The Domestic Adjusted Monetary Base. Richard G. Anderson and Robert H. Rasche

Transcription

1 WORKING PAPER SERIES The Domesic Adjused Moneary Base Richard G. Anderson and Rober H. Rasche Working Paper A hp://research.slouisfed.org/wp/2000/ pdf FEDERAL RESERVE BANK OF ST. LOUIS Research Division 411 Locus Sree S. Louis, MO The views expressed are hose of he individual auhors and do no necessarily reflec official posiions of he Federal Reserve Bank of S. Louis, he Federal Reserve Sysem, or he Board of Governors. Federal Reserve Bank of S. Louis Working Papers are preliminary maerials circulaed o simulae discussion and criical commen. References in publicaions o Federal Reserve Bank of S. Louis Working Papers (oher han an acknowledgmen ha he wrier has had access o unpublished maerial) should be cleared wih he auhor or auhors. Phoo couresy of The Gaeway Arch, S. Louis, MO.

2 Richard G. Anderson and Rober H. Rasche December 1999 This paper provides a consisen, monhly measure of he amoun of he U.S. adjused moneary base ha is domesically held, and of he amoun held abroad. Mos macroeconomic models ha address he role of ouside money as a deerminan of he economy s aggregae price level are closed economy models, suggesing a need o accuraely measure he domesic moneary base. To do so, his paper presens a new mehod o esimae he amoun of U.S. currency held abroad, a mehod which explois daa on he processing of currency a he Federal Reserve s 37 cash offices. Esimaes of domesic moneary aggregaes, including domesic M1 and M2, also are produced. Relaive o previous sudies and esimaes currenly included in he Flow of Funds and he Naional Income and Produc Accouns, our esimaes sugges larger currency expors during he 1970s and early 1980s, and a sharp slowing of expors since The auhors are vice presiden and economis, Federal Reserve Bank of S. Louis, and direcor of research, Federal Reserve Bank of S. Louis, respecively. Correspondence may be addressed o anderson@sls.frb.org or rasche@sls.frb.org. We hank Richard Porer for commens, and he saff of he Division of Reserve Bank Operaions, Board of Governors of he Federal Reserve Sysem, for daa.

3 The Domesic Adjused Moneary Base Economic heory predics ha he price level in a fia-money economy is anchored by ouside money, or he liabiliies of he moneary auhoriies. Early heoreical analyses of his relaionship include Gurley and Shaw (1960), Painkin (1963) and, somewha laer, Brunner and Melzer (1976) and Fama (1983); more recen general equilibrium models include King and Plosser (1984), Chari, Chrisiano and Eichenbaum (1995), and Coleman, Gilles and Labadie (1996); noable empirical examples include he series of papers by Benne McCallum and coauhors (McCallum and Hargraves, 1996; McCallum, 1988, 1990, 1993, 1995, 1997). Because hese and mos similar models omi a foreign secor, hey are in all essenial aspecs models of he deerminaion of he aggregae price level in a closed economy. Hence, applicaion of he models o real-world open economies requires one o measure separaely he domesic and foreign-held componens of a counry s sock of ouside money, ha is, of is moneary base. Such a pariion is paricularly imporan for wo of he world s larges economies Germany and he Unied Saes. 1 Ouside money for he Unied Saes economy is measured as is moneary base, which consiss primarily of he liabiliies of he Federal Reserve o households and firms. Currency in he hands of he public is he larges par of he moneary base, and expors of currency from he Unied Saes likely accoun for as much as half of he increase in he moneary base during he las decade. Reflecing is imporance, quarerly esimaes of foreignheld US currency are now included in he Federal Reserve Sysem s Flow of Funds Accouns of he Unied Saes and in he Bureau of Economic Analysis s esimaes of he ne inernaional invesmen posiion of he Unied Saes. 2 In he July 1997 revision of he Naional Income and Produc Accouns, he inroducion of esimaes of foreign-held US currency reduced he esimaed ne foreign asse posiion of he Unied Saes by more han $300 billion. This paper presens measures of he domesic and foreign-held componens of he U.S. adjused moneary base. 3 The principal problem in building such measures is separaing domesic and foreign holdings of U.S. currency. No direc observaions are available regarding he 1 On Germany, see Deusche Bundesbank (1995). Anecdoal repors also sugges ha large numbers of Swiss 1000 fr noes are held ouside Swizerland bu measures of he amoun are very enaive; see Andris (1997). 2 On he former, see Board of Governors (1996); on he laer, see Bach (1997) and Scholl (1997). 3 Tha is, he moneary base adjused for he effecs of changes in Federal Reserve sauory reserve requiremens on he quaniy of base money demanded by banks. 1

4 amoun of U.S. currency held abroad, nor of he oal amouns shipped o and received from oher counries. 4 Anecdoal repors sugges ha large numbers of US Federal Reserve noes in mos denominaions are in coninual foreign circulaion, and ha a leas some porion of each denominaion flows ino and ou of he Unied Saes wih business ravelers and ouriss on a more or less regular basis. Ye, hese and oher repors also sugges ha he overwhelming majoriy of he dollar value of foreign-held US currency is comprised of $50 and $100 noes ha end o remain abroad once here, and are held no only as a medium of exchange bu also as a hedge agains uncerain inflaion and poliical insabiliy. 5 In an imporan sudy, Porer and Judson (1996) examine eleven differen mehods o esimae he annual ne flows of US currency ino foreign circulaion during Combining hese flow esimaes wih a benchmark assumpion ha abou 50 percen of he sock of US currency ouside banks was held abroad a he end of 1976, hey conclude ha, by he end of 1995, he proporion of US currency held abroad had increased o abou 55 percen; 44 percen of he 55 percen is in he form of $100 Federal Reserve noes. 7 Their published esimaes have several shorcomings ha make hem unsuiable for consrucing a domesic moneary base, or any oher domesic moneary aggregae: no esimaes are available before 1977; he esimaes are only for annual ne ouflows; he benchmark assumpion ha half of US currency was held abroad in 1976 is implausible; and, heir preferred median-flow esimaor is impossible o updae in a imely fashion. The ne ouflow of U.S. currency o foreign circulaion has also been considered by Feige (1994, 1996). His preferred mehod, he shipmens proxy, is based on hree assumpions: all (or a leas, mos) foreign-held US currency is $100 noes; all (mos) expors and im- 4 Porer and Judson (1996) s foreign currency shipmens mehod uilizes specific, firm-level daa on he amouns of currency shipped abroad. The firms ha furnish he daa o he Federal Board saff regard he daa as proprieary, and hence he daa are no available o he public nor o oher researchers. 5 Bach (1997), p. 49, noes ha mosly lower denominaion noes ($5s, $10s, $20s and $50s) circulae in he U.S. economy, whereas mosly $100 noes circulae abroad. See also Porer and Judson (1996). 6 The mehods are summarized in Porer and Judson (1996), Table 4. The elevenh mehod, median flow, is no an independen mehod, bu raher a summary mehod, defined over he oher en. In correspondence, Porer noes ha he correlaions among he esimaes obained by he median-flow, cash-office, and shipmens-proxy mehods exceed 0.95, and ha he median-flow esimae is he cash-office or shipmens-proxy esimae in abou hree-quarers of heir observaions. Ye, in heir aricle, hey clearly focus on he medianflow mehod: he only annual flow daa in he aricle are he median flow esimaes. Subsequen publicaions, such as he Flow of Funds and he Naional Income and Produc Accouns, conain esimaes based on he shipmens-proxy mehod, which may be updaed promply each monh. 7 Porer and Judson (1996), p , Table 5. 2

