonmember vault cash are then denved
by muluplymg the estimated weekly vault
cash rat10 times the reported weekly-average
member bank vault cash Monthly-average
vault cash 1s denved from a prorat10n of the
weekly estimates Beyond the latest call report
penod, the current call report rat10 1s held
constant until another call report 1s available
It should be noted that the defimt10n of
"currency m circulat10n outside the Treasury
and Federal Reserve banks, less vault cash
held at commercial banks" mcludes an unknown amount of currency held m safe deposit boxes, sent out of this country, or lost or
destroyed Thus, the measure overstates the
true amount of currency 1n orculauon m the
Umted States

Member bank demand deposits
Data for the demand deposit component of
of the money stock are not so readily available as for the currency component The demand deposit component must be constructed
from a number of different sources These
sources include data available each day, as of
a smgle day once a week, as of a smgle day
once a month, and as of a smgle day on call
reports available four times each year
As shown m Table 3, by far the largest part
of the demand deposit component of the
money stock 1s demand deposits at member
banks The basJC source of data on member
bank deposits is the report of deposits submitted by member banks for determmat10n
of reserve reqmrements Unfortunately, because the purpose of this report is to measure
deposits sub1ect to reserve reqmrements and
not deposits to be mcluded m the money
stock, the deposit breakdowns available from

Improvmg the Monetary Aggregates Committee Report


TABLE 3 Construction of M1
Amounts m IDllhons of dollars, monthly averages, not seasonally adiusted
Line, item

I Currency ID c1rculat10n
Less Member bank vault cash
Nonmember bank vault cash
Equals Currency component of Af 1
5 Demand deposits at member banks'
6 \ Less F R float
Demand deposits at nonmember banks'



to M,,
December 1974

Source of data
Daily data reported by F R Banks and Treasury Dept
Daily data reported by all member banks
Estimated, based on member banks and call report data


Daily data reported by all member banks
Daily data reported by F R Banks


Estimated, based on daily data reported by small member
banks and call report data

CIPC associated with foreign agency and branch


Daily data reported by foreign related mstltut10ns m
New York City

Demand deposits due to foreign commercial


Demand deposits due to mutual sav1Dgs banks


Estimated, based on smgle day (Wednesday) data for
large banks and call report data for other banks
Estimated, based on smgle day (Wednesday) data for
large banks and call report data for other banks

Demand deposits due to banks ID terntones
and possessions
Af 1 type balances at foreign related mst1tutions
ID New York City
Deposits due to foreign official 1Dslltut1ons at
FR Banks
Equals Demand deposits component of M,


Money stock (M,) - currency plus demand deposits


Estimated, based on call report data


Estimated, based on last Wednesday of month reports


Daily data reported by F R Banks

Gross demand deposits less demand deposits due to the U S Govt , 1Dterbank deposits, and CIPC See text for explanation

this report do not match the deposit defimtlons that are needed for the money stock
Consequently, a number of ad1ustments must
be made to the basic data reported by member banks
M 1 does not mclude demand deposits due
to the U S Government nor demand deposits
due to domestic commercial banks These
items must, therefore, be deducted from gross
demand deposits reported on the reserve reqmrements reports as a first step m determmmg the demand deposit component of the
money stock This causes no problem smce
deposits "due to the US Government" and
"due to all commercial banks" are shown as
separate categories on the report of deposits
To av01d double countmg of demand deposits that are simultaneously shown on the
books of two banks at the same time, CIPC's
are also deducted from gross demand deposits
to derive the demand deposit component of
M 1 CIPC's are allowed as a deduct10n Item
m the computat10n of deposits sub1ect to
reserve reqmrements and are therefore available on a daily basis from the report of
deposits CIPC's shown on this report, how-
Federal Reserve Bank of St. Louis

ever, are not broken down by those cash items
associated with private demand deposits and
those cash Items associated with all other
operat10ns of the bank, only a gross cash
item figure 1s available
From past mvest1gat10ns 1t 1s known that
gross CIPC's overstate those items that should
properly be deducted from money stock deposits For example, cash Items associated with
mterbank deposits, with US Government
deposits, with redeemed coupons of US Government securities, and with bank credit cards
are mcluded m the gross cash items data It
1s believed, based m part on past mvest1gat10n
and m part on contacts with bank accountants, that the size of the d1stort10ns noted
above are not large, and further that these
d1stort10ns remam a fairly constant proport10n of total deposits Therefore, while the
level of the money stock may be distorted
slightly, money stock growth rates are probably not affected m a s1gmficant way because
of the overstatement of cash items from these
sources 5

Th1s note appears on opposite page

Measurement Issues and Recommendations

The subtotal denved from these first subtract10ns for member banks, as shown m
Table 3, is defined as gross demand deposits
less demand deposits due to the US Government and all commercial banks less CIPC's
That item is not the demand deposit component of the money stock, however, smce still
other ad3ustments are necessary to obtam
statistical estimates of demand deposits as a
component of M 1
Federal Reserve float is very similar to
CIPC's and is also deducted from pnvate demand deposits adjusted to move toward the
demand deposits component of M 1 FR float
is deducted because on some items cleared
through Federal Reserve Banks, credit is
passed to the sendmg bank before the paymg
bank has received the item and reduced deposits When the sendmg bank receives credit,
the CIPC's are reduced on that bank's books
even though deposit liabilities on the books
of the paymg bank have not been reduced
The amount of this double countmg is reflected m the float created by Federal Reserve
Banks Therefore, a deduct10n for float is
made to offset this double-countmg effect
Daily float data can be denved from the
dally reports of condit10n submitted to the
Board by each Federal Reserve Bank While
float can fluctuate widely from day to day, 1t
is a relatively small component of M 1 and
averaged about $2 7 billion m December 1974

Nonmember bank demand deposits
The second largest deposit component of
the domestic money stock is the domestic non6A much more senous problem, and one that will be
discussed m more detail later, concerns a sigmficant
portrnn of the CIPC's related to mterbank transfer of
funds associated m large part with the cleanng of
Euro dollar transact10ns m the New York City money
market These cash items should not be deducted from
money stock deposits, because the deposits to which
they apply are not part of M1 Therefore, smce they
are mcluded m total cash items, an estimate of their
amount 1s added back through a special ad1ustment,
item 8
Federal Reserve Bank of St. Louis


member bank component, "demand deposits
at nonmember banks" m Table 3 Data for
nonmember banks are available four times a
year from call reports In order to estimate
the deposits of nonmember banks for other
penods, the rat10 of the nonmember bank
demand deposit component of M 1 to that of
the smaller member banks' demand deposit
component of M 1 1s computed on each call
report date A straight-lme mterpolat10n of
tlus rat10, ad3usted for bank structure changes, 6
1s made between call report dates The weekly
rat10s so denved are then applied to weeklyaverage deposits data reported by smaller
member banks, m order to obtam weekly and
monthly average estimates of the demand
deposit component of the money stock at
nonmember banks (Monthly average estimates are denved by proratmg the weekly
estimates ) Beyond the penod of the most
recent call report, rat10s are estimated based
on a regress10n equat10n and 3udgment As
new call report data become available, these
nonmember bank estimates are revised and
"benchmarked" to the umverse data available from the call report

Further adjustments
A s1gmficant part of gross CIPC's (deducted
to obtam demand deposits at member banks
m Table 3) 1s related to the transfer of
mterbank funds related to Euro-dollar transact10ns 1n the New York money market and
should not be deducted from the deposits
properly mcluded m M 1 These mterbank
fund transfers create mterbank demand deposits, deposits not mcluded m the money
stock measure Smee cash i terns generated
from the transfer of Euro-dollar funds are
not associated with money stock deposits,
their deduction from money stock deposits
would cause an understatement of the level
of the M 1 senes And, if these cash items were
"Bankmg structure changes reflect shifts m me)llbership status, mergers, hquidatrnns, and the like


