View original document

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

DALLAS
FEDERAL RESERVE BANK OF OALLAS
1991
March 1991

1

••

•

CVICW
~conOInlC
cononnc eVlew

Forecasting the
Louisiana EcolZOJny
Economy
William C
C.. Gruben and
Donald W. Hayes

Modeling the Effects
ofInjlation
of
Inflation on the
Demandfor
Demand
for Money
Kenneth M. Emery

This publication was digitized and made available by the Federal Reserve Bank of Dallas' Historical Library (FedHistory@dal.frb.org)

Economic Review

_
..
.........---_....

Federal Reserve Bank
ofOal/as
Robert D. MeT"" Jr.

HIM\' RoMnblllm

s.- ....... ,...,.., INI

G,rlilli P. O' DriKoli. Jr.
~ t:low.- #t "'-'<II

-..........

W. MkhMlCo.

"'_

SIIplten P. A. Brown
,w,,1MII_~1'Jd

Economil1S
Nallonal af1d International
John K Hill

Evan F Koenig
Robel'1 T Clair
Joseph H Haslag
Kenneth M Emery

David M Gould
Mark A Wynne
Kevin J Yeats

RegIOflaI and Ensrgy
William C Gnb!n

Robert W Gilmer
Mine K Yilcel
Keith R. Ptuilips

LOri L Tay10f
Fiona 0 Sigalia
Editors
Rhonda Hams
Diana W Palmer
Vi/lI,oia M Rogers

The Economic Review II ll,IbIisiled bv ,tie
f edefll fIe1.,.... Baonk of o.lIiIS The \'Iew1 e.,
preued ale !hose of IIle aUltlln lind iXI not
rte(e$wwdv reHICI tile pOI!IIttlS of me Feder"
Re_ BIr* of OaDn 01 me Fedeli! ResertIII

"'~
s",b$CJ,ptJCIM Me IMIIabie ifw 01 thlrllll
PlUS! send ' !quesll lor "ngJ"f(lpy .nd
nUl~ SUCISeI~.

back WIllIS. ird
d'I¥lQes 10 lilt !'ubi.:; AIIIIfI DIc*t.
men!. fWeflllieserve 8irt of o.U•. SIJI>O'I K.
o.tIas. Teus 15Z22.121' J651-628S
At\Icllls ""Y bt reonmed (J:I1N anltt ...
l1li1 !hi IIUt$ IS credcrtd tnd the Aeswt:h
()epIrtmenl IS prOVIded WIth I COPY 01 till
pIbIic.I~on COI'IUIIftIn\I the IffIMted ....1eN1
~ohss

Contents
Page 1

Forecasting the
Louisiana Economy

William C. Gruben and Don:l ld W. Hayes have constructed a forccasling model that predicts mild overall growth
in the louisiana economy in 1991. The model predicts

William C. Gruben and
Donald W. Hayes

callnt. the civilian Jabor fo rce, nonllgricultllral employment,
hOllsing permits, real personal income, mi n ing, ,md sa les ta x
revenue. Only durable and nondurable goods manufactming
will decline, according to the forecast. Gruben and Hayes
conclude that while I.ouisiana is likely \ 0 experience an
economic expansion in 1991. a boom is unlikely.

expansion in seven of nine economic indicators: the rig

Page 17

Modeling the !ffects
ofInflation on the
Demandfor Money
Kenneth M. Emery

Because the Federal Reserve is responsible for controlling the value of money. economic policymakers are very
concerned wilh forecasting the public's demand for money.
Inflation is one of seveml faclors thai have made forecasti ng
the demand for money increasingly difficult over the past
fifteen years. In this article, Kenneth M. Emery reviews the
ways that inflation may affect the demand for money, and he
examines how well tr-.Jditionai money-demand models have
captured these effects in the postwar period.
Emery finds Ihat money-demand models would have
performt:d ocner during 1953-79 had they included Ihe direct
effects of inflation However, he also finds some cvidenc(! Ihat
the wide use of interest-bearing money and moderate rates of
inflation during the 1980s diminished the effects of inflation
on the demand for money.

William C. Gruben

Donald W. Hayes

SeniOl' Economist and Policy Advisor
Federal Reserve Bank of Dallas

Economic Analyst
Federal Reserve Bank of Dallas

Forecasting the Louisiana Economy
ur fort.·cbts n~ .n im.· t.·("ono l.nic indica to rs
a I)QS II1 VC t.·conOllllC o Ul look for
J.ouisia n :1 In 1991. Our fOI"l.:casls p rcdid g rowl h in
~",,\'t.'11 oj Iht.' n in .... indicators fo r I.ouisiana : tht.· rig
counl. Iht.' civ ilian lahor fOl"l·..... nonagricultu r.ll
t.·lllpl()YrlH:.·nl. hOLlsin~ jx:rmils. rl';11 JX'rsonal
iIKOlllL·. mining c mploymL·llt. ;Ind rcal salt.·s lax
lv\"t.·nUl· Ernploymt.·nt should (IL-di ne in hOlh
dlll~lhk ;lnd nondurahlt.· W)(xls ma nutaCiuring Wt.·
<.kriv....d our pn::diclioll., from an ori!!inal tim t'sL'ril.:.' modd. con"'lnlCl L'd "'1x:dlk ally to :.m;.!lyzl.:
Iht.' ].(luisian;1 economy u ... in~: 1 pnx:~durc WL' will
t.·... plain in thi:- :Inidt! Dt.·... p il .... o ur predit"lt:d
t.',,, p:ll1... ion in 1991. Loui"'I;1I1:1 i, unlikdy to Teach
ih t.'l1lphl} mt.·!lI peak-. (If 11ll' L':lrly 1980...
Firsl \\L' di-"'.:u....., l u.llsian;I·.... , tnK."lur.I1 and bboT
for<.·t.· ch; I1~lctL'rislic:-. 111....n " .... (k~rihe our fort.·c.I ...Im~ l1lodd of l..oui,ian:l·s l'conomic indiC,Jlors and
cxpbin how W~ \I~ IhL' Illtxk-llo fOR--C lst tbe statL'·...
t.·(:ooom": !,x:rfonnancc for 1991 \V~ t;()nc\udt: hy prt.....,.:ntin,g thL' forec, ...( and <.It ...nl.s... in,g it!'> implications.

O

.'ll~~i:'st

Characteristics of the Louisiana economy:
an Qven'iew
J.,llli ...ian:l·s L'("()f)(~niL Stl1lctlll'L' and pcrfonnanc....
arlO alll()n~ Ihe most LUHI .~lI; tI in tIH.: United Statt:s
lX:call'L' of thL' extl't:'llldy illlp0l't:lnt role that eneI'. i
L'.xtl~ll" li ()n play ... in 11lL' ...I:IIL'· ... L'COnolllY It is no
coinL"idt.·nct: Ihal hoth r~;11 oil prict.·... and LoLLi.siana·s
....mploymt.·nl 1'X':lk .... d in 19l-! I :\'01 only does ~nL·rgy
I.:xtraclion pl.l}' an imponanl din:t.1 rol~ in (h~
I.ouisiana ~("(momr. h UI il .Ibo strongly affects tb~
...lnl(·III1\: of Ihl.: stale', 0111t.·1' t.·(:onomic ~-'{,"lors. slldl
:I'" m;nlllfa<.1unnA
Ent.·rg} -rdated plwn( Illlena arc not thc only
UllliMla! :I!'>lx-"t."lS o f Iht: Loui,i:tn:1 L·cono my. E\'~n in
hoom IX·ntxi!'>. Ihe I.oui... i:tn:1 blxlr force duunic.llly
... \lI1"l'rs hum higher unelllplorlllt.·n! r.1lt.'S Ih:1I1 tht.·

n;ltion :11 aver.. ~..... I'-I0 rL'0\'l:r, I..oui.,iana's lahor
pal1iL"ip:ltion r:1IL'S :H\: Mlm.... whal 10wI.:r rh:11l the
n:ltion's AS:I con-"L'(jlICnCL', Louisiana's employl11l'nt r:ltL' (thL' 1~l\ i() of c illploymcnl 10 the civili:tn
l1()l1in~tiILilional bl)()f f()rL"~ ) rem:lins chronically
Ix:low the nation·s. WL' addl'L'ss Ihe implications of
I hi.s p hL'nolllt.·non in:1 !at!.:!' section.