5 pors of $100 noes occur a one, or perhaps wo, Federal Reserve Sysem cash offices; and none (or almos none) of he ne ouflow of $100 noes a hose offices reflecs changes in he domesic demand for $100 noes. 8 The esimaed amouns of foreign-held US currency shown in he Flow of Funds and he Naional Income and Produc Accouns are obained by applying his mehod o he New York Ciy and Los Angeles cash offices ne ouflows of $100 noes (Bach, 1997). 9 These esimaes, as well as Porer and Judson s, are discussed more fully below. The balance of he paper is organized as follows. In he following secion, we presen a model o pariion ne shipmens of currency ino he proporions enering domesic and foreign circulaion, respecively. In he second secion, we use he model o consruc esimaes of domesic and foreign currency holdings. In he hird secion, we compare hese esimaes o hose in previous sudies and hose included in he Flow of Funds and Naional Income and Produc accouns. In he fourh secion, we compare he implied domesic adjused moneary base o he oal adjused moneary base, and examine implicaions for moneary policy. Finally, we use our esimaes of foreign-held currency o consruc domesic measures of he moneary aggregaes M1 and M2. 1. A Model o Esimae Foreign Holdings of US Currency Our model o esimae he amoun of foreign-held U.S. currency uilizes daa on he receips and shipmens of currency, by denominaion, a he Federal Reserve s 37 cash offices. 10 We separae he ne ouflows of $50 and $100 Federal Reserve noes from hese offices ino a porion ha remains in domesic circulaion and a porion ha is expored ino permanen for- 8 The shipmens proxy is also examined by Porer and Judson (1996). 9 The mehod used in he Flow of Funds Accoun has apparenly changed hrough ime. Inroducory noes in he December 1996 Flow of Funds release (Z.1) refer he reader o Porer and Judson (1996); inspecion confirms ha he Flow of Funds daa are Porer and Judson s median-flow esimaes. Ye, in he Sepember 1997 release (for he second quarer of 1997), he Flow of Funds hisorical daa were changed o mach hose published in he July 1997 Survey of Curren Business. The laer, according o Bach (1997), are no he median-flow esimaes bu raher are a new se of daa obained by applicaion of he shipmens proxy mehod o ne ouflows of $100 noes a he New York Ciy and Los Angeles cash offices. This change in he mehod used o esimae foreign-held currency is no menioned in he Flow of Funds Z.1 release. 10 The Federal Reserve ships and receives currency a 37 cash offices naionwide (primarily Federal Reserve Banks and heir branches). Daa for hese offices are available on Federal Reserve elecronic daa bases beginning January, Hisorical daa back o January, 1958 are available on archival microfilm. Our esimaes uilize boh hese sources of daa. 3

6 eign circulaion. 11 This mehod furnishes monhly esimaes of boh he flow of U.S. currency o foreign circulaion and, condiional on our assumed benchmarks in 1965 and 1969, he oal foreign-held sock of U.S. currency. Because all required daa are available shorly afer he end of each monh, hese esimaes may be updaed on he same schedule as oher currenly published moneary aggregaes. 12 In our mehod, he fundamenal daa are he number of pieces of currency of each denominaion pu ino circulaion (E = emissions) and received from circulaion (R = receips) each monh by Federal Reserve cash offices. These flows ino and ou of circulaion are relaed o he amoun of currency in circulaion of a paricular denominaion, C, by he ideniy: (1) C E - R. Currency of a paricular denominaion pu ino circulaion (E ) eiher circulaes domesically ( E D ) or is expored and circulaes abroad ( E F ). Because no direc esimaes of eiher expors of U.S. currency nor he foreign-held sock exis, some se of idenifying assumpions is necessary. The assumpions we choose are: Firs, ha currency once expored ends o say abroad and hence has been permanenly removed from domesic circulaion. This allows us o assume ha currency received from circulaion by Federal Reserve cash offices reflecs (almos) exclusively domesic circulaion, D ha is R = R. Second, ha small denominaion noes $1s, $5s, and $10s carried abroad end o circulae ino and ou of he U.S., and hence ino and ou of Federal Reserve cash offices, in a manner similar o he inernal domesic circulaion of he same denominaion noes. Third, ha he emissions and receips paerns of small denominaion noes a Federal Reserve cash offices are good measures of he unobservable emissions and receips paerns of he large denominaion noes in domesic circulaion. In paricular, for reasons examined 11 Our mehod seeks o esimae he share of U.S. currency ha ends o remain abroad in coninual circulaion, eiher as a medium of exchange or as a sore of value. Our mehod is robus o rouine inflows and ouflows of small-denominaion currency; see he discussion of our Assumpions 1 and Daa on moneary aggregaes during a given monh are firs published by he Board of Governors on he second Thursday afer he Monday dae of he week, ending on a Monday, ha conains he final calendar day of he previous monh. On average, new monhly daa appear abou 2 weeks afer he end of he respecive monh. 4

7 below, we rely on he paern of emissions and receips of $10 noes a he New York Ciy cash office o consruc our esimaes of he foreign circulaion of large denominaion noes. Assumpion 1: Currency Abroad Tends o Say Abroad Anecdoal evidence suggess ha much U.S. currency held abroad is infrequenly repariaed o he Unied Saes. Alhough some currency eners and leaves he U.S. each day wih businessmen and ouriss, and some currency cerainly is reurned o he Unied Saes when foreign socks exceed desired levels, U.S. currency abroad acs as boh a sore of wealh and medium of exchange. Surges in currency expors have ended o be correlaed wih increases in economic and/or poliical insabiliy and, because of is dual role as a medium of exchange and sore of value, currency may be reained abroad as a hedge agains fuure insabiliy even afer he immediae crisis subsides. Phrased somewha differenly, once households and firms are induced by poliical and economic insabiliy or by ransacions needs o allocae some par of heir porfolio o U.S. currency, ha share perhaps ends o change slowly. In our analysis, we seek o idenify he proporion of US currency ha is in coninual circulaion, or permanenly held, abroad. We assume, as an idenifying assumpion, ha here is a permanen and ransiory componen o foreign-held US currency. As a maer of definiion, he permanen componen reflecs currency which is in coninual circulaion abroad and hence does no flow hrough Federal Reserve cash offices. 13 We assume ha currency held emporarily abroad, say due o ourism or business ravel, reurns o he Unied Saes (and hence o Federal Reserve cash offices) wih he same ransi ime as currency in domesic circulaion; an alernaive, equivalen assumpion is ha he raio of currency emporarily held abroad o currency in domesic circulaion is consan. Condiional on hese assumpions, receips of currency a Federal Reserve cash offices reflec he amoun of currency in domesic circulaion (ouside banks) plus any currency emporarily held abroad, and he mehodology developed herein provides a measure of he amoun of U.S. currency ha is in coninual circulaion ouside of he Unied Saes. Assumpion 2: Domesic and Foreign Circulaion Paerns for Small Denominaion Noes Our second assumpion implies ha, hrough ime, emissions of small denominaion noes from Federal Reserve cash offices should be approximaely equal o receips (ne of he growh of he economy), and ha he seasonal variaion in he quaniy of hese noes in circula- 5