Improvmg the Monetary Aggregates Committee Report

growmg rapidly relative to total deposits (as
they apparently have m recent year~) the
growth rate of the senes would also be distorted
In order to adjust for the CIPC's associated
with mterbank transactions m New York City,
data reflectmg the volume of these transfers
are collected from Edge Act corporations,
agencies and branches of foreign banks operatmg m the Umted States, and other foreignrelated mst1tut10ns m New York City These
data are used as a proxy measure for the
amount of cash items that are recorded on the
books of member banks and improperly deducted from member bank demand deposits
The deduct10n 1s improper because these cash
items have no correspondmg money stock
habih ty on the books of the reportmg banks
These data are available on a daily baSIS and
are reported to the Federal Reserve Bank of
New York In December 1974, as shown m
Table 3, this adjustment amounted to more
than $3 5 b1ll10n
The money stock as currently defined does
not mclude demand deposits due to domestic
commercial banks but does mclude demand
deposits due to foreign commeroal banks,
mutual savmgs banks, and banks m U S
terntones and possessions In order to obtam
items 5 and 7 (demand deposits at member
and nonmember banks), demand deposits due
to banks were subtracted from gross demand
deposits Deposits due to banks as reported on
the report of deposits, however, mclude not
only deposits due to domestic commercial
banks-which must be subtracted from gross
demand deposits to get the demand deposit
component of the money stock-but also
deposits due to mutual savmgs banks, foreign
commercial banks, and banks m terntones
and possessions 7 But the last three items, smce
they are defined as part of the money stock,

lt should be noted that deposit hab1ht1es of banks in
US terntones and possess10ns are not part of the
money stock It 1s deposit liab11It1es of banks m the
Umted States to banks m US terntones and possessions
that are part of the money stock
Federal Reserve Bank of St. Louis

should not be deducted from gross demand
deposits Thus, estimates of the deposits due to
these mst1tut1ons must be denved from other
data sources and added back to demand depos1 ts m order to obtam the demand deposits
adJusted component of M 1
The bulk of deposits due to foreign commercial banks and to mutual savmgs banks
1s held at weekly reportmg banks These
banks report full balance sheets each Wednesday-mcludmg deposits due to foreign
commercial banks and mutual savmgs banks
These smgle-day data are used as a proxy
measure for the weekly-average level of such
deposits at the weekly reportmg banks
Monthly-average estimates are based on prorat10ns of the weekly-average estimates Estimates of deposits due to foreign commeroal
banks and to mutual savmgs banks at nonweekly reportmg banks are denved from call
report data Between call report dates these
deposits are estimated on the basis of a
stra1ght-lme mterpolation Estimates of these
deposits for December 1974 are shown m
Table 3
Demand deposits due to banks m terntones
and possess10ns must be estimated differently
In order to estimate this component, 1t 1s
necessary to make a special tabulat10n of the
call report showmg balance-sheet data for
banks located outside the Umted States, sometimes referred to as "other areas" Included
m this tabulat10n is an item on the asset
side "demand deposits due from US banks"
This item is assumed to be eqmvalent to the
demand deposits due to banks m terntones
and possess10ns that are mcluded m demand
deposits due to banks on the books of US
commercial banks, it 1s used as a proxy measure for that item As shown m Table 3, the
amount of such deposits is small
In add1t10n to demand deposits at domestic
commercial banks, Mi-type deposits at Edge
Act corporat10ns, agencies and branches of
foreign commercial banks, and other fore1gnrela ted mstitutions are mcluded m the money
stock For reserve reqmrement purposes, Edge
Act corporations must file a report of deposits

Measurement Issues and Recommendations

on a weekly basis, as member banks do These
weekly reports, showmg darly data, are the
source of the Edge Act component of M 1
From 1972 through early 1975, estimates of
M 1 deposits at agencies and branches of foreign banks and other foreign-related mst1tut10ns were based on last-Wednesday-of-themonth reports filed by these mst1tut10ns Smee
Apnl 1975, darly deposit data have been
collected from each of these mst1tut10ns by
the New York Federal Reserve Bank 8
The final component of the money stock 1s
demand deposits of foreign central banks and
other official mstitut10ns at Federal Reserve
Banks This amount 1s shown dally on the
Federal Reserve Banks' daily statements of
conditron The amount is usually relatively
small (about $500 m1lhon m December 1974)

Very short-run (transitory) variations
in the monetary aggregates
Apart from seasonal and longer-term movements, the monetary aggregates are subject to
substantial short-run (transitory) vanat1ons
for two mam reasons Frrst, prelimmary senes
often exh1b1t vanatrons that are smoothed as
more complete data become available from
different financial mst1tut10ns that report at
different mtervals, and as reportmg errors are
corrected Second, unsystematic vanat10ns occur-from day to day and over slightly longer
penods-m payments among the public, the
Treasury, and banks, and other transitory
vanatrons are caused by reportmg errors and
delays of items m transit
These transitory vanat10ns tend to average
out over the longer run, but they often produce consequential vanat10ns m the weekly,
monthly, and even quarterly data It rs possible to estimate statrstrcally the size of such
"Pnor to 1972, M1 deposits at these foreign-related
mst1tut10ns were estimated on the basis of call reports
and monthly reports filed with the New York State
Bankmg Department In some cases back data were not
readily available, and estimates had to be made based
on end of-year call report data and other mfonnat10n
that could be gathered by the staff of the Federal
Federal Reserve Bank of St. Louis


vanat10ns We recommend that the Fed regularly publish the range of revlSlons m data
and uncorrected transitory vanatrons-to
warn users of the pubhshed aggregates agamst
unnecessarily confusmg the systematic part
of observed changes m the monetary aggregates with transitory short-run vanat10ns
Published money stock data that may appear
to represent large changes m growth rates
often m fact reflect only transitory changes

Revisions of preliminary estimates of
the money stock
Data on the money stock are compiled from
reports of banks and other financial mst1tut1ons that become available with varymg
delays A prehmmary estimate for each month
rs published about 10 days after the end of
the month As add1t1onal reports are received,
new estimates are substituted for the mrtial
estimates, and a revised M 1 figure 1s published
about 3 weeks after the end of each month
Tlus figure is sub1ect to add1t1onal rev1s10ns
when call report data for nonmember banks
become available, when the seasonal ad1ustment rs revised, or when special correct10ns
are made retrospectively (for example, that
m 1970 to ehmmate a cash item bias for
previous years)
These successive rev1s10ns can be sIZable
Table 4 summarizes the successive revis10ns
m M 1 smce 1968, showmg monthly, quarterly,
and annual estimates The differences shown
for monthly estimates, for example, are for
(1) the first rev1S1on (about 3 weeks after first
pubhcat10n) mmus the mitially published
estimate made 20 days earlier, (2) the final
estimate as of August 1975 (mcludmg benchmarkmg for nonmember bank data, latest
seasonal ad1ustment, and other special corrections) compared with the first revIS1on, (3)
the final estimate compared with the mrtial
estimate, and (4) the total rev1s10n due to
changes m the seasonal ad1ustment only The
final estimates used here, particularly those
for 1974 and 1975, may be revised strll further for seasonal adjustment and special cor-

Improvmg the Monetary Aggregates Committee Report


TABLE 4 Differences ID Estrmates of Growth ID M1 Between Imtially Published,
First, and FIDal Revmons, January 1968 to August 1975
Annnal rates of change m per cent

Difference between -

Successive monthly averages
I First rev1s1on minus 1n1trnl estimate
2 Final estimate minus first rev1s1on
3 Final estimate mznur 1n1tial estimate
4 Lme 3 due to seasonal adjustment•
Monthly averages a quarter apart
I First rev1s10n minus 1n1t1al estimate
2 Final estimate minus first re,1s10n
3 Final estimate minus 1n1ual estimate
4 Lme 3 due to seasonal adjustment•
Monthly averages a year apart
1 First rev1s1on minus 1n1tial estimate
2 Final estimate minus first rev1s1on
3 Final estimate minus 1n1t1al estimate
4 Lme 3 due to seasonal adjustment•



Range (±) w1thm
which 95 per
cent of rev1s10ns
m growth rates
will fall•


2 90
3 20




I 54
I 73
I 26

I 3






Assumes differences are normally distributed around the mean
Estimated from the 1mphed seasonal factors for the total money stock used at the time of
the mmal pubhcat10n and m August 1975, as applied to the final estimate of the unad.Justed

rect10ns, so the differences on the second,
third, and fourth Imes may not be entirely
accurate The differences shown for the period
as a whole, however, are reasonably mdicative
of the magnitude of successive revis10ns
The comparable differences for quarterly
and annual data are smaller than for monthly
data because most of the rev1s10ns pertam to
only the later month and not the base month
of the changes shown The mean differences
throughout do not equal zero, as would rev1s10ns for unbiased errors m est1matmg
unavailable items, because the benchmark
rev1s10ns and the special correct10ns have
tended upward over this period As 1s apparent from the standard dev1at10ns, both first
and final rev1s10ns have been substantial,
primarily because of revIS1ons of the seasonal
ad3ustments and of benchmarkmg when late
data for nonmember banks become available

Unsystematic day-to-day variations
The monetary aggregates exhibit systematic
movements attributable to mtraweekly and
seasonal fluctuations plus the more basic
movements mtroduced by Federal Reserve
policy After allowances are made for the
systematic movements, unsystematic (transitory) variations remam m the data To assess
Federal Reserve Bank of St. Louis

the importance of these unsystematic vanat10ns, the Committee requested the Board's
staff to make estimates based on a simple
analysis of variance The staff also explored
more soplusticated methods of estimat10n
noted later, which do not reqmre the same
assumpt10ns as the analysis of variance
In the imtial analysis of variance estimates
it was assumed that systematic mtraweekly
variat10ns are the same for all weeks m a
year (this 1s eqmvalent to no mteract10n between days and weeks) and that unsystematic
day-to-day variat10ns are not serially correlated The analysis begms by takmg the
difference between the money stock each day
and the average level for the week The average of these differences over the year for
Mondays, Tuesdays, and so on gives the average mtraweekly vanat1on Then these average
differences are subtracted from the correspondmg difference for each day of the year The
subtraction gives the daily residuals from the
average mtraweekly vanat10n and the week's
average The standard deviat10n of these
residuals 1s an estimate of the day-to-day
variat10n m the money stock
The standard deviat10ns of the residuals
for demand deposits, currency, time deposits,
and two aggregates are shown m Table 5 for
the separate years 1968-74 and for the full