Composition of the Louisi.ana economy
BcC.Jll'>t.· o f Louisiana ':- p(.-'Culi:tr L'conom ic
...lruCl urc. it is not unusu;11 for the !'>t:lt,,·s economy
10 Wow \\'h~n mn,1 ol her sl ales' economic... art:
dcdinin~ and to shrink \vhen OI:hl..·r stale ... · c("onomil's :lTC ~xIXlnd in~ Among thc OflY SI:I1t:S and
thL' Dislril..1 of Collim ili;l. only eight cronomit:...
\';uy le~, wit h n:llional Ollctu;Hions (Sht.'rw()(xl-Call
19Xfn. and o nly four aTe I~ ....... divcrsified
(Shcrwood-DII 1t)1.)())
The oil :lIld ~a ., t.'xlr.lction industry dominall.:s Loubia n;l·... !.:conomy :-0 p rofoundly that it
:It.·COlHll s for :ll1no.'t len ti llleS as large a ... han.: of
,gross Sl alL' p r(xlllt"1 (GSP) as in lilt.' a\'~r.I~t.: state.
Energy accOllnl~ tor ;1 ~rcaler ponion of GSI' in
Lou isi an:1 than in any olher state except Wyoming
(Schmidt 1989) :111(\ for more of l.ouisial):t output
than <I ny ot hl.'l' l11ajor L'<"onolllic secl or. A .~ of 19H6,
tht: most 1'L'(.'(;:111 yt.·:11' for w h ich d ata exist , oil and
,g:I .~ C.xIr.lClion ac(;()unlcd for one-sixth of
L()lIisiana'~ CSt' hUI for onl y one-sixtieth of the
n:lt ion ·.s gros... nal ion;11 product (GNP ) 1\·loreo\·cr.
1900 W:lS a particuhlrly WL':l k yea r for tht: L'nl'r,gy
indl1slrr: in 19H2 il produced more than 0111.:founh of l oubia na·... GSP

v,c wovId Me 10 InaoI< St_ Brown. l or' Tay10t and Oavrd
lor lne. he/fJM commentS

G~

Figure 1

Compos ition of the Louis iana Economy, 1989
(Percent)

(11

SOURCE: U S. Bureau 04 Labor Statistics

Figure 2

Composition of the U.S. Economy, 1989

"'"
SOURCE: u .S. Bureau oIlabof Statistics

l

~)

Energy extraction accounts for a much larger
portion o f louisiana's GSP than of employment.
Figure 1 :lOd Figure 2 depi<-, employmt:nt shares by
industl)' for Louisiana and the UnilL'<.i States in 1989.
iI'lining employment. of which 96.9 percent is in
oil and gas extraction, occupit:s I(.!ss of the swte's
work force than almost :.ny otht:r major sector.
Nevertheless, the energy industry still has a
~trong influence 011 employment in Louisi:lOa.
Brown and lIiIl ( 1988) show th;1I changes in
energy prices affected O\'(ToIIJ L"Inployment in
Lou isiana to a grealer degn.'c than in all hut three
U.S. states (Wyoming, O klahoma. and Delaware).
In lou isiana , a S; increase in tht: price of oil
yidds a 1.6-pcrcent increase in th~ swte's nonagricultural employment over tht: long m n
(Rosenblu m and others 1990).
Moreover, while mining ahsorbs only a small
portion of Lallisial13. 's work force , the Slate's
industri:11 composilion of employment is far more
heavily weighted toward mining than the nation 's
is. As Figures I and 2 show, mining accounts for
aoout five times as high a share of I.ouisiana 's
t:mployment as the nation's.
Government also employs a disproJXlrtio nateJy brge st:gment of Louisiana's work force
Almost six times as many Lo uisianans work in
government :15 in mining, and government's share
of t:mploymcnt in Louisiana is aoout one-fourth
hi~her than Ihe national average.
State and local government accounu, for this
disproportionate share. The fedeml government
t:mploys a relatively small JXlrtion of workers in
Louisiana, but 1000..ll government's employment
share is one-fifth higher than th~ national ~l.\'ef'"J.ge,
and slate govcmment"s share is higher by one-half.
Despite the relative impon:lOce of government as an employer in Louisian;t, the sector
accounts for only 10. 1 pcrct:nt of Louisiana's esp,
compared with 11.7 percent nationally. Even
though government employs r;l r more Louisianans
than mining does, mining accounl~ for :l substantially hlrgcr sh:lre of Louisiana 's GSP.
Manur:lcturing also employs more louisiana
workers than mining does and aC('OunLS for more
esp than government does. Comp:lred with the
muion, however. Louisiana's mllOllfacturing sector is
relatively small in tenns of jobs and output.
Another aspect of the role energy plays in
the louisiana economy may be seen from an
Federal Rcscrve Bank o f DaIla$

l:'xamination of th~ st:lte'<; m:lnuf:tl:turing .~t.'l'tor.
(k·cau.,,>~ the state h:ls :1 \\ ...:I k dUfahk' g<xKI...
"'l·cto r. manufacturing's s h:lI'l' of Louisiana·... GSP
<U pcrl:ent) is a ftood d ...:11 ... malk-r Ihan it...... han:
of the n.. lion·s GNP 09.6 per<.'Cnt). Ho\\'",\·~r.
IX·'<:;.Il1se lilt: t:nergy-linkt--d ch ... lllkal :mtl refining
indllstric~ :m:: so imponam in Louhi .. n:l. the nondur:lble go()d~ M!Ctor ;Kl'ounts for a We:ltl'T share
of L()u i ~ i :m:I's GSP than of the nation's GNP. I
Together. the refininj.( and chcmiG.11 industrie...
:Kcount for 6 3 percent of Louisiana·... GSP. <..'O1l1paR-xl wilh a 25-perct·nt .~hare n:uionally. ,., Ioreovcr,
the value of o utput in theSt: t\"\'o industries aCt'mmt<;
for 68.9 peret:nt of th ... value of total LOlli~i:ma
nondur:lble goods production. Like mining. tht'se
industries are capital intensive. so they :Lecount
for :1 lIluch sma ller shan:! of ... tate t:mploymt:nt
(27 percent) (han of GSP (6,3 pcret:nt).
Finally. hecau.'>I: of ent:rf!Y's imporlant rolt:.
g<xxls-producing indu... tries occupy a far brw.:r
share of Louisiana's GSP <36.3 pcrct:l,n Ih:m of
thl" n:llion's GNP (18 9 pl'rcl'nt) ~ The domin:lnee
of mining more than offsets the rd:lli\'dy sm:11I
contrihutio ns of l1l:lnllf:.cturing and ClIIl...lruelion.
thl" other goods-producing sectors

Thl" !lab nee of the differences between
Louisi;ma and U.s. employment rates occurs
IX'c'ltI."C a di.'iproportionatdy l:irge number of
johle... ~ working·age Loll isianans arc nO! even
~el'king work. which means they are nOt pan of
the blx>r forel" Thai is. Louisiana nOI only has a
chronically hifth unemployment r,lIe hut also a
chronic:.lliy low lallOr panicipation rate. Partly as a
rc.">ult or low employment and labor panicipation
r:lle."> :.lnd chronically high unemployment rates.
LOlli... ian:l r:lnked fony-sixth among the fifty states
in personal income pcr capita during 1989.