8 ion should vary relaively lile hrough ime; ha is, ha he paern of Federal Reserve cash office currency processing aciviy for small denominaion noes should be uncorrelaed wih boh he level and growh rae of foreign-held U.S. currency. E The raio of emissions o receips,, for $1, $5, $10 and $20 noes a Federal Re- R serve cash offices naionwide and a he New York Ciy cash office are shown in Figure 1. For aggregae naionwide daa, he raios for he firs hree denominaions display he sable paerns suggesed by assumpion 2. The raio for $20 noes displays a relaively less consan mean and seasonal paern, boh naionwide and a New York Ciy. On balance, we conclude ha $10 noes are he larges denominaion wih a relaively sable mean and seasonal paern, boh naionwide and a he New York Ciy cash office, and hence are mos suiable for our analysis. Figure 2 focuses on he raios of emissions o receips for $10 noes, a all Federal Reserve cash offices ouside New York Ciy, E R non NYC, in he op panel, and a New York Ciy, E R NYC, in he cener panel, for January 1965 Augus, (The daa are monhly, no seasonally adjused.) The boom panel of he Figure shows he raio of emissions o reurns a he New York Ciy office divided by he raio a all oher cash offices, E R NYC E R non NYC. Alhough here is some irregulariy in he New York Ciy daa during he lae 1980s, his compound raio displays several disinc characerisics: (1) a srong bu remarkably consan seasonal paern; (2) no disinc rend; and, (3) a mean for he numeraor, he New York Ciy cash office raio, equal o 1.11, abou en percen higher han he mean of he denominaor, he aggregae of he oher 36 Federal Reserve cash offices, equal o Noe ha because he mean of he denominaor is no significanly differen from 1.0, all growh in he ousanding sock of $10 Federal Reserve noes in circulaion has come (algebraically) from he New York Ciy office. 13 We of course do no mean ha precisely he same paper noes say abroad (alhough many perhaps do), bu raher ha he aggregae value held abroad ends o be sable. 6

9 Assumpion 3: The Domesic Circulaion of Large Denominaion Noes Our hird idenifying assumpion is ha he emissions-o-receips paerns a Federal Reserve cash offices of large denominaion ($50 and $100) noes in domesic circulaion can be measured by he emissions-o-receips paern of smaller denominaion noes. We believe ha i is reasonable o assume ha large denominaion noes in domesic circulaion will pass hrough Federal Reserve cash offices in he same fashion as smaller denominaion noes ($1, $5, $10). Since here are no daa on he separae domesic circulaion paerns of large and small denominaion noes, his assumpion is no esable bu i is suppored by he daa analyzed below. 14 Our subsequen analysis is based on he New York Ciy cash office raio, E R NYC, for $10 noes. We choose he New York Ciy raio, raher han he non-new York raio, so as o generae a conservaive esimae of he amoun of currency going o foreign circulaion, ha is, we prefer o under, raher han over, esimae he amoun being shipped abroad. (The reason for his bias will become clear below, when we compare our esimaes o ohers.) The New York Ciy cash office raio provides such a conservaive esimae because i allows for more growh in he domesic circulaion of large denominaion noes han would be obained if we used he naionwide raio for all cash offices: recall ha he aggregae daa for all cash offices ouside New York Ciy shows no secular growh in he ousanding sock of $10 Federal Reserve noes since Specifically, for large-denominaion Federal Reserve noes in domesic circulaion, le he raio of emissions o receips be denoed as: E (2) µ DL L = DL R 14 Some readers have objeced o his assumpion on he grounds ha he domesic velociy of circulaion of large denominaion noes may be much smaller han for small denominaion noes. This is a misinerpreaion. Our mehod relies on he raio of emissions o receips of noes a Federal Reserve cash offices, and no a all on he raios of emissions and/or receips of noes o he ousanding numbers of noes or o measures of aggregae economic aciviy. The appropriaeness of our mehod is independen of differences by denominaion in he velociy of circulaion of domesically held currency. One oher reader has objeced by assering ha banks end o reurn small denominaion noes o he Federal Reserve for processing more frequenly ha large denominaion noes, perhaps because newly issued small denominaion noes deeriorae in circulaion more rapidly han large denominaion noes. This objecion also is a misinerpreaion. Even if he proporion of noes ha banks reurn o he Federal Reserve differs by denominaion (and i likely does), our resuls which depend on he raios of Federal Reserve cash office shipmens o receips by denominaion would be affeced only if he endency for banks o reurn noes o he Federal Reserve has changed differenially by denominaion hrough ime. We doub ha his has occurred. 7

10 DL where E are emissions o domesic circulaion, DL R are receips from domesic circulaion, and he superscrips and subscrips are: D denoes domesic flows, F foreign flows, L large denominaion noes, and S denoes small denominaion noes. Our esimaor $µ L is he raio of S emissions ( E ) o receips ( S R ) of $10 denominaion noes a he New York Ciy cash office: S E (3) $µ L = µ S =, S R L D Recall from assumpion 2 ha R = R, ha is, receips of large denominaion noes from (permanen) foreign circulaion are zero by assumpion. An esimae of emissions of large denominaion noes o domesic circulaion is: (4) $ DL E =µ $ L R L or: (5) E$ DL L E L R $ = µ L L. E L DL FL Since E = E$ + E$, L (6) E = µ $ LR L + E$ FL and esimaed emissions o permanen foreign circulaion are: (7) $ FL L E = E µ $ L R L or: (8) E$ FL L E L R $ = 1 µ L L. E Boh E DL and E FL are emissions, and hence necessarily are > 0. Since R L is receips, i is also necessarily > 0. The facor µ S is > 0, since he raio E S S R is a posiive number. By (5) i is guaraneed ha he esimaed gross emissions (shipmens) of large denominaion noes o domesic circulaion is posiive, E $ DL > 0. This is, of course, reasonable: rue gross shipmens, E DL, canno be negaive. However, since he esimaed foreign shipmens, FL $E, is compued as he difference beween wo posiive numbers, here is no guaranee ha i 8

11 will be nonnegaive. We regard negaive values of E $ FL as a reflecing eiher daa errors or specificaion errors in our model. Esimaed foreign gross shipmens will be negaive, E $ FL < 0, when he raio $ DL E > 10,. ha is, when esimaed emissions o domesic circulaion exceed oal emis- L E FL sions of large denominaion noes. Because E < 0 is no feasible (acual gross shipmens of currency are always nonnegaive), we regard values of E $ FL < 0 as inadmissible and impose he ˆ DL E resricion in our mehod ha 1 or equivalenly E $ FL L 0. E The mos likely cause of E $ FL < 0 is ha, conrary o our assumpions, some receips of large denominaion noes a Federal Reserve cash offices came from foreign circulaion: Recall ha $ FL L E E $ R$ DL = µ, ha $ DL L R R R $ FL =, and ha R $ FL = 0 for all by assumpion. L Even if $ FL E R $ FL > for all, such ha he sock of large denominaion U.S. currency held abroad never decreases, some large denominaion noes received a he cash offices migh have come from abroad, ha is, R $ FL > 0. As wih mos foreign currency ransacions, such foreign receips, if any, are no observable and hence ineviably remain as esimaion error. Finally, noe ha FL $E < 0 should occur less frequenly for noes ha are more heavily in demand abroad ha is, for noes where he assumpion R $ FL = 0 is rue for more values of (or a leas R $ FL is very small) and should occur more frequenly for noes ha are less heavily (or decreasingly) in demand abroad; in wha follows, his is exacly he relaive paern we find for $50 and $100 dollar noes, respecively. 2. Esimaes of Foreign-Held Large Denominaion Federal Reserve Noes The monhly raios of emissions o receips a he New York Ciy cash office and a all oher cash offices for $10, $50 and $100 Federal Reserve noes are shown in Figure 3. In his figure, he raios for $10 and $100 noes begin January 1965, and he raio for $50 noes in January