Measurement Issues and Recommendations

TABLE!, Standard Deviation of Day-to-Day Vanation
m Monetary Time Series

devrnt10n (square root of the variance) for the
monthly growth rate 1s

Percentage of series level
Demand Currency










TABLE 6 Analysis of Variance of Log, M,., 1968-74
In thousands of dollars
Source of
Days of week




Sum of
28 07335
28 12978





2460 295
85 76026

Mult1phcat1ve day effect
Mon ay

I 00334
I 00161

1 Degrees of freedom
Standard deviation of residual m per cent 1s


v 00003143 X 100 = 561
per10d The ongmal data were converted to
natural loganthms m order to express the
standard devrnt10n as a percentage of the level
of the monetary data The analys1s of vanance
on wl11ch Table 5 1s based is 1llustrated for
the 1968-74 figure for M 1 m Table 6
From these standard devrnuons we can
calculate the vanat10n m money growth rates
due to the day-to-day variat10n On the assumpt10n of no serial correlat10n, the variance
of the residual variation m a 5-day weekly
average would be one-fifth of that m I-day
figures But the weekly average as currently
calculated 1s a 7-day average with the Friday
figure counted three times For this 7-day
average the variance is

V (¾) !~

(and then multiplied by 12 to express m terms
of annual rates) and for the rate between
successive quarterly averages (13 weeks m
each quarter) 1s

V (1D

(and then muluphed by 4 to express m terms
of annual rates)
These and similar formulas were used to
derive the implied transltory vanat10ns m
growth rates m Table 7 On the average,
roughly 95 per cent of all growth rates as
measured will be w1thm two standard devrnt10ns above or below the systematic component For M 1 this range is +5 percentage
pomts for month-to-month growth expressed
as an annual rate, + 1 percentage pomt for
growth between success1ve quarterly averages
expressed as an annual rate, and +0 1 percentage pomt for year-to-year growth These
ranges are about one-half as large for M 2
Given the usual magmtude of money growth
rates of about 3 to 8 per cent per year, only the
quarterly and yearly rates are reasonably mdicat1ve of systematic movements m the aggregates The quarterly and yearly rates have a

where u is the standard devrnt10n of the daily
For a month (assumed for s1mphc1ty to be
exactly 4 weeks) the variance 1s one-fourth as
large Thus the formula for the standard
Federal Reserve Bank of St. Louis



TABLE 7 Vanation m Monetary Growth Rates Due to
Transitory Fluctuations'
Annual percentage rate


growth rates


Successive monthly averages


Successive quarterly averages


u= 364u


Successive annual averages


Monthly averages
A quarter apart -

Range (±) w1thm
which 95 per cent
of growth rates
WIii fall

2 45

I 14














A year apart -


1 u for M 1 1s 561 and for M, 1s 261


Improvmg the Monetary Aggregates Committee Report

smaller range of variat10n, m part because the
rates for the longer penod average out more
of the day-to-day variatxon and m part because
there 1s less blow-up m annuahzmg these
growth rates
These ranges of vanat10n are estimated to
be somewhat smaller when the serial correlatxon still remammg m the daily residuals of
the analysis of variance 1s removed This serial
correlat10n reflects systematic short-run vanat10ns that last longer than a day but less than
a week and systematic vanations due to holidays and other particular days of s1gmficance
m the payments and clearmg process Estimates made by the Board's staff of these mfluences on the money stock account for
one-fourth of the variations shown m Tables
5 and 6 Consequently, the range of variat10ns
shown m Table 7 would, accordmg to these
estimates of the residual vanat10ns, be reduced by a quarter The danger exists, however, that these more wph1st1cated methods
of est1matmg the residual vanatxon may overstate the systematic component of movements
m the aggregates
While we recommend that estimates of the
range of transitory variat10n m the various
money growth rates be published regularly,
we are not prepared to recommend one specific method of estimation as necessarily best
among the reasonable alternatives available

Averaging of daily data
Weekly money stock data are now calculated as an average of 7 days Smee most
banks are closed on Saturday and Sunday,
they report the precedmg Friday figure for
Saturday and Sunday The Friday figure receives a weight of 3 /7 and each of the other
4 days of the week a weight of l /7 The Committee discussed the advisabihty of an alternative weekly average based on 5 days, m which
each day Monday through Friday would receive a weight of 1/5-on the ground that this
might reduce transitory vanab1hty m the
monetary aggregates and brmg them mto
closer correspondence with the pubhc's spendmg dec1s10ns
The day-to-day vanabihty 1s larger m the
Federal Reserve Bank of St. Louis

7-day average As shown above, the standard
13 of the
deviation of this variab1hty 1s
standard deviatxon of the day-to day variatxon,


while m the 5-day average 1t 1s


Hence a

5-day weekly average mstead of a 7-day average would reduce this variabihty by a small

(V!~ _V~ ) =


There would be a correspondmg reduction of
such vanabihty m the growth rates The _reductxon 1s slight, but the necessary computat10nal cost 1s also small
Although 1t would be desirable to reduce
tlus source of vanab1hty, the ch01ce between
7-day and 5-day weekly averages also depend~
upon the appropriate treatment of the systematic component m the data A 7-day
average appears appropriate 1f the pubhc
takes 1ts Saturday and Sunday holdmgs of
M 1 mto account m dec1dmg whether to purchase goods and services or to acqmre other
financial assets In view of the fact that
deposits cannot actually be transferred on
Saturdays and Sundays (except that many
banks are n,ow open on Saturdays for makmg
deposits and withdrawals), 1t might appear
that the weekend amounts are not a part of
transact10ns balances (that 1s, Fnday balances
count once) and so do not affect spendmg
Yet they may affect the holder's assessment of
the average amount of his money balances
over a week or a month and m that way mfluence his spendmg declSlons
The Committee did not find that a clear
case could be established, or that the potential
benefits would be sufficiently large, to JUSt1£y
changmg the standard practice of averagmg
the monetary aggregates from a period of 7
days to a 5-day period

Nonmember bank deposits
The estimat10n of demand deposits adJusted
for commercial banks that are not members of
the Federal Reserve System has been a particularly significant source of error and uncertamty
m current statistics of M 1 Wlule nonmember

Measurement Issues and Recommendations

deposits account for only about one-fifth of
the money stock, large rev1s10ns of early estimates of M 1 have repeatedly been reqmred
because of revis10ns m the nonmember bank
deposit senes Even after rev1S1on, the Iustoncal estimates for nonmember banks are
much less reliable than those for member
banks, except for two to four call report days
each year
The problem arises because reportmg by
nonmember banks is so mfrequent, the umverse reports on deposits are for only 2 days
each year-call reports at the end of June
and December-although FDIC-msured banks
also report on one day m sprmg and one m
autumn Moreover, about 4 months elapse
between the date of each call report and the
availability of processed mformat10n from the
report that can be mcorporated mto money
~tock estimates
Usmg histoncal mformation on nonmember
bank deposits for those mfrequent reportmg
dates, the Federal Reserve staff has found
that the rat10 of such deposits to those of
smaller (country) member banks can be approximated roughly by a regress10n relat10nship that uses, as explanatory variables, lmear
and quadratic time trends and the 90-day
Treasury bill rate (the sigmficance of the
latter bemg that nonmember banks are not
affected m the same way as country member
banks by changmg credit market condit10ns)
When any weekly or monthly estimate of M 1
is mitially made-say, monthly for Julythe nonmember bank demand deposit component is simply extrapolated from the latest
available reported nonmember bank data,
which at that time would be for the last day
of December, usmg that regress10n estimate
(with a Judgmental ad1ustment) and the
known total for country member banks m
Experience has demonstrated that these
extrapolat10ns often produce substantial errors, m recent years they have averaged more
than $1 billion When the actual data for
msured nonmember banks for the next call
report date-which, m the example, would
be durmg the spnng-become available about
August, a benchmark revis10n is made Some
Federal Reserve Bank of St. Louis


of the recent revis10m m M 1 reqmred by the
call report data have been especially large,
mcludmg ones of $1 7 bill10n (upward) for
mid-1973, $1 2 billion (downward) for mid1974, and $2 4 billion (downward) for sprmg
1975 Moreover, use of the current-or any
alternat1ve-extrapolat10n procedure might
become sub1ect to even greater errors m the
event of more structural changes m the bankmg system
Once the spnng call report benchmark data
are available, the level of nonmember bank
deposits for the entire 8-month mterval srnce
the precedmg report date m December must
be mterpolated to match a new estimated
ielat10nship of the rat10 of nonmember bank
demand deposits to country-member bank
demand deposits usmg the sprmg call data
Moreover, M 1 and related aggregates are estimated as weekly or monthly averages Thus,
because the "final" nonmember call report
data cover only a smgle day, some assumption
must be made as a basis for estimatrng the
weekly average of nonmember bank demand
deposits, even for the week of the call report
It is currently assumed that the rat10 of the
weekly average for that week to the known
I-day figure for nonmember banks is the same
as it is for country member banks
The same procedures and thus the same
kmds of problems apply to the estimat10n of
nonmember bank time deposits for mclus10n
m M 2 In practice, however, the size of benchmark revis10ns for time deposits has been
significantly smaller, although far from tnvrnl
For a penod from summer 1974 to sprmg
1975, the FDIC asked (although 1t did not
reqmre) a sample of 573 rnsured nonmember
banks to report dally aggregates of selected
balance sheet items, and it transmitted summary mformat10n from tlus sample to the
Federal Reserve,' although with a lag of at
least 2 months The sample was stratified by
size, mcludmg (m prmciple) universal coverage of about 177 nonmember banks with total
deposits rn excess of .$100 mill10n (large nonmember banks) Explorat10ns with the sample
data pomt to several revealmg conclus10ns
First, under present procedures the problem
of mferrrng a weekly average from a 1-day