ConstructJon of the model: an overview
The sol~ plI T])().st: of our time-series model is
to forcC:l.~t nine Louisiana economic indic:nors as
:ICCUr.llely as po~sible In its final form. fhe model
indudt·s ninl' foreC:lsling equ:nions, one for each
rorCC:I.~t variahle. Thl" model is not designed to
te~[ hypotht'se~ :Ihout the stnlcture of louisiana'S
economy ;lIld was not constructed in accordance
wilh :l ny thl'"'()retieal economic modeP
-111c model-huilding approoch we use is a
purely statiSlie.tl process. involVing :l two-stage

Louisiana's labo .. force
Louisiana's labor force <:har.Jclen~tiL'" are also
unusual. In louisiana. empIO}'111l"nt as a percentage
of the cj"ili:m noninSiitulioll:.l fXlpuiation <or Ihe
employment mte) h:ls f\.."tnaincd helow the national
:IVCr.lgC even when the sta (~·... employment wa ...
growing more rapidly th:m the n:llion's Thi... means
th:.t a dispropoltionatcly brge numher of workingage Louisi;mans dtllcr lire out uf work but :tt1ivdy
seeking it or arc not even part of thc labor force. In
1980 and 1981. for example. total employment in
LouLsi:ma grew at more tha n fin~ times the n:.llion:!1
rat~. but l.ollisian:I·S employment r.lle remained
lx:low the nation:!1 aver:lgc in both years.
Most of the difference.. hetwl.'Cn the Louis;;!n:!
:md U.S employment rates Gm Ix: explained hy the
statc's unemploYlllent roue, whil'll is usu:llly well
al)()\'c the nation:11 :!"erage. Unemployment rates
for hlack males. hbck females. white: malt-:-, white
femalt'S. and for young persons of b()(h sc:xes are
llnifonnlr higher in i..ouisiana th:!n in the nation
nlC le&:ll definition of unemployment in\'olvl"S only
jobless workers who arc :loivcty St:eking work
Eco n Ofld c Rc" icw -

March 199 1

, 18CI1ntCafly. /llerefiNng data Sle8Clual1y b" pelro/eum and
coal prOduCts. but ref.rung dotrwIatcS this motu aggregate
category
> BeSIdes mmmg and fIOfldurab16 goods manufacturing. the

onlymator mdustrycatcgory mat has a /arf}(If-tflan.nat.onal.
average shllre 01 GSP m LOUIsiana is transportallon The
(ast. mcludlng af}"culrur6. durable goods manufacturmg.
who/esal6 trade. relail trade. {mance. services, and government. have shares below the naliona/average
, Whlla tha LOUISIana forecastIng 8qVal1QflS were not con·
stIVCledlf'laccordaflCCWlthany IhooreDCalCCOflOfflICmodcl.
thIS statement sholJla nor be taken 10 mean that we ignored
all "'formatlOrl aOOUI LOUISIana If! all aspects 0I/he model.
bulldof'f} procedvre We rehed on our MowIedge 01 /tie
LOUISlaml economy LmcrI we selected me OVCfall set 01
polenDal fOfecastng vanables. from wh1ch /he candidate
vanables were SBfecI8d IlIStlS why we inf;1udfJd 011 pnees
8S a vBnBO/(J We did fIOf. however. fotcc /he oil pnce
vortable 10 be 8 candidate I131l8b1e Of 10 appeal'" any frlal
Iorccas/lflO cqu8110tl me Sl8tISIICSldecJ$lOll rr.Jles were the
baSIS lot all SIJCf'IctIOICes. and they were COfISISlen/ WIth Ihe
.o1C>\I$oOn 01 ff'lBOII pnce V3fl8b1c til 5(1V(Ifsi final forecaSlmg

-''''''

,

system of dt..'(:i.~ion niles.' We llSC tht.:.-.c mk.·s to
choose which of ninetlXn national indic:.llor... and
nine l.uubiana indic.Ltor:-. to lbC ;L;o. fon.·(:;Jstin~
variahle:. in e;lch of t.he nine final L'tillations. hut we
conslmct many prelimin;lry L'(lllation~ before we
I'C'Jch the final nine.
In the following three .~L....·ti()n.~ . we diM:lL.......
detaib of the construction of the mudd. First. we
describe the two-stagt;.· pl1X,.·t.....lure u . . L....1 to seleci
fOreGlsting \'ariables for each I.: quation 1'ht:
modd-building procedure for loui.. . iana n:semhles
those presented in pasl anidcs on regional
forecastin!i in the Ecol/ omic /{el'iell i. hllt we havt'
made a signilk.mt change to accolllL11od:lle the
oil-related events of 1990, :u,d we disnlss this
change in Ihe sectiO!, laheled 'The u .~e of consensus forecasts in the modd .. In brief. we lise
external, (:unsensus-b:l.~ed foreC:lsts of nation:11
indicators instead of Ihe cndoRenous. moddbased forec;lsts that h;Lve been depit-Ied in past
anidt..'s The configurJtion of the final model.
di.scllSSt..'d in a later .-.cction. refle{.ls in many ways
the :.tTl.lcture of the l.ouisi;ma Ix-onomy. E\'en
though the model I:' con:.tnl('tcd on time-serit:S
principles, energy-related indicators ofll.·n ;Ippt:ar
in the equations that forL'Glst tlK' aAArq.pte
variable:.. and sc,'cr..11 Dlher c har..Ll·teri... tic:. of tht:
model arc con~b(ent with what one might expt:ct
from a slructur..11 model.

Tests of variables for the forecasting model
The COl1s(rm:tion of tht..' Louisiana (ime-series
model involves a two -~t;Lgt: procedure for selecting forecasting variables for each of the nine final
equal ions. It, both stages, the Ill(xld-huildl:r
constructs a simple foreCaSlil\l.\ regressiol' equ;l -

• I hiS d iSCUSSion of inC modeImg procedure IS very general
Ny ITI(ItC derailed descflOiIOflS oIlhlS mode/lOg procedure.
S8t! Grubon and Long (19880) and f/oehf1 and Balazsy

( /985)
J

To plCSCfWJ

IlI'1I8-ser~ SllIllOnBflry. 11/11 dalll are placed"

frrsl dlfferCflCCs 01 /ogIIrllhnllc Itxm fills means /hal all
vCtflaOles are clfptesscd

as

growrh fares >NhIch are /Jle

same 85 firSI dlffflfflflCes oI1og8l11hms. and nalas absolute
/eveI$ ~ WIt COtIS/rUC/ I~ress.on equ<ll/Clns. we are
IDglCSSNlg
QfOWIh lale oIa V8IIaOJe on lIS own paSI
grOWlh f/lles II/IlQI\::Ir on past growrtl rales 01 011>« VCtf.ables

m.