12 A. $100 Noes The emissions-o-receips raios for $100 noes are much differen han hose for he $10 noes. A boh he New York Ciy cash office and he oher cash offices, he emissions-oreceips raio is considerably in excess of 1.0 hroughou he sample period (averaging 2.3 for he New York office and 1.3 for he ohers). Such high raios of emissions o receips are no characerisic of he smaller ($1, $5, $10) denominaion noes. This conras beween he raios for small and large denominaion noes srenghens he case for our assumpion 2. A comparison of he emissions-o-receips raios for he New York Ciy and he oher cash offices suggess several reasons o quesion he shipmens-proxy assumpions of Feige (1996) and of Porer and Judson (1996) ha all emissions of $100 noes o foreign circulaion came from he New York Ciy office. 15 Before he 1980s, he emissions-o-receips raios a he New York Ciy and non-new York Ciy cash offices have approximaely he same mean. This lends suppor o he view ha expors of currency were more closely linked o business ravel and ourism han o poliical and economic insabiliy abroad, and hence also were perhaps more widely disribued among cash offices. Alhough no srong, here appears o be a sligh posiive rend in he emission-o-receips raio a he New York Ciy cash office, a leas prior o he 1990s, and a sligh negaive rend a all oher cash offices, suggesing ha expors have ended o become more concenraed a he New York Ciy cash office during our sample period, from An implicaion of his rend is ha a leas some significan porion of expored currency mus have come from non-new York Ciy offices prior o Addiional suppor comes from comparing he raios of emissions-o-receips of $100 and $10 noes a New York Ciy, E R $100 $100 NYC E R $10 $10 NYC o he comparable raio for non-new York Ciy cash offices, 15 The shipmens proxy used o obain esimaes for he Flow of Funds Accouns and he Naional Income and Produc Accouns includes he Los Angeles cash office, as well as he New York Ciy cash office. These daa are examined furher laer in his aricle. 16 Noe ha, o he exen his is rue, esimaes based primarily on shipmens from he New York Ciy cash office will end o underesimae he amoun of currency expored; his perhaps also explains, in par, why Porer and Judson s esimaed foreign-held shares of US currency decrease during he 1970s and 1980s. This opic is explored furher below. 10

13 E R $100 $100 non NYC E R $10 $10 non NYC. If lile of he currency shipped from cash offices ouside New York Ciy was expored, hen one migh expec he emissions-o-receips raio for he oal of hese offices o more closely resemble heir raio for small-denominaion currency. In fac, he raio for offices ouside of New York Ciy seems oo large o suppor he posiion ha all foreign shipmens of $100 noes originaed from he New York Ciy cash office. Focusing on he New York Ciy emissions-o-receips raio, inadmissible occurrences of $ DL E. L E > 10 (ha is, E $ FL < 0 ) based on equaion (5) appear fairly ofen for January observaions, as shown in Figure 4. Albei parially seasonal, his volailiy also perhaps reflecs he ebb and flow of currency expors in response o evens abroad: Inadmissible values are more frequen during periods of slower expors, such as he mid-1980s, and less frequen during acceleraions of expors, such as in he early 1990s. As discussed above, we cap (ha is, allow a maximum E value of) $ DL a 1.0 and apply equaion (7) o he emissions and receips of $100 noes a he L E oal of all Federal Reserve cash offices during January 1965 Augus Esimaed ne ouflows of $100 noes o foreign circulaion are shown in Figure 5, and cumulaive socks of domesically and foreign-held $100 noes are shown in Figure 6; he laer are consruced by assuming ha he quaniy of foreign-held $100 noes was zero in December As described above, he ne change in he domesic sock of $100 noes equals he esimaed domesic emissions, $ E DL, minus oal receips of $100 noes, R. The ne change in he foreign-held sock of $100 noes equals he esimaed foreign emissions, $ E FL, because, by assumpion, receips from (permanen) foreign circulaion are zero. 18 Our esimaes sugges ha of he $336 billion (ne) of $100 noes ha were emied o circulaion from January 1965 hrough Augus 1999, only an esimaed $116 billion remained in domesic circulaion, while an esi- 17 To examine he robusness of our resuls o his consrain, we have also examined quarerly and annual raios of currency emissions o receips. A he quarerly level of aggregaion, very few observaions of he emissions-o-receips raio exceed one, and hose few observaions are very close o one. Our esimaed flows o foreign circulaion are lile changed. These resuls are available from he auhors on reques. 18 By definiion, any $100 noes reurned o U.S. banks, and hereafer o he Federal Reserve, canno be in coninual, permanen circulaion abroad. 11

14 maed $220 billion were expored abroad. On Augus 1999, he esimaed cumulaive expors of $100 noes accouned for abou 45 percen of oal US currency held ouside banks (he currency componen of M1). The benchmark from which his esimae is consruced is discussed furher below. 19 B. $50 Federal Reserve Noes In addiion o $100 noes, some repors sugges ha significan quaniies of $50 noes are in permanen foreign circulaion. The raios of emissions-o-receips for $50 Federal Reserve noes a he New York Ciy cash office and a all oher cash offices are shown in Figure 3 for January 1969 Augus The characerisics of hese raios sugges ha: pas expors of $50 noes are oday a significan porion of foreign-held US currency; ha mos of hese expors came from he New York Ciy cash office; and, ha relaively few $50 noes have been expored in recen years, relaive o expors of $100 noes. The firs noable characerisic of hese raios is heir size. A boh he New York Ciy and oher cash offices, he raios are considerably in excess of 1.0 hroughou he sample period, averaging 2.2 for he New York office and 1.1 for all oher offices. Furher, he mean of he emissions-o-receips raio for he New York Ciy cash office, 2.2, is almos equal o he mean of ha office s emissions-o-receips raio for $100 noes, 2.3, during January 1965 Augus The emissions-o-receips raio for he New York Ciy cash office does no exhibi any rend, while he raio a all oher cash offices has a negaive rend and is close o 1.0 in recen years, suggesing few expors from cash offices oher han New York Ciy. Inadmissible values of $ DL E. L E > 10 (ha is, E $ FL < 0 ) are observed more frequenly for $50 noes han is he case for $100 noes (see Figure 7). Before 1983, hese observaions occur mosly in January; since hen he fracion of such observaions during each year has been increasing. This increasing frequency reinforces oher evidence which suggess ha in recen years relaively few $50 noes have been expored. As above, we impose a maximum value of 1.0 on 19 The benchmark from which his esimae is consruced assumes ha he quaniy of foreign-held $100 noes was zero in December Our esimae is robus o he benchmark assumpion. Because he oal value of $100 Federal Reserve noes in circulaion in December 1964 was only $7.6 billion, even if fify percen of he sock of such noes a ha ime was held abroad, which seems unlikely, he addiional accumulaion of esimaed expors of $100 noes would only increase he fracion of oal currency held abroad o 47 percen. 12