lmprovmg the Monetary Aggregates Committee Report

report is itself s1gmficant Day-to-day vanat10ns m demand deposits are sizable When
the ratio of weekly-average deposits to callreport-day deposits for the sample (mstead of
the correspondmg weekly-average to 1-daydeposit rat10 of country member banks) 1s
used to convert the I-day figure for the umverse of nonmember banks to a weekly average, the estimates differ by amounts between
$200 million and $1 b1lhon There 1s every
reason to beheve that the estimate based on
the sample 1s more reliable These discrepancies produce permanent errors m the h1stoncal
senes on M 1 as now constructed
Second, demand deposits of small nonmember banks behave differently from those of
large nonmember banks, hence, the sample
data for the large nonmember banks cannot
be used to improve sigmficantly the estimates
for the smaller nonmember banks
Third, the pattern of weekly changes (as
distmct from levels) m demand deposits for
the sample 1s not drastically different from
the pattern estimated for nonmembers by
ex1stmg procedures on the basis of changes
at country member banks
Fourth, the statistical properties of the
FDIC sample imply that, when the sample 1s
used to extrapolate data from the latest call
report date, 1t should provide m1tial estimates
of the umverse total of nonmember demand
deposits with a standard error of a little more
than $300 million When the sample 1s used
to mterpolate between data from the precedmg and succeedmg call report dates, the
"final" estimate should be obtamed with a
standard error of at the most $240 mill10n
Fifth, the sample improves accuracy to a
maJor degree The evidence from three call
report dates 1s reasonably consistent with the
a priori estimate of the standard error If the
data from the sample had been available on
a current basis and If they had been mcorporated mto the mitial estimates of Mv the
benchmark revis10ns for the reportmg dates
m the fall and m December 1974 and m the
sprmg of 1975 would have been no more than
$200 million mstead of rangmg up to $2 4
Federal Reserve Bank of St. Louis

After reviewmg current procedures, the
sample explorations, and various alternative
proposals, the Committee concluded that the
maccuracies m the estimate of demand deposits of nonmember banks represent a maJor
defect m up-to-date monetary statistics and a
sigmficant defect m historical statistics of
M 1 and that marked improvements are
feasible at reasonable costs for both reportmg
nonmembe1 banks and the Federal agencies
Encouraged by the results obtamed from
the daily reportmg sample of nonmember
banks, the Committee urges the resumptron
of such a procedure on a contmumg basis and
the development of techmques to permit the
sample reports to be processed as rapidly as
are data from country member banks The
Committee has considered carefully the possibihties of improvmg prov1s10nal estimates
and mterpolauons of nonmember bank deposits by new methods that rely on currently
available data from member banks We recogmze that such methods would be less costly,
but m our Judgment they would be far less
reliable We are convmced that a currently
reportmg sample of nonmember banks 1s
essential to develop up-to-date estimates for
the nonmember umverse with a reasonable
standard of accuracy-which, m our Judgment, would be met by a standard error of
around $300 millron
The Committee also recommends that all
nonmember banks be asked to report weekly
data m addition to I-day figures on call report
dates, and that the processmg period for the
call report data be shortened substantially
from its current 4-month length It 1s our
understandmg that tlus change would not be
costly to either the reportmg banks or the
Federal data processors
In combmauon, these two reforms would,
for practical purposes, substantially solve the
currently sigmficant problem of maccurate
estimates of nonmember bank deposits m the
monetary aggregates 9
As this report 1s bemg completed, we are mformed
that begmnmg m March 1976, the FDIC plans to col
lect 7 days of deposit data for each call report week We

Measurement Issues and Recommendations

Consolidation of data from different
financial institutions
Conceptual issues m definmg monetary
aggregates discussed above pomt to special
problems of measurement m av01dmg double
countmg The Committee paid special attention to the problems that anse m the treatment
of bank fl.oat

Present treatment of -fl,oat
The elapsed time between sendmg, depos1tmg, and collectmg checks gives nse to changes
m bank deposits that are referred to as fl.oat
As we noted earlier, there are two kmds of
fl.oat-mail float and bank fl.oat When A
wntes a check, he records a debit to his checkbook balance Until B receives the check and
deposits it m his bank, deposits on the books
of the banks have not been affected Dunng
this period the check IS m transit between A
and B, and the dollar amount of such checks
1s called "mail fl.oat " When B receives and
deposits the check, his deposit balance IS mcreased, and his bank sends the check for
collect10n to A's bank, recordmg a CIPC as
an asset on its books When the check clears,
A's deposit account 1s debited, and reserves
are transferred from A's bank to B's In the
mtenm between B's deposit of this check and
its debit to A's account, B's deposits are higher and A's have not yet been reduced, thus
total recorded deposits m the bankmg system
are higher by the amount of the CIPC This
mcrease m recorded deposits, or double
countmg, due to checks m process of collection between banks 1s called bank float
As noted prev10usly, we have accepted the
trad1t10nal procedure of deductmg bank float
but not mail fl.oat, for the reasons mdicated
If all banks kept their records properly, bank
note that the FDIC staff 1s convmced, rev1ewmg the
same evidence we have used, that availab1hty of this
improved regular call report data plus mformat10n pro
v1ded by the expenmental sample will make 1t possible
to meet our standards of accuracy without mst1tut10n of
a regularly reportmg sample of nonmember banks We
doubt that this 1s possible
Federal Reserve Bank of St. Louis


float would be measured by the CIPC between banks

Current errors in the treatment
of bank float
The process of ehmmatmg double countmg
of deposits by deductmg bank float can lead
to error when checks are drawn on accounts
that are not mcluded m the money stock or
when checks m process of collect10n are not
reported as CIPC
The first error anses from checks drawn on
US Government, mterbank, and foreign accounts, as well as collections of nondeposlt
Items, compnsmg mamly postal money orders
and nonbank traveler's checks, redeemed savmgs bonds and coupons on Government securities, food stamps, and credit-card shps
Inclus10n of these items m the CIPC produces
an overly large subtraction from gross deposits and hence an understatement of domestic demand deposits held by the pubhc It
would be very costly for banks to count these
Items separately, and no attempt 1s made to
do so m calculatmg M 1
The second error anses from the practice
of many banks of clearmg checks through
correspondents and of reportmg checks m
process of collect10n as "due from banks"
rather than as CIPC Suppose, for example, a
check drawn on Bank B 1s deposited m Bank
A and collected through correspondent Bank
C Instead of creditmg "due from banks,"
Bank A should credit CIPC unul the next
day when its account at Bank C 1s credited,
but 1t has no mcenuve to do this Indeed, some
nonmember banks have a special mcentive
not to do so (Whereas for member banks,
both due from banks and CIPC are deducted
from gross deposits m calculatmg reserve
reqmrements, for some nonmember banks the
amount due from banks, but not CIPC, 1s
counted as part of the bank's legal reserves)
In addition, Bank A relieves Itself of the
bookkeepmg cost ot transferring Items from
CIPC to due from banks by shortcuttmg the
CIPC stage of accountmg Once Bank C receives the check and credits Its CIPC, the


Improvmg the Monetary Aggregates Committee Report

money stock adJusted for CIPC 1s agam the
same as 1t was before the check was depos1 ted
m Bank A On the first day, however, the
calculated money stock adJusted for CIPC 1s
higher than 1t should be because Bank A does
not record a CIPC
Pr10r to November 9, 1972, there was another source of error that was opposite to the
due-from-banks bias Just noted-the Federal
Reserve gave Bank B I day to remit payment
This pertamed to all checks that cleared
through the Federal Reserve payments system
In that I-day grace penod, Bank B debited
the payer's deposits and credited, not its
CIPC, but a nondepos1t hab1hty to the Federal Reserve (which reduced Bank B's reserve
reqmrements) On this day total deposits were
as they had been before the check was depos1 ted m Bank A, but Bank C still earned a
CIPC Hence the money stock was measured
as bemg lower than 1t should have been Smee
1t was too high the first day and too low the
third day, on the average durmg the full
transaction 1t was correct
A change m Regulation J on November 9,
1972, reqmred banks to remit payment to
Federal Reserve Banks on the same day 10
This produced the present clearmg process
and thus removed the offset to the due-frombanks bias on the first day described above
As the check-clearmg process speeds up m
the future, the overstatement of the money
stock because of due-from-banks bias will disappear smce all the transact10ns m the threebank clearmg process will be completed on the
same day However, the attamment of sameday clearmg 1s some years away
In add1t10n to the change m Regulat10n J,
two other steps have been taken m recent years
to correct maJor errors m the measurement of
bank float First, effective July 31, 1969, Regulat10n D was changed to reqmre member banks
to mclude m their deposits subJect to reserve
reqmrements any checks sent by them m pay10
"Revmon of the Money Stock Measures and Member
Bank Reserves and Deposits " Federal Reserve Bulletin,
February 1973, pp 62---o4
Federal Reserve Bank of St. Louis

ment of foreign Euro-dollar transact10ns Such
deposits serve as an offset to the mcrease m
CIPC that 1s produced when the rece1vmg
banks remit the checks for payment 11 Second,
a s1m1lar kmd of bias ongmated m the checks
sent by agencies of foreign banks and Edge Act
corporat10ns Pnor to 1970, deposits with these
mst1tut10ns were not mcluded m the money
stock, so subtract10n of float generated by
checks drawn on these deposits improperly reduced the calculated money stock This understatement was corrected m 1970 by mcorporatmg mto Mi the deposits and officers' checks
reported by these mst1tut10ns 12