tion. adds a variable to it, and tests to sec if the
new variahle ;Idds to forL'Ca"'Ting aCCllr.Lcy .~ The
entire proces.... of Illodd constnlltion i.. JUSt a
series of variant:.... on this procedure. The stages of
Ilux.ld construction rt..'M;~lllhle those taken to dect
an individual to puhlic office under thc pany
system in the L'nileu State:. .
In thi.~ tWo-sl:tgc :....y.. . tcm . the purJXlse of the
first stage is 10 :-ell.....·! fOfCl·a:.ting v:lriables that will
he c;mdid:Lft..'."; for indu.. . ion in ;11 lea~t one i..."'qualion in the lLltiLllatt: forecLsTing modcl. In continuing thc :1Il;Llogy to :L politic.L1 campaign. this stage
Illay be comp;lrcd to :1 party primary. We test
every L()ui.~i;m:t variable's ability to forecast itself
and il .~ :Ibilil)' 10 foreGI .. t other Louisian;l variables
We ,llso Il.,':-.t T1K' ability of each of (he nineteen
nation:ll indkators to forecasT lluctu:Hions in the
l,()uisian;1 indicators.
JU.... I as the viClOr of;1 primary election may
not win Ihe gener.L1 election. (hI.,,' Gmdidatl.,,'S
:-elcctl.,,'d in this first st;lge may not ultimately find
thl'ir way into :lny t'quation in thl.,,' romplcted
forL....-:lsting model E\'en so. only vari:Lhles thaI
have bt.'Comc Glndidate:-; through the firsl SI:lge of
[csts will recd,·c cun.. . ider..ltion for an L,<]lIation in
thl' SI..·(:ond :-.tage of m<xlding. Note. however, that
:I fort'GI:-.ting \';Iriablc Ih:lt ht.....'Olllcs a Glndidate
for one final t.,<]u:ltion ocL'<.1 not be ;t caooid:tte for
other:..
To Ix'gin thc fiN! staRt: of model building.
we ex;uninl.· tht' ;Ihility of past fluctuations in a
variabk to fon:Glst its own fUllLre 1ll0\'ements For
each variahle. wc tl·.~t whether an equal ion
containing ;. conSI;lnt tcrm-plus one- and twoqU;1I1er bgs of tht..' dcpcndent variable-forecasts
more accur.II\:ly Ihan ;1 :-.imple random walk
A .... t..'cond .~tep in thi.~ first stage of the model
is to add other v;lriahle ....--ont.: ;1( a (imi...'-to the
L'qu:Ltion that conTains onc- and tW(K]Lmrter lags of
tht.· dept:ndem v;lri;lhle. In this stage, l.,,';lCh of the
independent v;lri;lhle."; is te:-.TI.'{1 ;Igainst an equ:ltion
that con!:lins only two b!{.... of the dependcnt
vari;lble. -n):lt i .... we do not test several vari:tbles at
the S:LIlle lime in Ihe same l.'{lu;uion.
For t..'x:lmpll·. 10 te:....T Louisiana housing
pcmlil'" a:. a predictor of louisiaml's drilling rig
count. we add two l:Lg.~ of louisi;lOa housing
pennits. We tlll'n tc:....t the il1formlltioll 8(1il1. or
increase in fO!'L'(.·a:-.ting accur.KY, [hat the inclusion
of thi ... variable Ill.. y h;I\'c over tht' modd Ihat
F~dera l

ReK...·~ Hank of Dallas

conta i n.~ only t\VO b~s of the d epcn(it.:nt va riabk.h
T:lhk' 1 pn: ...c nts information g:lin st:nistics. and
Appendix A <.Jdines tht:: variabk's listro in T;lhk' I.
To :Ichic\·c candid;lcY, :l v,lriahl\' must otTer
:10 i nfon n ation ~a in \h:11 is significanl al Ill<: 10pt."rn: nl k '\·d . Lo ui:.ian:1 hous in~ pemlils do nol
olTl."r an info nn:llio n gain th:1I i ... Significa nt at the
IO-pcrn: nt It: \·d fo r the t ouisi:ma rig counl :md
a<'."l"o rdingly cannOI become a e.mdid:lte variahle
for the fin:11 rig cou nt equation. We also tested the
pn.xlun: r p rice index for the refiner cost of crude
oil as a prediclor for the Lou bi:ma rig count. This
va ri ah le o tTers :1Il infonmllion gai n th:11 is signifi{";'1I11 even at Ihl." I -perccnl Icvd . so il becomes a
candid ate
the rig (·Olmt equation
Once Wt examim! evc l1' variable hy these
test... and select Gmdidate variables for each of the
nim: l.ol lisi:m:1 forec asting equations. we move to the
sC("Ol1d stagc of model h u ilding. This second stage
is a n a logou.~ to :t gentral t:lection , in w hich the
winners of c:lCh pa ny ·s primal)' election compete as
GUldidates. I !ere, \'ie allow C""dch candidate variahle
to he considen..>d in a ool1"eOlion:11 ordinary-least...qll ;l n..·~ step\visc regressio n procedure
Th b procedure selects varbbles \h:1I expbin
the m O....1 in-sa mple \'ariatio n of some i nd iGuor
thai \\'(;.' want 10 fo reca ...!. such as the Louisiana rig
{·ount . To remain in the final fo rec:.lsl ing equatio n.
a ,"ari:l b le must havc a I stati:.l ic thai b significant
at lilt: IO-pcrccnt Ic \'cl.- O nce this St.'cond step has
IX"t:"n finished for each of the variahles, we han: a
model Ill:.t, in p:1SI examinalions for Ohio, T exas.
and New r-,·Iexico, h:ls been sho wn 10 fonx;ast
w il h rl.":1....anah le accur.:lCy

tions of oil-price shocks more full y Ihan :.tandard
timc-serie ... models can . Consequently, we resol\·ed to fo rgo time-series fo rt-alsts of l1alional
\·ari ahle~ :lnd to suhstitute natio nal consenslls
fo n.:c l sts w henever they we T(" I1c(:d (;.'"(1 in the n ine
... tepwisc l.ouisiana equations.

, The lerm I'IlormaioOfl ga .... relers 10 /he percenlage redt.Jc-

IIOtl If) slandard _ra lhal results /rom the /fIdI.JsIorI of a
partICr.sIM varrable III an equatlOtl ConsIder two equaliOllS.
eOl.l8to()fl A and eqlIallOtl B. which arc dcslf}flCd 10 fOl8CaSI
/flo SBm6 ondependenl variab/8

ror

The use of consensus forecasts in (he model
\'(Ie fo llowed the time-ser ies mood-building
plUctdure described aoovc to constn lCI forcC:lsting
equations for I.."ach of the nine I.o uisi:m:t v;lri;lhles
In sim il:ir sll Kli!..'S for New Mexico (Gruhl."n ;lI1d Long
19HHal. Ohio (lloehn and 1:kII:lZ."y 1985 ). and Tn :ls
(Gmben and Long 1988b), the model-builders :Iiso
l"Ol1stnlCtL-d time-seri!..":> fOr('Clsting equ:ltions for Ihc
n:ltion:.1 indiC:IIOrs :Ind u!>Cd fo rcc lSIs from Iho~:
t'qualio ns 10 pl't:.'llict the region:11 v:lriahk-s
The pre ...enl fon..' Clsting mCKle!. ho wc \"{'r.
m U~ 1 :In:ommod;lIc the o il-price ~ h ()<.: k... of the
th ird qu:m er of 1990. It !>Cemed compelling to
(:hoo.'>C foreclsts that ca n incorp orate the impliGIEco nomic Review - March 199 1

Suppose /flal equatlOll B IS

idenlICallo BqUSIIOll A a}(C9pltllai equa loon B conlallls IWO
Jags of one acid,llOfIal varisb le Lei SEE A represenl tile
siandard 9"Of of CSllma lC 01 Co ulllIon A and SEE. be the
Slandard error of estlmal9 01 equa llon B The gain in
'nlOfma llOll of equa lion

l3

over e q uallon

A is

due 10 the

Inc/uslOll of the vafla/)/o m CqUlll1Ol1 B bul nol in equation A
The illformallon {J8m IS the percentag e r:lJfference If! Ihe
slandard errOf 0I9SIIfll81ol for BqUSDOfl B compared with
/hal of equsliOn A ThaI IS.

and F values

The I'Iforma/JO(l g8WI- vanabic I

CCfTf!-

spend 10 one IJfIOlher ,., /he following fomt
I~I

{(n - I<.

I . q~

1<.-I .. qF/I!I.

whereq IS /he numIJer 01 iJdd9d varl8bles."