15 DL $E and apply equaion (7) o esimae emissions and receips of $50 noes o permanen foreign L E circulaion. Esimaed ne ouflows of $50 noes are shown in Figure 8, and he accumulaed socks of domesically and foreign-held $50 noes (condiional on he assumpion ha no such noes were held abroad in December 1968) are shown in Figure 9. Of he $49.9 billion of $50 noes ha were emied o circulaion from January 1969 hrough Augus 1999, we esimae ha $20.3 billion remain in domesic circulaion, while an esimaed $29.6 billion are held coninually abroad. On Augus 1999, he fracion of currency held by he nonbank public accouned for by foreign holdings of $50 noes is approximaely 6 percen. This esimae is consruced by benchmarking oal $50 noes ouside he Unied Saes a he end of 1968 o zero. 21 C. Growh Raes of Toal, Domesic and Foreign-Held Currency Monhly growh raes of he esimaed domesic and foreign currency componens (of oal currency in M1), seasonally adjused, are shown in Figure The esimaed foreign componen displays significan seasonaliy, especially before This seasonaliy perhaps reflecs a largely ransacion-based demand for, and use of, U.S. currency abroad during he early par of our sample; ha is, he quaniy of U.S. currency abroad flucuaed wih seasonal flucuaions in business aciviy as i moved ino foreign counries. During he laer pars of our sample, and especially afer 1980, he increased demand for U.S. currency abroad as a sore of value caused by poliical unres and inflaion insabiliy migh end o mask seasonal flows, paricularly when holdings reained abroad are growing rapidly. 24 The well-known srong monhly seasonaliy of currency demand suggess an addiional es on he reasonableness of our esimaes. If he quaniy of currency demanded by US resi- 20 As noed elsewhere in his analysis, emissions-o-receips daa exis in machine-readable daabases beginning January Daa for before 1974 were colleced by hand from microfilm. Alhough daa exis prior o 1969, anomolies in he daa sugges ha he daa are unreliable before As wih he $100 Federal Reserve noes, his esimae is robus o he benchmark assumpion. Since he oal value of $50 Federal Reserve noes held by he public a ha ime was only 4.19 billion dollars, even if fify percen of he sock of such noes a ha ime were held abroad, which seems unlikely, he addiional accumulaion of esimaed expors of $50 noes would increase he fracion of oal currency held abroad by less han one percen. 22 Simple monhly percenage change a annual rae. 23 Because of he major shifs in level, he foreign currency daa are no direcly seasonally adjused bu raher are obained as he difference beween oal and domesic currency, each seasonally adjused. This perhaps also conribues some residual seasonaliy. 24 We are indebed o Richard Porer for his inerpreaion of he daa. 13

16 dens is seasonal while he quaniy demanded by foreign residens is no, hen changes in he ampliude of seasonal facors should be correlaed wih changes in he foreign-held proporion of US currency. A well-esimaed domesic currency componen should, perhaps, display relaively consan monhly flucuaions. Currency seasonal facors are shown in Figure 11; he upper panel shows he seasonal adjusmen facor for he oal currency componen of M1 as published by he Board of Governors, while he lower panel shows a facor for our domesic currency esimaed via X11 (wih sandard defauls). The ampliude of he seasonal facor for all currency decreases rapidly during he laer par of sample, presumably reflecing he increasing share of U.S. currency in coninual circulaion abroad. In conras, he esimaed seasonal flucuaions in domesic currency display a near-consan ampliude. During he 1990s, when large currency ouflows are alleged o have significanly changed he esimaed seasonal facors for oal currency (Porer and Judson, 1996), he esimaed facors for our domesic currency are essenially he same as hose in he lae 1960 and early 1970s. Separaing he domesic and foreign componens of U.S. currency growh has a significan impac. In recen years, foreign currency shipmens have accouned for a large par of monhly flucuaions in currency growh. The recen slowdown of foreign shipmens and acceleraion of domesic currency growh also are apparen. 3. Comparison o Oher Esimaes In his secion, we compare our esimaes firs o hose obained by Porer and Judson (1996), and nex o hose currenly published in he Flow of Funds and he Naional Income and Produc Accouns. A. Median-Flow Esimaes of Porer and Judson (1996) In he mos exensive published analysis, Porer and Judson (1996) examine eleven differen mehods o esimae he ne ouflow of currency from he U.S. ino foreign circulaion during Condiional on a benchmark assumpion ha a he end of 1976 abou half of US currency was foreign-held, heir esimaed ouflows sugges ha, a he end of 1995, abou 55 percen of he oal sock of U.S. currency held by he nonbank public was held abroad, 44 percen of which was in he form of $100 noes (Porer and Judson, 1996, p ; Table 5). In he inroducion, we noed several reasons why Porer and Judson s median-flow mehod esimaes are no saisfacory for consrucing a domesic adjused moneary base, nor any oher domesic moneary aggregae. In his secion, we expand our concerns regarding, heir 14

17 assumed benchmark. Alhough a benchmark assumpion is necessary o consruc any esimae of foreign-held socks from flows, heir choice is inappropriae: Absen direc daa on he sock of foreign-held currency during any poenial benchmark period, he only saisfacory benchmark is a period sufficienly early in ime so ha one may reasonably assume ha he foreign-held sock was approximaely zero. Their benchmark does no saisfy his crierion. 25 To compare our esimaed monhly ne flows ino foreign circulaion o he annual flows repored by Porer-Judson (for he shorer ime span ), we sum our monhly flows from December o December. These annual flows, and Porer and Judson s preferred median-flow esimaes, are shown in Figure Through 1988, our esimaes of he amoun of currency expored annually are always larger han he median flow esimaes of Porer and Judson; in he 1990 s heir amouns are larger han ours. Neverheless, he wo series end o move ogeher quie closely. Alhough he esimaed flows differ somewha, year-o-year acceleraions and deceleraions are quie close. The larges excepion is , where our daa show a much more rapid decrease in he rae of currency expors han is shown by he Porer-Judson daa. Esimaing he sock of US currency held abroad is a more suble maer han esimaing ouflows. In heir aricle, Porer and Judson emphasize ha heir preferred median-flow mehod suggess a foreign-held share of 55 percen, as of December Bu, like all sock esimaes, his esimae is condiional on he seleced iniial benchmark. Lacking a direc measure, Porer and Judson consider wo polar-opposie alernaives, as of December 1976: Firs, ha no U.S. currency was held abroad and, second, ha all US currency in circulaion ouside banks was held abroad. When combined wih heir median-flow esimaes, hese alernaive benchmarks imply wo ime series of he implied shares of US currency held abroad during Porer and Judson selec he mid-poin beween hese series, shown in Figure 13. Noe ha Porer and Judson s benchmark produces esimaed foreign-held shares for ha are approximaely he same as assuming ha half of he ousanding U.S. currency sock has been held abroad each year since We find esimaes based on Porer and Judson s benchmark implausible, for wo reasons. Firs, boh anecodal evidence and domesic surveys sugges ha foreign holders prefer large de- 25 Porer and Judson s analysis is discussed more fully in a separae appendix, available on reques from he auhors of his paper. 26 Porer and Judson (1996), Table 6. For each year, he median-flow esimae is he median of he flows suggesed by Porer and Judson s en esimaors. 15