Tentatively recommended alternative
calculation of Mi
We tentatively recommend an alternative
method of ehmmatmg bank float m calculatmg
M 1, which we believe will produce s1gmficantly
improved estimates of M 1 and related aggregates To remove the due-from-banks bias resultmg from madequate use of CIPC, we recommend that amounts due to banks be
mcluded m gross deposits and then those due
from banks be subtracted In this way, net
domestic commercial mterbank deposits
would be ehmmated from the gross deposits
on banks' books Usmg this net figure ex11"Rev1s10n of Money Supply Senes," Federal Reserve
Bulletin, October 1969, pp 788-89
"Revis10n of the Money Stock," Federal Reserve Bulletin, December 1970, pp 890-92
In further mvestigauons of CIPC problems for this
Committee, the Board staff discovered that the cashitems bias adjustment has been overstated m some cases
One source of overstatement is that some banks have
been accountmg for checks deposited by agencies of
foreign banks and Edge Act corporations m a different
manner In these banks these checks are not credited to
the accounts of the receivmg mst1tut10ns on the day of
deposit and therefore do not produce a CIPC that day,
as is assumed by the bias adjustment, rather they are
credited on the next day, when the officers' checks drawn
agamst these deposits clear A second source of overstatement is that the checks deposited by these mstitu
uons m New York City banks for collection do not give
nse to CIPC if, as 1s true to some extent, the checks are
drawn on the same bank m which they are deposited
The Board's staff has undertaken to correct this overstatement of the cash items bias adjustment m the pub
llshed data

Measurement Issues and Recommendations

eludes from M 1 any CIPC mappropnately
reported by banks as due from banks
Under this recommendat10n M 1 would thus
be calculated by deductmg amounts due from
domestic commercial banks and CIPC from
gross deposits exclusive of Treasury deposits
but mclus1ve of amounts due to US banks
(As recommended earlier, amounts due to
foreign commercial and central banks would
also be excluded) The recommendat10n has
the further advantage of makmg data collect10n easier than 1t 1s now, because the separate
estimates now reqmred of deposits of mutual
savmgs banks and banks m terntones and
possess10ns would no longer be necessary
(Such deposits would be mcluded m the dueto-banks item of the recommended defimt10n )
The reasons why we propose this recommendat10n only tentatively are explamed m the
followmg d1scuss10n
Table 3 illustrates m detail how M 1 1s currently estimated Table 8 repeats that calculat10n m the first column and shows m the
second column how M 1 would be estimated
for December 1974 by usmg the tentatively
recommended alternative approach The more


simple method recommended should eliminate the CIPC bias that has troubled the
ex1stmg method, and should avoid the necessity of makmg the difficult estimates shown m
Imes 7-10 under the present method On the
other hand, the alternative method reqmres
special data on Edge Act corporat10ns and
their CIPC's that are not required under the
present method, but the sums mvolved are
relatively small
As Table 8 md1cates, M 1 calculated by the
tentatively recommended alternative 1s $8 billion less than when estimated by the currently
used method In pnnc1ple, the current and the
tentatively recommended alternative methods
should give 1dent1cal figures for M 1 Thus, the
choice 1s not one of concept, but rather a stat1st1cal issue as to which method makes possible the closest approx1mat10n to the concept
mvolved Detailed staff analysis suggests that
the precise reasons for the discrepancy are centered m the recordmg of foreign transact10ns,
mamly at large New York City banks, and that
they have developed mamly over the years
smce 1970, although there was a s1gmficant difference m the late 1960's

TABLE 8 Comparison of Methods1 for Calculat10n of Demand Deposit Component of M 1
Monthly average, m1lbons of dollars
December 1974
Line, item




Gross demand deposits at commercial banks
Demand deposits due to commercial banks
U S Govt demand deposits


CIPC adJustment
Demand depos1 ts due to foreign banks
Demand deposits due to mutual savmgs banks
Demand deposits due to banks tn territories and possessions
Equals Commercial banks' component of M,
Plus foreign related institutions
M 1 type balances at Edge Act corporations, agencies, and mvestment
Foreign demand deposits at FR Banks
Gross demand deposits of Edge Act corporat10ns, agencies and
investment compames2
CIPC at Edge Act corporations, agencies, and mvestmcnt compames'
Demand deposits due from banks at Edge Act corporations and agencies
Equals Total demand deposits
Plus Currency
Equals Total M 1
Difference Recommended less current



R float
Demand deposits due from domestic commercial banks







1 For an explanat10n of the method currently used, see Table 3 and the accompanymg text
'Branches mcluded m 1974 data
Federal Reserve Bank of St. Louis




lmprovmg the Monetary Aggregates· Committee Report

TABLE9 Alternative Estunates of M,, 1968--74
Averages of daily figures for December
In bill1ons of dollars

Line, item


Current M,
Plus Net interbank deposits
Minus Adiustment for cash llems bias
Minus Adiustment for overstatement of
remittance payments bias

5 Equals Recommended M 1
6 DJlference (5) mmus (1)

208 I
I 8



214 8 225 7 240 4 262 6 278 6
2 -20 -42

291 3




205 6 212 0 220 2 235 3 257 7 272 0 283 3
-25 -28 -55 -51 -49 -66 -80

Table 9 shows annually for the period 196874 the maJor items mvolved m estlmatmg M 1
by the tentatively recommended alternative, m
comparison with M 1 as currently estimated M 1
under the alternative method (lme 5) 1s derived from the current M 1 by addmg net mterbank deposits and subtractmg two adjustments
The first, shown on lme 3, removes the current
cash-items-bias adjustment because the mclus10n of amounts due to banks provides an offset to the Euro-dollar transfers and other checks
that may now be improperly mcluded m the
CIPC The second, shown on lme 4, adJusts
for the fact that the remittance-bias correction
m the money stock, made for the period before
the change m Regulat10n J m November 1972,
1s too large for the recommended defimt10n
Prior to November 1972 the I-day grace
period for rem1ttmg payment to Federal Reserve Banks created a CIPC bias because the
rem1ttmg bank debited the deposit account on
which the check was drawn and at the same
ume credited, not a CIPC, but a nondepos1t
liability to the Federal Reserve This resulted
man understatement of CIPC However, some
nonmember banks, which remitted through
correspondents, debited their due-from-banks
account on the day before their correspondent
simultaneously remitted to the Federal Reserve
and debited the due-to-banks account, which
offset the understatement of CIPC Because of
this offset, adjustment of CIPC for the remittance bias 1s too large for the recommended
alternative method that deducts due-frombanks deposits directly The Board's staff has
estimated the amount of this remittance bias
prior to November 1972 that is not applicable
to the recommended defimt10n These estimates, on lme 4 of Table 9, are subtracted
Federal Reserve Bank of St. Louis


from the current M 1 to obtam the alternative
estimate of M 1
Lme 6 shows that M 1 estimated by the alternative method is persistently smaller than M 1
as now estimated The difference before
I 970, which is m the $2 5 billion to $3 bilhon
range, appears to reflect largely the exclus10n
of checks m transit that were mappropriately
reported as due from banks, exclus10n of these
cash items as proposed should remove this sigmficant source of spurious vanat10n m the
money stock
Begmnmg m 1970, however, the difference
between the current and alternative methods
becomes much larger for reasons that have not
yet been fully determmed In 1972 agencies of
foreign banks and Edge Act corporat10ns mst1tuted the Paper Exchange Payments System
(PEPS) to facilitate the clearmg of Euro-dollar
transact10ns, and it seems likely that this system accounts for most or all of the recent mcrease m the hne 6 difference, smce data for
banks outside New York City do not exlubit a
comparable mcrease In addit10n, the mstitut10n of the New York Clearmg House Interbank Payments System (CHIPS) m 1970 may
account for some of the sharp mcrease, m the
difference m 1970-71
M 1 senes as estimated by the current and
alternative methods are compared on monthly
and quarterly bases m Tables 10 and 11 Both
were seasonally adjusted by the same Census
Bureau X-11 program without any Judgmental
adjustments These tables were constructed m
early 1975 and so do not mclude recent revis10ns m the data The comparison should thus
be considered tentative pendmg mcorporat10n
of these revised data
The differences between the two M 1 senes