B ccmpared

WJlhA. n IS IhenumlJer 01 0bSefva11Of'lS. and I<. IS /henumber

01 vanables 1'1 B Thus. Iho rnonomum valve 01 I

/hal a /JoWs
a new varl8b1e in B 10 bollflCluded .... a forecastlflg equallOl1

CCfTCSportds 10 Ihe value

01 F IhallS srgmf>eanl at /he a ,

level
T

The slepWise procedur9 we use coolalns a declSIOI1 rule
thaI reiCCls somo candidale vanables Ihal origInally ar9
mcluded 1M the IOffJCaSlmg oloualoon The rule worl<s as
follOws Suppose some V/lflsbi8 x, ad ds enoug h IOfecaslmg pawer 10 warra nl including II In some noarly final
predICllYO equa twn I/ I/le subSCqU6nt mcluslOIl of other
predICtively powerful vllflabJes (say

x, Ihrough ~J pushes

the t StallSIIC for /Ile cocffoclem of x , below thea IS level of
srgrnl>eance. lhen

x,

IS deletlJd

'rom

the equsliOn. and

Tho only CxCCpllOt'110 thiS re}eCIKXI rute was
IMIIWO own I8gs were used any lme tlI6y resr.slled '" a
kMer mean squarc CffOf /tlan ono'y 8 firSl own lag Of ono'y a
~ 1"

~" remaln

sCC(l(H1 OWl' lag. rflgiJId!ess

011 slatlS/1C II not even

one

own lag was selecled for anyequallOl1. WtI did not force any
." Even mcscdcc&Ofl rules aremr:xerestnclJv8 than those
oIHoehnSlldBalarsy(r985J, who used two own lags ." a.
cases Wc. Hoehn SlId BaIaLsy (1985). and Grueen and
Loog (1988a. 1988b) apply some speoaI, less rCSlrro;lNe
rule for /he IfIC/usIotI 01 own lags oeclluse we have S/roog
{XI(}(

behBfs aOOuf Ih6If fQfscastmg pq.Nef

,

Table 1

Information Gain Statistics
LARIG

LACIVlF

LANAG

1.6616

8.1853
8.7443

LAHQUSE

LAPV

LAMINE

LADUR

13.2320
5.8189
31 3110
2.9310

0 193
3.3464
2.1234
.6472
4.8635

3.4795
1.7916
12232
3.6038
4778
2.1176

lANDUR lASALES

Regi onal Variable s
LARIG
LACIVLF
lANONAG
LAHOUSE
LAINCOME
LAMINE
LADUR
LANDUR
LASALES

2.9453
2.4097
.7046
.2491
9.1935
9.1429
.3589
2.1244

2_1099
3.1255
3.4313

2.4938
3.5250
5.6310
7,4125
1.8068
2.0287
2.0386

6.8539
.7993
6.0856
6.6385
.8966
6.5702

.4619
.4049
4.4981
4.0652
3.2327

16.8368
15.3678
8.0740
.8573

3.8229
116948
4.3610
.7502
.2402
.6730
2.7242
88399
54624
3.0811
7.6331
6.4417
4.1853
.3555
5.6633
1.7168
.4215
1.4401
.4678

.5674
2.8264
15385
3.1039
.3479
2.4537
8.5124
7.6676
.1739
3.9612
.4410
.6688
8.5021
.3106
2.3191
1 1345
1,4018
11 .8296
2.8637

6.2078
4.3366
24.1981
14.9124
16.5106
8.8614
3.6471
5.4135
2.9158
18.3035
16.1936
141299
12.4646
' .7843
10.2659
11.7320
2.7969
9.5220
1.9768

1.6480
4.4937
5.1955
1.0702
2.5825
.2901
8.4950
6.3242
6.2224
.2257
2.0539
1.9240
1.5568
7.4026
.8842
.0152
.4339
5.5698
,3123

8.2093
1415
5.3289

4.5442
4.5699
5.9017
5.5021
5.7348
.7240
13.6964

5.8868
7.3784

1.1639

7.2933
.0570
4.1261
3.2825
1.0379
2.1945
2.9407
7.5858
1.4237
2.7128
.4620
.0158
2.7897
24073
2.9304
2.0627
.9981
1.9797
5.5017

14.5606
4.5463
5.1457
23.2370
1.5532
18.5228
2.1330
.3708
1.4553
16.3177
8.5981
4.2702
15.9263
9105
2.7067
4.2651
5.7575
13.6274
8.0893

6.8791
.1556
6.9515
.1089
6.4556
6.2576
11 .2339
5.5768

National Variables
REALGNP
.4009
GNPDEFLATOR 3.8777
CPIMSA
4.3095
USNONAG
.6921
MASOND
5.6227
INDUSPROD
1.9254OILPRICE
22,8751
PPIFGA
12.0540

M1

23858

UNEMPLOY
2.4354
TBILL
1.4855
PRIME
44341
CIVEMPL
2.1458
INCOME
2.1790
EXPORT
.3146
IMPORTS
12.4990
PERSCONSUM 1.3079
NONRES
5.1068
RES
1.4211
LevelNalue'

.10
.05
.01

F
2.39
3.15
4.98

1.9242
10.3982
9 .3603
.7220
95561
1.4076
19.0901
16.5807
3.5865
1.53545.7978
69030
1.0223
.9839
1.77541.7000

.9961
6.3207
.9743

2.2023
1.6620
1.6550
1,9856
1,2756
.9875
.3567
.8217
4.7597
.7063
.9413
1.8661
.6763
.54-57
.6943
.6606
.1108
5.9358
2.0008

1.8271
2.7846
4.9800

• For a mathemallcat derlnl\lon or the relationship beIWeen /values and Fvalues. see footnote 6
NOTE: AppendIx A dehnes the VlInab~ used In tns table .

•

F~-dt:-r:. 1

Rl"SC.· ,....• Hank of r>aUas

R;nher Ihan huilding :1 mouo.:! to forcc:!.'>t oil
prices, Wt' indutk<l quant'rly :lverages of futures
m:lrket prkt's for each uf Ihe qU:Hters to IX! forecast" For the rest of the forecasts of n:ltion:d
indicaIors. we lI,'i(."(lthe const.'n,~us fOI"{_-'(:<lsts from
the Scptemlx:r 1990 issue of Bille Chip ECOllomic
flidicato/~ :md thl: quarterly sLlpplt'rnent that also
:tppea~u in thl: S:HTIe month. T heSl..' mile Cbip
conSt;'nsus foreclsis arc b:lst.'<l on :m ;l\·erage of the
fon_"Clsts of tiff)' finns that participate in the Blue
Chip p:lIlel. so thl:Y incOfJXmHI: much infonnation
that h:ts IX'"l·n captul\.'d by m:my fo~caslcrs.q
Consensus forecasts are oftt'll relatively
accurate I n:1 sludy of seven macro<:'conomk'
vi.lriables that :tppear in the mile Chip consensus
forecasts. McNIXs (1987) finds that about only
one-fourth of the individual fOf1.:.'clsters were
consistently lllOTe accurate th:ln tIll' BIlle ChiP
consen,~llS fOrt:CISt Anothl:r group, about onethird of all foreGls! panel memhers, Wl're consisten Tly less :tu:ur:.He Ihan the consenSllS. The
n.:maining panel members, :t link: less than half
the total. wert' slightly inferior to the consensus.
For any particular variable. the BIlle ChiP
consensus W;IS more accur:l1e than mast indi\·idual forecastc.:rs. even thou~h every forecaster
was more :In:uratc.: than the consensus for at leas!
one \'ari;lble. Non!.: of the forecasters oUlperfonned the consensus for all sevc.:n variables.