18 nominaion noes, while domesic residens hold relaively few such noes. 27 In December 1976, $81.0 billion in US currency was in circulaion ouside deposiory insiuions, of which 32 percen was $100 noes ($26.7 billion) and 11 percen was $50 noes ($9.0 billion); by December 1995, $376.2 billion was in circulaion, of which 64 percen was $100 noes ($241.5 billion) and 12 percen was $50 noes ($46.4 billion). Hence, during he period sudied by Porer and Judson, large denominaion noes increased four imes as rapidly as he sum of all smaller denominaions, and almos doubled heir share of currency held by he nonbank public. Ye, Porer and Judson s esimaed share of US currency held abroad decreases hroughou he 1980s, and is only slighly larger in December 1995 han in December Second, consider he problem of algebraically accouning for he half of currency in circulaion as of December 1976, or $40.5 billion, ha Porer and Judson assume was held abroad. This amoun, a ha ime, would equal he sum of all $100 and $50 noes in circulaion, plus abou one-sixh of all $20 noes. Alernaively, if domesic residens are assumed o have held one-fourh of $100 and $50 noes, he amoun would encompass half of all he $20 noes in circulaion, in addiion o he remaining hree-quarers of $100 and $50 noes. Such high proporions seem improbable, given anecdoal evidence and he exensive shipmens of large-denominaion currency from he US during he lae 1980s and early 1990s. 29 In conras o Porer and Judson, we consruc socks of foreign-held US currency by accumulaing our esimaed ne ouflows of $100 and $50 noes from an assumed benchmark amoun of zero, beginning, respecively, in December 1964 and December Alhough some US currency cerainly was foreign-held on hese daes, our sock esimaes are neverheless robus o our benchmark assumpion. In December 1964, he oal value of $100 and $50 noes in circulaion was $11.8 billion (versus $35.7 billion in December 1976). Even if half, raher han zero, of hese noes had been held abroad which seems unlikely he effec on our esi- 27 This evidence is surveyed by Porer and Judson (1996). Bach (1997, p.49) noes ha a 1995 survey of U.S. households found ha hey could accoun for a mos a lile more han 3 percen of oal holdings of $100 noes. 28 In one mehod ha exends back in ime o earlier years, Porer and Judson noe ha heir seasonal mehod suggess ha 40 percen of U.S. currency was held abroad as early as We find his esimae also implausible. Toal currency held by he nonbank public in January 1960 was $28.7 billion. Fory percen of his oal is a dollar amoun equal o he sum of all $50 and $100 noes in circulaion plus half of he $20 noes. See Banking and Moneary Saisics , p This analysis suggess an alernaive benchmark, no considered by Porer and Judson: selec he midpoin beween he shares implied by zero or all $100 noes held abroad. This alernaive suggess ha 16.5 percen of currency was abroad as of December 1976 (oal currency in circulaion ouside banks as of December 1976 was $81 billion, of which $26.7 billion was in $100 noes). 16

19 maed foreign-held sock is minimal: by December 1995, he share of US currency held abroad increases by less han 2 percen, relaive o our esimaed 52 percen. The robusness is a direc resul of using a benchmark dae sufficienly early so ha he amouns of $100 and $50 noes in circulaion are small, relaive o subsequen issuance. In summary, for flows of currency o foreign circulaion, our mehod relaive o Porer and Judson (1996) suggess relaively larger ouflows during he 1970s and mos of he 1980s, abou he same annual ouflows during he lae 1980s and early 1990s, and significanly slower currency ouflows during he mid-1990s. For he foreign-held sock of US currency, we esimae ha in 1977 abou one-sixh of US currency was held abroad, versus Porer and Judson s assumpion of one-half; during he 1980s, we esimae ha he foreign-held share increased almos coninuously, while he Porer and Judson esimae ha he proporion decreased up o abou 1989; during he 1990s, we find a much sharper slowing of foreign demand for U.S. currency han is repored in Porer and Judson. A he end of Porer and Judson s sample period, our esimaes are remarkably close o heirs, despie difference in esimaion mehods. Porer and Judson esimae ha foreign holdings of U.S. currency were abou 55 percen of he oal currency held by he nonbank public in December 1995; our esimae, based on $50 and $100 noes, is 53.2 percen. Of heir 55 percen, Porer and Judson esimae ha 44 percenage poins was accouned for by $100 noes; we esimae 46 percen. 31 For $100 noes alone, Porer and Judson esimae ha 74 percen of ousanding $100 noes were held abroad a he end of 1995; accumulaing our esimaed emissions of $100 noes o foreign circulaion since December 1964 suggess 72 percen. B. The Shipmens Proxy: Feige (1994, 1996), Survey of Curren Business, and Flow of Funds Accouns Feige (1994, 1996) also consruced esimaes of he fracion of U.S. currency held abroad. Among his various mehods is he shipmens proxy. 32 In ha mehod, he assumes ha he enire ne ouflow of $100 Federal Reserve noes from he New York Ciy cash office is ex- 30 These daes are pragmaically chosen as he earlies for which suiable currency processing daa are available. 31 Porer and Judson do no aribue o any specific denominaion he 11 percenage poins no accouned for by $100 noes. Our examinaion of $50 noes suggess ha abou an addiional 7.2 percenage poins may be accouned for by foreign holdings of $50 noes. 32 The shipmens proxy also is among he mehods considered by Porer and Judson. Alhough heir aricle does no include enough deail o permi a direc comparison, in privae correspondence Richard Porer indicaed ha he correlaion beween heir median-flow and shipmen proxy s esimaes (applied solely o New York Ciy daa) is approximaely

20 pored and remains in coninual circulaion ouside of he U.S., and ha such expors from New York are he sole source of currency leaving he counry. 33 In noaion similar o ha inroduced in secion 2, he foreign-held sock of US currency is obained by accumulaing he ne emissions L, NYC L, NYC ( E R ) of $100 noes from he New York Ciy cash office. The esimaed flows of US currency o foreign circulaion ha were inroduced ino he Flow of Funds and he Naional Income and Produc Accouns during he laer half of 1997 are based on he shipmens proxy, excep ha daa from boh he New York Ciy and Los Angeles cash offices are included, beginning January Esimaes of he socks of foreign-held US currency are obained by accumulaing he ne quarerly flows from a benchmark which assumes ha he foreign-held sock was 49.2 percen of oal US currency in circulaion as of December Overall, hese more recen annual flow esimaes are very similar o Porer and Judson s median-flow esimaes (Figure 12); our concerns regarding he laer apply o he former. The close correlaion beween he wo ses of esimaes is no myserious, for wo reasons. Firs, prior o 1990, emissions and receips of $100 noes a he Los Angeles cash office were approximaely equal, and ne ouflows from he office were small. Only during he second half of 1991 did hese begin o diverge, when he Los Angeles cash office began receiving considerably more $100 noes han i emis o deposiory insiuions, a siuaion ha persiss. The reasons for his change are uncerain: perhaps he noes came (and are coming) from he Far Eas, or perhaps from underground economic aciviy along he Mexican border. Alernaively, he change migh represen a shif of some aciviy o Los Angeles from oher cash offices: he relaive growh of noe receips a he New York Ciy cash office slowed during his period, for example. Regardless of he reason, we emphasize ha he esimaes presened in his analysis are based on he emissions and receips of large denominaion Federal Reserve noes a all 37 Federal Reserve cash offices, no jus wo, and hence are likely much more robus o such evens. The sock of foreign-held US currency suggesed by he shipmens-proxy mehod, expressed as a share of US currency in circulaion ouside banks, is compared o oher esimaes in Figure 13. Due o he benchmark assumpion ha half of US currency was held abroad as of December 1973, he shipmens proxy mehod, like Porer and Judson s median-flow mehod, is ap- 33 Feige s mehod is discussed more fully in an appendix available on reques from he auhors. 34 Bach (1997), Table 3, page 49. No explanaion is offered for how he benchmark foreign-held percenage was obained. 18