TABLE 10 Comparison of Current and Recommended Mi-Monthly
Seasonally adjnsted, amounts m uullions of dollars
Demand deposits m Date






Annual growth rates (m per cent)

mended M 1




194 655




3 65
3 77
4 40
9 01
9 80
6 72
10 18
8 60
8 13
12 58
9 83

3 99
3 93
4 13
6 77
10 31
9 64
7 03
8 57
9 41
II 18
7 85







column 7











5 80
4 26
3 60
2 91
l 27
4 36
2 57
l 06
4 57
4 52
3 80
I 41

4 7!:I
4 30
2 29
3 04
I 48
4 34
I 14
5 69
4 34
4 92

-I 01
- 02








11 09
-6 69
9 58
7 27
4 32
2 84
3 39
12 23
JI 49
3 34
4 85
5 36

6 or;
7 76
8 40
-3 94
7 57
7 75
5 72
5 26
I 68
8 30

-5 04
-1 83
I 13
-5 26
4 18
-5 77
I 92
-3 17
2 95


I 79,270






7 66
8 01
9 05
6 92
12 26
7 91
5 65
5 98
3 45
3 09
2 99

8 72
10 60
8 88
8 07
10 83
6 03
7 42
4 59
4 56
3 15
4 50
5 65

I 06
2 59
I 16
-1 88
I 77
-1 38
3 84
2 66








5 03
9 01
11 27
5 60
2 91
4 04
11 25
JO 17
JO 44
7 80
6 10
15 41

2 54
11 28
14 29
5 72
5 57
11 12
8 61
10 16
7 42
- 28
14 43

2 27
3 02
I 52
-1 ,6
- 27
- 38
-6 38








10 47
l 71
4 33
12 75
10 69
4 00
2 53
- 48
4 29
11 35
9 18

7 30
2 47
4 00
10 03
9 64
5 39
5 49
4 99
9 33
4 63

- 33
-2 73
-I 05
I 38
2 96
I 29








4 56
6 17
7 07
5 38
3 74
6 73
3 48
3 51
3 18
5 13
4 51
2 39

9 56
7 80
12 14
-3 38
6 53
2 06
- 62
3 87
6 15
4 27
7 35

3 3<)
6 76
- 20
1 02
4 95

Federal Reserve Bank of St. Louis





2 91
- 03
I 28




lmprovmg the Monetary Aggregates Committee Report


the current mvest1gat10n by the Board's staff
of the effects of PEPS and CHIPS, and of other
possibly important factors, m est1matmg M 1 13

were tested for two kmds of economic relat10nslups First, the rate of change of personal mcome was regressed on the rate of change of
each Mi, lagged 1 to 12 months m 12 different
regress10ns, where the rates of change of both
vanables covered 6 months at a time Second,
a standard money-demand equat10n was used
to regress each Mi senes on the 3-month Treasury bill rate, gross nat10nal product, and a 1penod lagged money stock, where the vanables
were loganthms of quarterly averages In both
sets of regress10ns, Mi calculated by the recommended alternative method gave a slightly better fit Although not stat1st1cally srgmficant,
the differences between the regress10n fits could
be mterpreted as margmally supportmg the
1ecommended method of calculatmg M 1
Thus, on both theoretical and emp1ncal
grounds, the alternative method of consohdat10n of accounts that we suggest seems preferable to the current method However, smce we
are not yet sure of all the reasons for the growmg discrepancy m M 1 as estimated by the two
methods, we recommend the new alternative
only tentatively, contmgent on the outcome of

Remaining sources of error in the
treatment of float
It would be desirable to correct float to exclude the amount of cash items not ansmg
from checks wntten on 1pnvate demand deposits This correct10n 1s not made under our
tentatively recommended calculat10n because
of the unava1lab1hty of the needed data There
1s reason to believe, however, that the resultmg understatement of the money stock 1s small
Most of the nondeposit cash items (food
stamps, coupons, and credit slips) are quant1tat1vely small The only large item is U S
13Estimatmg complex concepts by alternative methods
often provides d1ffermg results, although the figure obtamed should be the same whichever way 1t 1s estimated
For example, somewhat different estimates of GNP are
generally obtamed by usmg the value-added and finalexpenditures approaches In such cases (mcludmg M1) ,
1t 1s difficult to say which 1s the true amount of the concept bemg calculated

TABLE 11 Companson of Current and Recommended M1-Quarterly
Seasonally ad111sted, amounts m m111ions of dollars

Demand deposits m Period



Federal Reserve Bank of St. Louis




I mended kt,




<\nnual grol\ th rates (m per cent)






189 732

- 205
- 952
- 905
- 997




7 79
8 56
10 26
2 86
3 25
4 65
4 83
4 53
8 30
9 09
5 05
2 25
8 49
4 20
10 71
9 84
3 94
9 32
2 02
8 33
5 96
5 31
3 40
4 03

7 II
8 47
9 55
3 81
2 96
2 75
3 09
I 95
I 16
7 05
5 IO
9 48
8 37
5 55
4 45
9 43
4 93
JO 05
7 22
3 24
7 94
3 91
6 35
6 97
5 10
I 77
5 95

I Column 8

column 7



- 16
-2 05
I 18
- 73
2 20
- 67
-2 62
- 70
-1 39
I 89
I 00
- 21
-I 63
I 92

Measurement Issues and Recommendations

Treasury check!f Although tlus quantity
vanes between $0 5 billion and $1 5 billion
per day, the Board's staff estimates that up to
70 per cent of these checks do not give nse
to cash items m collect10n because they clear
the same day, the cash items that Treasury
checks do generate are apparently on the
order of only $300 m1ll10n per day Because of
the difficulty of obtammg separate data on
these items, no attempt to correct for this
overstatement of float 1s recommended
Smee these errors would still remam m the
treatment of float, we considered the alternative of not deductmg float at all That process
would overstate the level of the money stock
on a bank-record basis but might give more
accurate rates of change through the elimmat10n of fluctuatmg errors m the float component A test of this suppos1t10n 1s whether variat10ns m the money stock are reduced by elimmatmg the ad1ustment for float Tlus test was
conducted for monthly money stock data for
1960-73 The standard deviat10ns and coefficients of vanat10n were calculated for M 1 , both
mcludmg and excludmg the various components of float For both levels and rates of
change, M 1 ad1usted to elimmate float shows
substantially smaller vanat10n Other evidence
based on fittmg standard money-demand regress10ns for gross and ad1usted-for-float concepts of money also leads to a shght preference
for the ad1usted money stock 14

Seasonal adjustment

Basic approach
In many economic senes, seasonal vanat10n
results from natural causes changes m temperature or ramfall In others, 1t 1s a consequence of stable sonal mst1tut10ns number
of workmg days m a month, consumer purchases of Christmas tree ornaments In still
14These regressions followed the form of those presented m Stephen Goldfeld, "Money Demand Revis
Ited," Brookings Papers on Economic Activity, 1973 3
The regress10ns cited m the text above were performed
for the Committee by Professor Goldfeld The regress10ns m which the standard defimt10n of M 1
were used had marginally lower standard errors than
those not adJusted for various components of float
Federal Reserve Bank of St. Louis


others, 1t 1s a mixture, m part a result of
natural causes or of stable social mst1tut10ns
and m part a consequence of human act10ns
m response to natural phenomena crop
yields, fuel consumpt10n, product10n of
Christmas tree ornaments
The approach to the mixed type of seasonal
depends on the user's purpose If the analyst
1s the economic agent who 1s reactmg-the
producer of Christmas tree ornaments, for example-he will want to separate clearly the
"natural" or "exogenous" seasonal (which 1s
typically outside his own control) from his own
react10n to 1t Given the seasonal 1n consumer
demand for Christmas tree ornaments, he will
set his product10n schedule m light of costs of
storage versus costs of bunchmg product10n
The seasonal m his product10n schedule will
be a "policy" seasonal deliberately arrived at,
not a natural seasonal, but 1t will of course be
strongly mfluenced by the natural seasonal m
consumer demand
If a stat1st1cian 1s analyzmg busmess act1v1ty,
employment, and the like, he will observe the
end result of the combined natural and policy
seasonals of purchasers and producers-for example, m the product10n of Christmas tree
ornaments Insofar as this pattern 1s repetitive,
he may want to abstract from 1t m order to
isolate more sharply the effect of longer-term
changes m the output of Christmas tree ornaments If so, he will want to construct a descriptive seasonal, which applies to the actual performance of the senes mcludmg the results of
both natural and policy seasonals
The admixture of "natural," "policy," and
"descnptrve" seasonals rs particularly troublesome for the Federal Reserve's seasonal ad1ustments of monetary aggregates On the one
hand, the Fed 1s faced with such natural seasonals-from its pomt of view-as fluctuat10ns
m i etarl sales and the associated fluctuat10n
m the desired rat10 of currency to demand
deposits, corporate tax payment dates, and,
on an even more subtle level, the effect on all
of these magmtudes of such variables as fluctuat10ns m pnces and mterest rates On the
other hand, it has nearly complete control