Configuration of the final model
Wt' fol1owt:d the limc-series lmxleling
procedure dcscrihed above to construct forecasting equ:llions for t;'ach of the nine Louisi;lna
variables. WI: used quarterly <lat:t for the period
1970:1~1990:2 10 forecast rTloVem<:'nts in indicators
for 1991. Appendix t\ define.~ the variahles, and
Appendix B contains the n:l1ional consensus
forecasts that were used in the Louisiana forecasling equations . Appendix C de.'i(."fibes the cquations. Recill th,l{ hecause thi.~ is a time-series
model. its equations arc purcly predictive; th!.:y uo
not represent :llIempts to explain the Louisiana
t'conomy hut onl}' 10 forecast It
Ne\'erthdcss, il is i!1leresting to note how
often the purely statistical v:lriab[c.;: selecl ion rules
r<:'sult in c<..juations suggestive of the actual
structure of the Louisiana econo m y. For example.
the import:Hlce of energy t:xtr:.K"tion to Louisiana
F.cooomic

Revkw~M3rch

1991

i.~ ,~uggested by the regularity with which energyrdated varia hIes appear in forecasting equ:lt io ns.
Oil prices appear directly not only in the equations for the louisiana rig count and mining
emp loyment hut also in the equation for
nonagricultural employment. More(wer, beCluSC
i.ouisiana nonagricultural employmem is forecast
by oil prices. and such employment fOf{'CaSL'i
Louisi:ma personal income. oil prices may be seen
as indirectly foret·:lsting Louisiana personal
income The ri~ count is a predictive variable in
the Louisiana hOUSing permit equation.
Considering the state·s relatively low labor
force partidpation ralC', it is also interesting to
note thitt cbanges in rC<l 1 person:ll income forecast
c nt!), into the Louisiana civi lian labor force. While
changes in the labor force mily suggest inmigration in response to rising real income, it may
:tlso suggest :1 discouraged worker phenomenon.
The years 1980 and 198 1 not only saw Louisiana's
highest r:.Ues of growth in re:ll personal income
during the 1980s but also the decade's highest
employment r:.llcs.
Becausc few states' et·onomies vary less
consistently with national nuctllations than

II should be noled Ihal we used the prOOVC6r price rndex for
the refme,·s cost of crude 011 or, Ihe coos/ructlOll of lhe
forecasl'ng model but relied on forecasls 01 the price 01
I,ghl. sweel crvde as mpu/s to lhe I()(~st Movements m
these 011 priccs are highly cQffelaled We based our dec,·
SJQ(1/Q use two different prICes of crudeoo the avaifabililyol
e/ongcr time scncs for the reliners cost 01 crvde end the
ava,lablllty of 8 de facto consensus lorecast, VI8 futures
prices, for light. sweet crude Wecons/luc/ed a regression
equalion /0 lelale movements of one to the olher and
estimated a forecasl growll! rate IOf reliner's cost of clude.
based on the term structure 01 futuros prices IOf IIghl, sweet
crude
When following the model·buildmg procedUfe for our Louls,ane forecasllng equation. we restricted the set of nalional
vanables from whICh candidates could be selected Only
varoables thaI were /otecas/ m Blue ChIp EconomIC indi-

cators.

lIS quarterly Sl.Ipplemenl. or by means of futlJles

market prICes. were a/lowed /0 be conSidered lor cand,·
dacy Ollhe nmet8fJfl nalJQfl3/ val/abies m Ihls set. Ween
IQlJlld thelf way infO

IfIC

linal /otecasting equations Tile

nallOf"lal indicators that appear In lhe fmal forecasting equations are. tnemselves, forecast 00 tile bas,s of tile above
sources

7

1.0ll l .... i.IIl,I .... ('.... IlL·f\\()<)(I-CIII

11Jx,,'{). I IlL'

n.:gu bri l ~

w i l h \1 11Il1l Illlll ..o i l 1l;lI io n :d \ ;I r i.lh].,· .... fin d IhL'i r

w:.y

inlo IhL' llillL' for\,.·c .....1lI1g ...·qu,nion .... i ........ Inl.. ing

Table

:1 1),. Ihll'CII:l l illlh in .... i\ (If lIlt· Ilin..: Ilm..'''·: . .... t I:.ri ..
. Ih l l •....

F. II'

t!L·di n..: . . il) l ong Il'1111 i llllTe.....1
th L· AA \ h Ol1 d 1:ItL') prL'din i l wl .....;!.... e ..

L·\:1I 11pk.

1:l tL·.... (herL·.
in 1,c )lI l ..... :l n .1
11ll' p n nw

2

1991 Forecasts of Growth Rates of
Nine louisiana Economic IndIcators

C h .mgL· .... '11 1l:,lulI1,. 1 111I(·r\,.·.... 1 r: II L· .... lor(·L';t ....1 1.0lII .. i -

hc'u. . i n .!.!:

.:I1\,.· prL'd ill inl re: .....L· .... l1l11K' l,oui .... I.IIl;[

LARIG
LACtVlF
LANONAG
LAHOUSE
LAINCOME
LAMtNE
LADUR
lANDUR
lASAlES

rig COUll t On Ih...· othl'r h .llul . real G\P 1)t,,'(oIllL·..... 1
/ol'l·la .... l tn g 1.l ri.lht". in only 1110 t.'{jIl, lI nm ..... thlhl'

of LOllI .... I.IIl.1 dlll~lhk· good . . \:l1lp[OI1l11.:'1lI .lIld IllL'
1.01l1 .... i.m.1 ('II Ik lll 1.11)<11" fo rct.'.
\ l llfl·(lll·r. :-.IK"nI I)()(I·C, II ·.... fi l)d l ll).: .... 11;1ll'
in:.ln1tllll· 1I11pl ic ilUlIh for lilt' Illie I h ,1I n .ltlon :ti
1·;ll"i;ll1l..-.... pby i n ... HilL'

o f Ille

Forecast rales
(Percem)

lnd;cat()(

IXTlll il ... I\hik ,k·din ..: .... In

l orl'c .....l i n ,!.: L·qualion ....

I S JX'T'<()Il:d i n colIIL'. for e '"" ;l mp t..-.
fO!\.·l ·a.... 1 i l1 o \ ·:1:.\·:, in Lou i:.i:lll:t ['ll:t':'ool1aJ iIKllIll\.·

9 .•
.9

10
16.0
2.1

2.

- 2.1

-5
3.9

D e d i Ill ' . . i n

EIl'l)' onL' u t l llL' nilK I'OI'L·C;.I..... l ill).: L·qll:ltlon ....

NOTE: Appendix A defines the variables used
in this table.

l'(Hl ..

1:1I1h :11 le:I ....1 OIl\' nOll-oil 11:lIlo n:.1 1 :U1,lhk

Forecasting the Louisiana economy
T.lhk .! prl· ....L·ll h 11K' Ilil)(lvl'-. 11J) 1 l u n.·cl"'h
ot' ninl' l'COll01llll IIld ICllOI'" t'or LOlli:'l.lI1.1 To

qll\'llt " 'nL'I):1 ( ·r: I .... I1(: .... led l.oui .... i; t!1 :1 L'rl1p ln~ !l1t..'Il!

lInd l ....... I;lnd Illl' lI11plicliioll'" o j I ll",...",. jOI"L'l·;I .... ". 11 I..

10 f.lr ,,~·:tI..L·r .1\l·T~'gC L'''P: II1,' ' lon tlLlIltlll' n ali o n ....

\·I1\'r).:~ 11I)()tn ~l·;l r ....

of II)HO :lnd II)X I. 11t\ · .... ull . . \·

lI"l'f lill u n lll ... i d L'" P:l... 1 I1 l1(1 l1:11 1011 .... ill 11lL' ....U I L' "

I, If (11\' 11),0.\( h ( F i ).:u I"e :~) Dv.. p I( l' ....1 Illll' .l : 1'I III l it i 11

l ·\·OIlIJ!l!il'ind l(;l h l)".. I.iI.. L· ih L'l'0I101Il il' ....lrtl"·l lI rL' .