21 proximaely equivalen o assuming ha half of US currency has been held abroad each year since The wo mehods share anoher characerisic: excep for a small increase circa 1980, he esimaed shares boh decrease during mos of he 1980s, even while large denominaion noes are growing four-fold faser han small denominaions (hereby doubling heir share of currency in circulaion), and surveys sugges ha US households are no increasing heir holdings of large denominaion noes. 4. The Domesic Adjused Moneary Base and Moneary Policy In his secion, we discuss how our pariion of domesically and foreign-held currency may be used o build a domesic adjused moneary base daa series. We use he adjused moneary base as published by he Federal Reserve Bank of S. Louis, because all he required daa are readily available. 35 This measure of he adjused moneary base has been recenly revised, and is now consruced as a chain index; see Anderson and Rasche (1996 a, b). During each of several separae inervals, he adjused moneary base is consruced by adding a reserve adjusmen magniude, or RAM, o he moneary source base. Differen, specific RAM series are used in each inerval. The final adjused moneary base is formed by chaining he separae pieces ogeher o form a single coninuous ime series. 36 Our separaion of currency eners via he moneary source base, which is defined as he sum of currency in circulaion ouside he Federal Reserve and he Treasury, plus he deposis of domesic deposiory insiuions a Federal Reserve Banks. 37 The domesic moneary source base equals he moneary source base minus he esimaed amoun of foreign-held currency. We have calculaed his measure of he domesic moneary source base monhly from January 1950 hrough Augus The overall domesic adjused moneary base is consruced by adding, during each ime inerval, he domesic moneary source base plus he appropriae RAM adjus- 35 So far as we could discover, he daa necessary o reproduce he Board of Governors adjusmen for changes in reserve requiremens has never published. 36 The S. Louis adjused moneary base is chained, or spliced, in 1972, 1975, and 1980, and seasonally adjused by X11-ARIMA. 37 Noe ha he Federal Reserve Bank deposis included in he moneary source base already reflec a separaion of domesic and foreign componens. Deposis due o deposiory insiuions doing business in he U.S. (including branches and agencies of foreign banks) are included, bu deposis due o foreign cenral banks are excluded. (No oher offshore deposiory insiuions excep cenral banks hold deposis a he Federal Reserve.) 38 Before 1965, he domesic and oal moneary source base are he same; we choose 1950 because ha is he firs dae for which he S. Louis Fed publishes a seasonally adjused, adjused moneary base. Daa on he oal moneary source base are available from he S. Louis Federal Reserve Bank beginning in

22 men; he final series is chained ogeher in he same manner as he oal S. Louis adjused moneary base. The oal and domesic adjused moneary base measures are shown in Figure 14. The difference beween hem increases seadily afer 1965, for wo reasons: an increasing share of currency is being held abroad, and currency is becoming a larger share of he source base. Yearover-year growh raes of he oal and domesic adjused base are shown in Figure 15. Currency expors have frequenly accouned for hree or more percenage poins of growh in he adjused moneary base during he sample period. In recen years, our esimaes sugges very slow currency ouflows and small differences in heir growh raes. 5. Domesic moneary aggregaes, he k raio, and velociy Beyond he adjused moneary base, ouflows of U.S. currency o foreign circulaion have been large enough o disor inferences regarding he sance of moneary policy based on broader moneary aggregaes. Year-over-year growh raes of oal (published) and domesic M1 and M2 are shown in Figures 16 and Growh raes of he published daa for boh aggregaes are significanly larger han hose of heir domesic counerpars during he mid-1970s and he lae 1980s, hrough For boh M1 and M2, he exen of he secular deceleraion of money growh since he mid-1980s has been parially masked, in published daa, by he acceleraion of currency expors, especially beween 1987 and In conras o shipmens proxy-based mehods (no shown on hese figures), our esimaes sugges ha currency expors have slowed significanly since 1995, and ha increased currency growh during recen years reflecs srong domesic economic aciviy. Finally, we noe ha, hisorically, differences beween he growh raes of published and domesic aggregaes have ended o increase when money growh was slowing, a relaionship ha may prove imporan in empirical models of he effecs of moneary policy. Expors of currency also have disored componens of money mulipliers. The k raio is equal o he raio of he currency componen of M1 divided by ransacion deposis. 40 Four measures of he k raio are shown in Figure 18. The upper wo lines measure k wih published 39 The M1 daa shown in he figures have been adjused by adding he amoun of ransacions deposis ha he Federal Reserve Board esimaes is being swep by banks from checking deposis ino savings deposis. Noe ha hese reail sweep programs began only in January 1994, and do no include sweeps of checking deposis ino repurchase agreemens or money marke muual funds. For discussion, see Anderson (1995) and Benne and Hilon (1997). The daa are from he S. Louis Federal Reserve Bank a hp:// 40 For discussion of money mulipliers and componens, see Rasche and Johannes (1986). 20

23 oal currency in circulaion, and wih and wihou, respecively, adding o published ransacion deposis he Federal Reserve Board s esimae of he amoun of reail sweep programs. 41 The lower wo lines measure k using domesic, raher han oal, currency, and also, respecively, wih and wihou including he amoun of reail sweep programs. The boom line, based on domesic currency holdings and including an adjusmen for he amoun of ransacion deposis swep ino savings deposis beginning January 1994, displays remarkable sabiliy around a generally downward sloping rend hroughou This behavior of he relaive quaniies of currency and ransacion deposis lends furher suppor o our esimaes of foreign-held U.S. currency. Because he k-raio remains close o is mean during he enire 50 year period, repors of he imminen deah of currency as a domesic ransacions medium perhaps are oversaed. Finally, hree measures of he velociy of he adjused moneary base are shown in Figure 18. The measures published by he Board of Governors and he S. Louis Federal Reserve Bank have very similar behavior. The velociy of boh measures increases seadily unil 1980, afer which i decreases. This behavior clearly reflecs he pah of shor-erm ineres raes and inflaion, and hence he opporuniy cos of holding base money. The velociy of he domesic base increases more rapidly han he velociy of he oher wo measures before Afer 1980, however, i becomes saionary, wih no discernible rend. This paern migh sugges a higher ineres elasiciy for domesically held base money, compared o he foreign-held componen. A a minimum, i suggess ha empirical sudies ha use he adjused moneary base mus be cauious regarding a break in he rend of velociy circa Conclusions U.S. currency held abroad is an imporan asse for residens of many counries, and a subsanial source of revenue o he U.S. Treasury. Expors of U.S. currency during he las several decades have significanly disored growh raes of he adjused moneary base and of boh narrow and broad moneary aggregaes, including M1 and M2. The increase in foreign-held currency also has disored componens of he money muliplier, such as he k raio of currency o checkable deposis, and perhaps lie behind some conclusions regarding he assered insabiliy of U.S. money demand relaionships. Recen revisions o boh he Flow of Funds and he Naional Income and Produc Accouns recognize hese, and oher, aspecs of he imporance of measuring he amoun of U.S. currency held abroad. Furher, a pariioning of he moneary base beween 41 See foonoe 37. Transacions deposis prior o 1959 are he demand deposi componen of he hisorical M1 series in Rasche (1987). 21