Improvrng the Monetary Aggregates Committee Report

over the monetary base and through the base
over other monetary totals, so 1t can mtroduce
any policy seasonal 1t wishes mto some one
total For example, 1t could decide that the
base should have no seasonal, m that case,
Mi, M 2 , and so on would have whatever seasonals are produced m them by the natural
seasonals m the relat10nslups lmkmg the base
with the other totals Or alternatively, 1t
could decide that M 1 should have no seasonal,
m which case 1t would have to mtroduce
whatever seasonal m the base 1s reqmred to
elimmate the seasonal m M 1 Or to be more
general, 1t could use its control over the base
to modify seasonals m several totals, given
that with one mstrument (the base) 1t can
control only some combmat10n of seasonals
Or on an even more general level, 1t could
use other mstruments, such as discount rates or
reserve reqmrements, to widen the policy alterat10ns 1t could mtroduce mto the several
In principle, the policy seasonal and the
monetary total mto wluch 1t 1s to be mtroduced should be chosen m terms of the obJeCtives one wants to aclueve We have not found
any explicit discuss10n m Federal Reserve releases of the cntena for choosmg a policy seasonal That om1ss10n should be remedied
The evidence suggests that the Fed's policy
on seasonals has been designed mamly to offset a natural seasonal in the demand funct10n
for money-that 1s, m the relat10n among the
quantity of money demanded, mcome, and mterest rates The mam policy ob1ect1ve of mtroducmg a seasonal m money has apparently
been to reduce the amplitude of the seasonal
m mterest rates 1 ~ However, the actual seasonal
m the base or other monetary aggregates has
apparently not been an explicit policy decmon
arnved at by combmmg an explicit, desired,
muted seasonal m mterest rates with a determmat10n of the seasonal m money reqmred to
produce such a seasonal m mterest rates The
actual seasonals m the monetary aggregates appear rather to have ansen almost advent115
Th1s view 1s supported by the decidedly smaller
amphtude of the seasonal m mterest rates after the
estabhshment of the Fed m 1914
Federal Reserve Bank of St. Louis

t10usly Insofar as the Open Market Committee
has stated its ob1ect1ves m terms of monetary
totals, 1t has done so m terms of desired rates
of change m seasonally ad1usted totals The
staff has then computed the changes m seasonally unad3usted totals reqmred to achieve the
targets m seasonally adjusted totals It has
done so primarily by calculatmg a descriptive
seasonal for the past and extrapolatmg mto
the future
Tlus procedure may mtroduce umntended
changes m monetary policy, particularly 1f the
descnpt1ve seasonal 1s calculated by a strictly
mechamcal movmg seasonal
method such as the Census X-11 method Suppose, for example, a monetary total 1s expanded m March for several years m sequence
by substantially more than the prior seasonal
amount-whether because of random disturbances or because nonseasonal policy considerat10ns happen to call for a more rapid expans10n A mechamcal seasonal ad3ustment would
tend to mcorporate this deviat10n m the movmg seasonal for subsequent years, and the
deviat10n would then be validated by policy
dec1S1ons expressed m terms of seasonally ad3usted totals The staff has been aware of this
problem and for this reason has not simply
accepted the X-11 seasonal ad3ustment, but
has made Judgmental correct10ns designed to
av01d umntended policy results of the sort
JUSt described However, this seems to us an
unsatisfactory procedure
A more satisfactory procedure would identify
explicitly the followmg items (1) the seasonal
policy ob1ect1ves-the target seasonals m nommal mcome or mterest rates, (2) the seasonals
m Mi or other totals reqmred to achieve that
ob1ect1ve, and (3) the seasonals m Federal Reserve open market operat10ns or other policy
mstruments reqmred to aclueve item (2) At
the moment, our knowledge of the relauonslup
between items (1) and (2) 1s too meager to enable us to allow for anytlung but ma3or developments altermg that relat10nship For example, a number of tax law changes culmmatmg
m 1968 produced a ma3or alterat10n m payment dates that clearly reqmred a change m

Measurement Issues and Recommendations

item (2) to achieve any specified policy ob1ective of the kmd listed m item (1)
Makmg use of the hmited available evidence
bearmg on these issues, the Committee mvestigated the descriptive seasonal for past years m
currency, demand deposits, Mi, and M 2 We
conclude that, smce the changes m tax payment
dates culmmatmg m 1968, there has been no
economically substantial change m the descriptive seasonal We therefore regard this average
descriptive seasonal as mcorporatmg the Federal Reserve System's present policy 16
Individual members of this Committee have
divergent views about what the seasonal policy
ob1ective of the Fed should be However, that
policy decis10n 1s not withm the scope of our
assignment In the context of our assignment,
we are convmced of the importance of distmgmshmg the policy seasonal from the descn ptive seasonal Failure to distmgush between the two may lead to mcorporat10n m
future monetary changes of seasonal vanat10ns from the past that may have been
mtended or unmtended, desirable or undesirable In this way, the mechamcs of seasonal ad1ustment may unmtent10nally become
determmants of monetary policy

Accordmgly, we recommend the followmg
I The Fed should choose and publish m
advance its best estimate of the seasonals that
1t mtends to use as its gmde to policy decis10ns
for some substantial penod ahead (say, a year)
We suggest also that further research be mstituted on the relat10n between seasonal movements m policy mstruments and m monetary
totals, and on the relat10n between seasonal
movements m monetary totals and m the more
basic ob1ect1ves of nommal mcome, real mcome, and mterest rates-m order to improve
the basis for policymakmg and to permit a
more prompt and more accurate allowance
for changes m the natural seasonals lmkmg
policy mstruments with monetary totals, and
monetary totals with basic ob1ect1ves
2 As a service to persons usmg such data

•More detailed mformation supportmg this conclu
Federal Reserve Bank of St. Louis


for research and other purposes, the Federal
Reserve should mclude m its retrospective publicat10n of monetary data seasonally ad1usted
senes mcorporatmg Its best estimates of descriptive seasonals
3 Seasonally unadjusted data should also
be published retrospectively for those who
want to study past pohcy or to use other approaches to adJustmg for seasonal vanat10n
These statistical reforms would establish a
clear distmct10n between the policy seasonals
and the descriptive seasonals Any change m
the policy seasonals would reflect an exphot
dec1S1on by the Federal Reserve to change its
seasonal ob1ect1ve or to ad1ust to perceived
ma1or changes m natural seasonal forces (as m
1968) But the policy seasonals would not
change JUSt because the descriptive seasonal
recorded a change m the seasonal pattern of
the money stock over recent years

Suggested technical change
In mvestigatmg past seasonal movements,
the Committee developed a statistical method
of computmg a descriptive seasonal that takes
advantage of the availability of daily data and
therefore permits allowance m a rather simple
fash10n for changes m weekly and monthly
seasonal factors that reflect the occurrence of
holidays, different number of days m the
month, and the like
The fost step m the piocedme is to allow
for the mtraweek seasonal, that is, a systematic
patte1n that makes Monday systematically different from the other days of the week, and so
on This was done by first expressmg observat10ns for each day as a rat10 to the average for
the week of wluch 1t is the central day 17 The
averages of these rat10s for all Mondays, Tuesdays, and so on, give a day-of-the-week seasonal Vanous tests were made to determme
whether the day-of-the-week effect vaned over
s1on is provided m a staff memorandum m the second
volume of this Report
1 -Because most banks are closed on Saturday and Sunday, daily figures for Saturday and Sunday are essen
tially identical with Fnday Hence, we experimented
with both 7 day averages and 5 day averages, which
explams the roundabout statement m the text


Improving the Monetary Aggregates Committee Report

the course of the year or from year to year
The evidence thus far md1cates essentially constant day-of-the-week effects, so a smgle correct10n was used throughout
The second step was to d1v1de each daily
observat10n by the relevant day-of-the-week
factor The ad1usted value for each day was
then expressed as a rat10 to the average of the
365 days of which 1t 1s the central day A
Fourier series was then fitted to these daily
rat10s, the Fourier terms were arranged 1n
order of amplitude, and the 30 sme or cosme
terms with the largest amplitudes were retamed A multiple regress10n was then fitted
expressmg the daily ratios as a funct10n of
these 30 terms plus 11 dummy variables for
days either precedmg or on holidays The
value for each day calculated from this regress10n and then multiplied by the relevant
day-of-the-week factor 1s the seasonal ad1ustment factor for that day The rat10 of the
observed value to the seasonal ad1ustment factor 1s the seasonally ad1usted daily value It
can be summed for weeks or months to get
seasonally ad1usted weekly or monthly values
This procedure was applied for various periods and for md1v1dual years There was a
clear break at 1968, but year-to-year differences
after 1968 were very small
The Committee believes that the above procedure offers a number of advantages and that
1t should be seriously considered as an alternative to the present Judgmentally adJusted Census X-11 method of seasonally ad1ustmg money
stock data Each step 1s clear and simple, which
should aid m mterpretmg and adaptmg to
puzzlmg circumstances that might be encountered The use of daily data offers the opportumty to develop efficient stat1st1cal estimates
and tests and makes 1t easy to mcorporate
knowledge of holidays and special events mto
the procedure This should make 1t possible to
Federal Reserve Bank of St. Louis

reduce substantially troublesome ex-post rev1s10ns of historical adJusted data
This procedure, first suggested by Professor
Friedman, was developed m detail by the
Board's staff The Committee encourages contmued analysis of this method by the Board's
staff and recogmzes that useful mod1ficat10ns
of 1t may be found 1s