I ll\' I. II ~· I I}HO..... l .o u i .... i.lIl;l Ii.h n eIL'!, ;I).:ai ll !'L';ld 1l'd

1.l lll l "I:ll l.I ·" "l'l,'n l l'lll!1(l! " ic pI:r!(lnn;l n u,: de\ 1;1 1..: ..

11](' \'l1lplo~ tnt.'111 Jl\'ak il IlI h l \'d in I <)HI

II Ilk·" f)"olll 11K' 1I.lI lon ....

OIL'!' I ll ... ' p :I .... ' declll\· hOIl\·\\·r. Ill,,· I.oll l .... i·

'" Ihl' lOl1nJ.:Ur:ll iol) of Iht..· 1ll1)(ld "'1I~t.''''''.
I.t )lI i .... i;I11,( .... l'(fI!1( l illy ).:11)\\ ..... ;K I: l'Ier:lI l· .... . ... 11)11 .....

\' lllplo} Illl'nt 1\:l;Ii n .... :.hou l fhe Il lIIl· ... ,I .... 1.l rg,,' . 1

:md l'OIll r:IU", II lIlt I ll\' priel' o f t..· ner).:~

....I u n..'

111 1<):-;( •.

.In.1 \·\"j)o tll } lu .. Ch.lIlgL·d

. \ lth(lll~ 1i lIlltllll).:

of l.oU! .... I.uu l ·lll ploylll\ ·111 . t-. th.1l 0 1 ti ll'

II Ill'l) \'Ill'rg~ p rin: .... ld l pn:dpil ou ..ly. t h L' 11111);1\'1

11;111011. lIlil1il1~ ' ", .. h :lrL·

11la dL· I,I)llhl.ll1;1 I lllL' ()fllnl} .... ix "'1,lI l', ,,11I ):'L'

I h \· ....I ;IIL·. !:d l m).: lrom "i .H pel'"ell l III ]1),'>1 I I I :~6

h: ... d ..:d

i!ll.: d 1l1.u'I..L·d l~

!1)

11(lIll ill. 11 (;:-' 1' ..I,·di llL'd I" I.lllli 'I;I I).II'll:).::I11 .1

rl1.Hl·111 i n 11)1'1) Co m·~· I~ dy . joh .... III f t..·I :1 1I a mi

rL'('(l1 L' T'y d uring I I) K"' . :111 hough :t n nll:d :II·L' I: I).< L· ....

....I T \ i( l' 1'l'I:1(l·d illdl..... lril·.~ 11 ;1\ l ' 1l1 ; 1I1 l'l ll ~ 111,'f L':! ~ l'd

le lr 1I 111.... 1 (l j' llll' .... 1: 11..:· .... ,lg.l.!rq':;l I\· ind icl l())" .... )"L"

I l ll'ir ~ 11. lrl· " (II (1 I l'l";l J l l' I'l p lll ~ lllL' n l '1'111' l· l ll·r).< ~

1ll:l Il lL'd h ..: loll 1\)1';(1 1L·1t.'1, . :Illd Ilil' .... tIlL· h:I" .... i lln·

indu"lr~ Ivlll,u n .... \ L'r~ i mpOt1:lTlIII . I.I IUi .... l.lIl. l . I lUI

).,:1'(11\11 1111111,: :.111\\J~ 111:111 Illl' n;UI(1Il L'IL' )") ~ ·l·.lr I n

inll'\·.I ..(· .... in 1111 p r iel '" nIl IOIl).:L·r 1llt..';l ll .1 .... t1llllll 10

JI) h·<).).<ro ........ 1l01ll 1ll,IIIXT-.e,n ,t1I1l(,(II)lL· rO ....l· "' ,(I

1I1l' J.olII .. i.I1l.1 L'COllom\':'" II1\'\ IInu' d id

fk.· ...."·111

1).lI lon.111I hlll (lnl~

5

I Ix·rn·nt III 1 ~)[ 1t "1.1I),1

\ hllllUgll 1.IIIII .. j;l ll;I " l'nl pl l)~l l ll'nl gr,,·\\

"lllbl.IIlIL I IJ~ b ....I ..:r Ih:lIl tilL' n;lliol1" dll r in).< 111\·

'l'h\' 1')<)1 f OI'l'C"'h indll(k I11ll1.kr:Ul· (0 h ig h
l·\p.I O.... I{)1l III 1 ;m.l h lt.' .... reL llcd III I h ...· 01 1 .lIld g.l ....
\·\tr.I( lllltlllldu .. t0 hul .... 10\\ ).:rll\\ lh .•tt k""'1 h~
n'U1< ' I1.1 1 .... t.ll1d.ll1.1..... in thL· .. 1, 1(L··." .1).:).:I"\ ·).<.l l l· Li l liit'
fllf,\· ind ll,ltl)t... Ik.11 1'\"1',,111:11 I !H, U))L' ).:\'(11\1 11 I ....
IlItld "' I~lll' \\ 'h ik- l lll'rl' i:. ;l good \1\':\1 III 1;l ri; lI ion

in
AI<tsJ<a New Mt1I 'c() ()j..1<111011~'
1{",M .mtl WYOUJ/1l9 afO also _9,(,,,,00"(:'''9 51,>les
filo)

•

olhef he sl•• rcs

tl1\'

prnl ll'1l 'd 1lI01 "'l11l'n h in nl h e!' I.lllIi .... i:ln:1

\'U ' 1lI111111 IIltliL;l tOr... ( 1)91 i:-. gl' n er: l lt~ II 'I\:I';I"'! .1 ....
.1 IlI).... 1(ill· ~l·.lr lor thL' ..;tatt..· \ \·I\·l'Il w k:. ..... dl· ... pll ....
r t'tkr,1llk SC,' r'\'': U.:m k " flh l b s

Figure 3

Figure 5

Nonagricultural Employment

Louisiana Housing Permits

InOex, JIIfIuary 1980.100

'"
'"

.....

n,

-

'"

""
''''

'"
'"

.

SOU RCE US Bureau 01Labor

...
...""
...

$+a1lS1>es

SOU RCE OF PRIMARY OAT" : U S Bureau 01 the Census

Figure 4

Lou isiana Nonagricultural Employment

"

""",,, f\.- ............

,."'"

""
SOURCE OF PRI MARY DATA, U S Bureau 01 Labor Sla~StlCS

imTL·a.......·d oil prit"o.:... Illl· mudd dol..''' not pn:dit"t:1
hool11 for Loui .. iana

Thl: I ·pl..'rc~ nl raiL' of L·XP:ln.~10n prL'(licIL·d
for 1.001l~i:tna n()t1a~rinlhtlr:d \\";Igl..' :lnd ..alar),
l'mploymcnt h mort.' than I hree time.. as f;lSI :1;'0
\\"h:1I W:I~ :Khl:llly p()~ll'd in 1l)H() (0 ..3 pcKl'nO,
Fi~lIrc 'l til·piet.. I);'~t Illlctlialion" ill Loubian:I's
nOlla).trlcliliur.:t1 W;l!-ll' :uld ,alar)" I..'mploynlL'nl.