24 is domesically and foreign-held componens seems necessary if furher progress is o be made in empirical models o link he behavior of he economy s price level and he balance shee of is cenral bank, he Federal Reserve. The mehod of esimaion of foreign-held currency proposed in his analysis provides he firs consisen monhly measure of he domesic moneary base and foreign-held U.S. currency. Unlike previous analyses, he ime series is benchmarked o daes sufficienly early in ime (December 1964 and December 1968) ha relaively lile U.S. currency was held abroad, making our esimaes robus o he seleced benchmark. Our esimaes also permi consrucion of domesic moneary aggregaes, including domesic M1 and M2. The esimaes may be readily updaed (by Federal Reserve saff) a he close of each monh from Federal Reserve currency processing daa. Furher, because he esimaes uilize flows of large denominaion Federal Reserve noes from all 37 Federal Reserve cash offices, raher han jus New York Ciy and Los Angeles, hey should be less sensiive han previous series o shifs in currency processing paerns among cash offices. 22

25 References Anderson, Richard G. (1995), Sweeps Disor M1 Growh, Moneary Trends (Federal Reserve Bank of S. Louis), November. and Rober H. Rasche (1996b), Measuring he Adjused Moneary Base in an Era of Financial Change, Federal Reserve Bank of S. Louis Review, November/December (1996a). A Revised Measure of he S. Louis Adjused Moneary Base, Federal Reserve Bank of S. Louis Review, March/April Andris, Felix (1997), Das Geldnachfrageverhalen der Haushale in der Schweiz (Bern: Verlag Paul Haup). Bach, Chrisopher L. (1997), U.S. Inernaional Transacions, Revised Esimaes for Survey of Curren Business, July, pp Benne, Paul and Spence Hilon (1997), Falling Reserve Balances and he Federal Funds Rae, Curren Issues in Economics and Finance (newsleer), Federal Reserve Bank of New York, April. Board of Governors of he Federal Reserve Sysem (1996), Flow of Funds Accouns of he Unied Saes, (Z.1 release, quarerly), December. Brunner, Karl and Allan H. Melzer (1976), An Aggregaive Theory for a Closed Economy, in Jerome Sein, ed., Monearism (Norh-Holland) Chari, V.V., Lawrence J. Chrisiano, and Marin Eichenbaum (1995), Inside Money, Ouside Money, and Shor Term Ineres Raes, Journal of Money, Credi and Banking, Par 2, November. Chrisiano, Lawrence J., Marin Eichenbaum, and Charles L. Evans (1997), Modeling Money, Federal Reserve Bank of Chicago Working Paper 97-17, December. The Circulaion of he Deusche Mark Abroad, Deusche Bundesbank Monhly Repor, May Coleman, Wilbur John, Chrisian Gilles and Pamela A. Labadie (1996), A Model of he Federal Funds Marke, Economic Theory (7). Fama, Eugene F. (1983), Financial Inermediaion and Price Level Conrol, Journal of Moneary Economics (12). Feige, Edgar L. (1996), Overseas Holdings of U.S. Currency and he Underground Economy, Exploring he Underground Economy, Susan Pozo, ed. (Kalamazoo, MI: W.E. Upjohn Insiue for Employmen Research) (1994), The Underground Economy and he Currency Enigma, Public Finance and Irregular Aciviies, Werner W. Pommerehne, ed. (Supplemen o Public Finance, vol. 49), pp Gurley, John G. and Edward S. Shaw (1960), Money in a Theory of Finance (Washingon, D.C.: Brookings Insiuion) 23

26 King, Rober G. and Charles I. Plosser (1984), Money, Credi, and Prices in a Real Business Cycle, American Economic Review, June McCallum, Benne T. (1988), Robusness Properies of a Rule for Moneary Policy, Carnegie- Rocheser Conference Series on Public Policy, 29. (1990), Could a Moneary Base Rule Have Prevened he Grea Depression? Journal of Moneary Economics (26). (1993), Specificaion and Analysis of a Moneary Policy Rule for Japan, Bank of Japan Moneary and Economic Sudies, November. (1995), Moneary Policy Rules and Financial Sabiliy, in K. Sawamoo, Z. Nakajima, and H. Taguchi, eds., Financial Sabiliy in a Changing Environmen (S. Marin s Press). (1997), Issues in he Design of Moneary Policy Rules, paper presened a he conference Recen Developmens in Macroeconomics, Sanford Universiy, March and Monica Hargraves (1996), A Moneary Impulse Measure for Medium-Term Policy Analysis, Inernaional Moneary Fund, papers for he World Economic Oulook, December 1996 (also, IMF working paper ). Painkin, Don (1961), Financial Inermediaries and he Logical Srucure of Moneary Theory, American Economic Review, November. Porer, Richard D. and Ruh A. Judson (1996), The Locaion of U.S. Currency: How Much is Abroad? Federal Reserve Bullein, Ocober, pp Rasche, Rober H. (1987), M1 Velociy and Money Demand Funcions: Do Sable Relaionships Exis?" Carnegie-Rocheser Conference Series on Public Policy, 27, Auumn. and James M. Johannes (1987), Conrolling he Growh of Moneary Aggregaes (Kluwer Academic Publishers) Scholl, Russell B. (1997). The Inernaional Invesmen Posiion of he Unied Saes in Survey of Curren Business, July, pp

27 $ Figure 1: Raio of Emissions o Receips a Federal Reserve Cash Offices monhly, January June 1999 All Cash Offices New York Ciy office Noe denominaion $5 $ $

28 Figure 2: Raio of Emissions o Receips, $10 noes monhly, January June 1999 (varying verical scales) All Cash Offices Excep New York Ciy New York Ciy cash office New York Ciy raio divided by non-new York Raio

29 $ Figure 3: Raio of Emissions o Receips a Federal Reserve Cash Offices monhly, January June 1999 (varying verical scales) All Cash Offices Excep New York Ciy New York Ciy cash office Noe denominaion $ $

30 Figure 4: Esimaed Domesic Share of $100 noes Shipped based on circulaion paern of $10 noes 1.50 Unresriced esimae Esimae capped a

31 Figure 5: Esimaed Ne Emissions of $100 noes Based on New York Ciy cash office daa for $10 noes billions of dollars per monh o domesic circulaion o foreign circulaion

32 30

33 Figure 7: Esimaed Domesic Share of $50 noes Shipped based on New York Ciy emissions-o-receips raio of $10 noes 2.00 Unresriced esimae Resriced esimae capped a

34 Figure 8: Esimaed Ne Emissions of $50 Noes based on New York Ciy cash office daa for $10 noes 1.6 o domesic circulaion billions of dollars per monh o foreign circulaion

35 33