Ownership of demand deposits
Informat10n on ownership of demand deposits by different groups m the economy 1s
also important m un~erstandmg the uses made
of such deposits by these groups of spenders In
order to understand the role of money m the
economy,· 1t 1s necessary to analyze the demand
for money balances While a great deal of research has been done m this area, 1t has been
seriously hampered by the lack of adequate
ownership data
The Federal Reserve mterm1ttently before
1970 and regularly smce that time has collected
and published mformat10n on the ownership
of demand deposits by broad classes of private
owners Although these data do not yet form a
long enough historical series to be of great
analytical value, their contmued collect10n and
publicat10n seem to us important m order to
bmld up an historical senes extensive enough
to be of analytical value We recommend,
therefore, contmued collect10n of these data
and improvement of the series bemg collected
They should be mcreasmgly valuable over the
years ahead 19
Further details of the method and the results of
applymg 1t to past data are provided m a staff memorandum published m the second volume of this Report
The case for contmuat10n of the demand deposit
ownership survey and suggestions for its improvement
are presented m a staff memorandum on this topic m
the second volume of this Report



Summaries of Staff Papers

During the course of dehberat10ns by the Advisory Committee on Monetary Statistics, a large
number of staff papers were prepared for the
Committee by members of the staff of the Board
of Governors Some were written to provide general background mformat10n, while others were
directed at specific issues The Committee suggested that some of the studies should be made
available to the public Consequently, the Board
of Governors of the Federal Reserve System plans
to publish the s1gmficant staff papers m a compamon volume to this Report of the Advisory
Committee on Monetary Statistics
The staff papers to be published draw on those
prepared for the Committee, most reflect an mtegrat10n of md1v1dual studies In preparmg the
papers for pubhcat1on, the Board's staff attempted
to mclude only those materials that had been
presented by the staff to the Committee
The staff papers, of course, do not necessarily
reflect the op1mons of the Board of Governors of
the Federal Reserve System
The eight papers to be published are summarized below

Transitory Variations in the
Monetary Aggregates
by Agustm Maravall, Darrel W Parke,
and Richard D Porter
The monetary aggregates are sub1ect to a variety
of very short-term transitory mfluences that impart
day-to day variauons (noise) m the series Though
these variations are unrelated to longer term movements m the series, It 1s useful to isolate their
impact on measured growth rates, and this paper
explores two empmcal methods of estlmatmg such
One method utibzes daily data, together with a
simple analysis of a variance model, to estimate
the transitory variance The second approach 1s
based on a time series analysis of the series It 1s
Federal Reserve Bank of St. Louis

shown that under relatively weak assumptions, the
transitory variance can be estimated from the autocovanances of the observed time series Advantages
and pitfalls of each method are exammed and

Foreign Demand Deposits at
Commercial Banks in the United States
by Helen T Farr, Lance Girton,
Henry S Terrell, and Thomas Turner
The paper 1s d1v1ded mto two parts The first
provides a general descript10n of foreign banks,
foreign md1v1duals, partnerships, and corporat10ns, and foreign official demand balances at
banks m the Umted States This section pays particular attent10n to mst1tut10nal arrangements m
which foreign commercial banks hold demand
balances with commercial banks m the Umted
States to compensate for the clearmg and other
services provided to these foreign banks by
domestic commercial banks
The second part of the paper attempts to estimate empmcal demand relat1onsh1ps for the three
types of foreign deposits at commercial banks m
the Umted States The general empmcal results
suggest that the three categories of foreign deposits are not emp1ncally related to domestic
macroeconomic variables m the same way as
domestically owned deposits

Nonmember Banks and
Estimation of the Aggregates
by Darrel W Parke
Rev1s10ns of the estimates of nonmember bank
deposits have, m recent years, led to substanual
benchmark rev1S1ons of the money stock By usmg
8 months of daily deposit data for a sample of
nonmember banks collected on an experimental


lmprovmg the Monetary Aggregates Committee Report

basis by the Federal Deposit Insurance Corporauon, this paper explores ways of 1mprovmg the
Federal Reserve's esumatmg procedure It 1s concluded that not enough mformauon 1s presently
available to the Federal Reserve staff on a contmuous basis to improve the esumates but that
s1gmficant improvements could be obtamed 1f more
data were available Such data mclude (a) deposit data reported on a weekly basis by a sample
of nonmember banks s1m1lar to those mvolved m
the FDIC experiment, and (b) 7 days of deposit
data reported by all nonmember banks on each
call report

Seasonal Ad1ustment of the
Monetary Aggregates
by David Pierce, Neva Van Peski,
and Edward R Fry
This paper discusses the problems and concepts
mvolved m seasonally adJustmg the money stock
and compares alternauve methods for domg this,
mcludmg the development of a daily seasonal
adjustment procedure
The paper mcludes a d1scuss1on of the concept
of movmg seasonality, mcludmg tests on the
money stock to examme whether s1gmficant
changes m seasonal factors have occurred over
recent years A new daily seasonal adjustment procedure 1s then presented and analyzed It has the
feature that once daily seasonally adjusted data
are determmed, weekly, monthly, or quarterly
seasonal adjustments can immediately be calculated
and are consistent with each other Fmally, there
1s a comparison of four seasonal adjustment pro
cedures the current procedure, the ordmary and
the "fixed-factor" X-11 procedures, and the daily

Demand Deposit Ownership Survey
by Eleanor M Prmtt, Helen T Farr,
and Arthur Havenner
This paper gives a brief techmcal description
of the demand deposit ownership survey, reviewmg the System's experience with the survey over
the past 5 years and its potenual usefulness for
analytical purposes The paper also presents the
results of recent staff research on money demand
/ functions for the various ownership categories
Federal Reserve Bank of St. Louis

Sources of Data and
Method of Construction of the
Monetary Aggregates
by Darwm L Beck
Information on the various sources of data used
m the construcuon of the historical monetary
aggregate measures (1959 to 1975) 1s provided
This paper describes the vanous methods used to
estimate components of these measures that are
not reported or are reported only mfrequently
The construct10n of the narrow money stock
measure, M 1, 1s discussed m the greatest detail,
but mformat1on 1s also provided on the M 2-M 5
measures The report also mcludes mformat1on
on the mstituuons and types of holders mcluded
m each of the monetary aggregate measures

An Alternative Method of
Calculating M 1
by Anton S Nissen (Federal Reserve Bank of
New York) and Darwm L Beck
The first part of this paper discusses the theoretical nature of the alternative method proposed
by the Committee for calculatmg the money stock
and presents reasons why 1t should yield essentially the same results as the current method It
explams how the alternative method should correct for both "cash items bias" and "due from
banks bias" m the money stock and, therefore,
should s1mphfy the construction of Mi The second part of the paper describes the alternative
senes constructed by the Board's staff and the
dilemma produced by the nearly $8 0 billion d1fferei{ce between the two series m December 1974

Developing Money Substitutes Current
Trends and Their Implications for
Redefining the Monetary Aggregates
by Steven M Roberts
In the past seveTal years financial mnovat1ons
and regulatory changes have mcreasmgly blurred
the d1stmcuon between demand deposits and
savmgs-type deposits at both bank and nonbank
financial mst1tut10ns To cite a few of these mno-

Summaries of Staff Papers

vat1ons 1s sufficient to md1cate the trend NOW
(negotiable orders of withdrawal) accounts, telephone and third-party transfers from savmgs
accounts, shares of money market mutual funds
that can be transferred by check or wire, and
transfers via CBCT's (customer bank communication termmals) directly from the customer's to
the merchant's savmgs account for purchases of
goods and services In add1t10n, smce mid-1973
banks have been required to impose mterest
penalties for early withdrawal of time deposits
prior to maturity, a requirement that has sharpened the d1stmct10n between savmgs and Ume
deposits The drivmg forces behmd the changes
that have taken place have been mcreased competltlon among financial mstituuons for deposits
Federal Reserve Bank of St. Louis


(fueled by the lack of demand deposit authority
for most thnft mstituuons) relatively high mterest
rates that have mcreased the cost of foregone
mterest on demand deposits, and the existence
of mterest rate ceilmgs on time deposits In add1uon to documentmg both the changes and their
apparent causes, this paper discusses the 1mphcauons of recent developments for the mterpretat10n
of the monetary aggregates as currently defined
In the future M 1 may well tend to reflect a
decreasmg share of transactions balances while
the time deposit share of M 2 will consist of an
mcreasmg amount of deposits that are more like
secunues Such changes suggest that a new array
of monetary aggregates will need to be considered
by both monetary policymakers and economists
Federal Reserve Bank of St. Louis
Federal Reserve Bank of St. Louis
Federal Reserve Bank of St. Louis