10gL'l hL'r \\ illl Iht.: lorL'Cl..t v:l llIl'S 'I1H: figure
.. ho\\ .~ thaI. while Illl' forccl .., gl'Owlh r:lle I~ low
hy nalional .~I,lnd:tr<.h. it i .. compar:lIivt'lr high hl'
rL'i.:t:nl LOIII:.i:m:1 ..t:u1d.lrd.. .\'L·\ enhdc::.... the 1991
fort:G' . . 1 of 1.0 u i . . I;ln:l L·mployment ..till doc:. not
n.·ach Ihl' It:\·c:! . . of thL· .,:arty 1980s.
Likc\\i . . e.tlu..' mild O.9'pL· rcen ll~lI l' of
expan.. ion fo rt.:e l'l for I.oubi:ma ·... bhor force i... a
good de:11 fa:.lcr 111:111 Ihl..' dt.:dl1ll' it realizL'(1 in
19H9 (OA pt,·rcL·nl) Till' 199 1 fort:"<.·;I.""t.~ abo
:.lIAAc:.t f:I,'t.:r I: lhor market indicalor /-:ro\\,lll than
whal :Ippr.::trs 10 ha\'L' O<XUI'fL·d III 1990. \\o!"L·on:r.
the lll(xkl foreGl .. . \...."itnJllg~·r gro\\th III rL':11
pn:.on:ti income (2 I \ )Cf(:~nt) I h:m for L'ithL'r
nonagriclIillII':ti ,,:mploymL·nl or the lahor fore..·.
The 1990 ;nl'l"t.:;I~t.::' in o il rricL·s hl..'lp to
forcca ... t a 9o.percent illl'1"e;t."i~ in drilling activity
:I nd a 26·rx·I"CL·nt incrr.::tsL' in mining employment
in l.o11i~iat1:t for 1991 ov(.."r 1990 ['ositivL' 1-::lin." :Ire
;, ).,0 fore<.<ht fo r rL':11 .. .:til'S 1;(.1( eol!cctiot1 .... which
\\ill irKrl..'a"L· ';.9 pt.·1"("..'nt in 1991.
Thl..' highL· . . t r:.ttL· of (,."xp:msiol1 (If any
\'ariabic fOI" I..oUl"ibna in 1991 is housing \x'llnit...
\\ hieh :lrL' prr.:di{:ll·d 10 intTe:lM! 16 pert.:enl I\:.
Figure 5 .. ho\\:.. L·n.'11 with this r.tte of \.'xp:ll1sion.
hou:.ing p<:rJmls n:m:lin f:lr hL'IO\\ their kVl'b of
mo... t of t h~ 19HQ-..
'111t: pn.·uil1ed expan .. ion:. in i.oui,i:ln:("i
pt:r:-.onal JOC·OI1K·. lahor fo".-e. nOnf;ln1l employrnt.:Tlt .

•

urill mg at·[i\"i[y. Ilou ... ing periniI:-.. ;lnd re[ail S:IIe:[ax colll..·(·[ion ... do nol im ply ('mplnymenl gro\\ t il
in Ihe :.1;111.:':- manufa('luring "'I.:clor Durable good...
empl(lym(,.'nt f:ll1:-. Il)' 1.1 percenl. whih: nlllldur.lble good ... ('rnployment ... Iip:- h y 0.5 pern:n(.
Despit(· I lie continued impon ance o f energyre!all.-'(I C\'t'n", 10 lI)l' I.oui.... iana l·('·onOIll}'. thc :-t'l
Ilf flll"l.-'(.':I:-ling t'(jll:tl i()n:- incllt(k ... fifteen n:llion:ll
fOI"l.·c.l"ling \·ari;lhll.·:- nut of t he n il1l'lcen origina lly
con:-kk'red in thl' fir-I :-t:tge of Illodd bllilding
Appl..'ndix U (it'pit"l:-. Iht' foreGI:-!.... of till'....c \ ':IIi:tblc::md oITt'l":'> :t pil'lllI"I.· of ... Iow on:'l~ llll·xpa n:.io n i n
[Illal (IUl pul :l nd empl(IY!1lell!. For 1991. l)(lIh re:ll
G~ P and I is n()n:twindHIl~lll'mplo}"lnt'nl
i ncre:tse hy I Iw rn'n t
It is intt'rt':-ling to nOll' Ih:Ll. (lespi\(;.· Louisi:tna 's
typical (kvi:tliolh from U.S p:lltcrns of t·x[u llsion.
Ihl.: fOfcGlst for I r.s no nagril'u itur.t l wagl' :tnd
sal:IIY l' l11ploYllll'nt Wmvl h for 1991 is i(kntict110
Ihat for l.o11i ... ian:l II :-llOUld hl' noll.:d thai Ihl'
fOn.:GI:-1 ItlT i.<llli:-.iana·... ntmagrinlilural !.:l1lpl(lY111en[
growl h in 19l}1 i... ha:-'l'd. in pan . on foreGI....1:-. of
SOIllI: dedin(':' in oi l pli('I..·:- in Ihe -.c<:oml half
of th:lI ye;lr
I.<)ui:-i;lna·s lal xlr market indic.llor:- wew more
r.lpidly in 19HH. w hen IIIl" :.Iate was l akmg o il'
aflt'r con......nLlivl... :tnnuallkdme:. oycr the period
198s-B7, Ihan Ihe J-tTO\\·th our model pn.'dicl:-' for
1991 BUI a ... i<lt' frum 19HH. thc fOrCGI'it for 199 1
impJi!.:., Ihal loui ... i:ln:(s (,'!l1ploymcnl :tnd lalx)r
forcc will g row b:-.[er I ha n a[ ;tny lime Mncc Ihe
early 19HOs. I.oui:.i:m:l will nOI boum in 1991. hut
the ye:lf will hI..' better Ih:m almo:-I :my rear :-.inn:
Ihe IXlom yeaTS of :1 d!.:Glde aJ.:o

10

Conclusion
l.olli:-.i:ln:(:-. Cl'onomy dillep. :-lllht:tnti:tlly
from lil:1I of Illo:-t ... tatc .... hOlh in ...Iflll ·turl..· :tnd
dYlla111il·:-. Stfllctur.llly. tilt' I.()ui:-.i:ln:l l'(,(lIll)IllY i:Ullll:-U:l l in llle cxtent 10 whid! till' l..'nl·l'/-:y industry
dOl11inatl..·:- it Ix,·all-.c of Ihe profound t-rfl~CI of
cncrgy price 1l10\'c melll:- on till' :-.t:l\t·, l,oui:-.i:tn:t·:-.
c("(lIlornit" Illl('lu:lljon:- differ frolll lhl' n;ujon·:-. not
only in rn:lgnilude bu[ :1 1-..:) :-,ollleliml'''' in dirl,('llon
Alt hou/-:h \\C u ...n l :-.i:tli,ticilly h;l:-l'(l
d •..'cisioll fuk:- 10 \,:o n:-t fll<:1 a fOfl·c l.... ting model.
cncrgy·rt"!ated \ ': In:lhlc.... appcarl 'd in equ;lIlOlb
not only for thc 1..ouhian'l rig ('OUllt and mining
cmployment hut al:.o for th \;.' ... t:llt·· ... nonawil"llltllr:ll Clllpl(lY1111:111 l\'1<lrC(l\'l'f, t'nt'rgy prlt.·!.:s
indirt'Ctly I(lf!.:ca:-I [.()[li.o;ial1a ]1""rsl)n:ll in("(lIlK' :lnd
houo; ing permils
Ik:-pitc 11K' pecliliaritit's of [hc l.ouisbn:1
,,·("(momy. SI..·\'cr:.11 \·ariahlc:-. rdlcl'li\'l' of national
tlul·l\lalion... found their way into tlK' fOfl..·c:.lsting
l'(lu:lIion:-.. 1\":ltion:l1 e\'I:Il1'" dearly h:IVl' f( ll"l.·l:l:-'Iin).:
pO\\l'r on Ihc [.<lui ... iana t:<;onomy. l'\'('n thollgh
tiK'ir I)()\\'('r rnay Ill.: le:..... lh:!n for 1I10:-.t olhl..·r ... t;ttl·...
Our model fOR'(."";l:.b ).:I1>wlli in 11)9 1 for !'>I..·\·l:n
( If the nine 1.<)ui... i;lIl:\ CC()I)(lmit.· \':Iri:lhk':-,. '111e Ill( xld
prl..'dit.·ts mOlll-r.lle 10 high l.'xp:lI1:-.ion for \';tri:lhle:-.
rd:llt'd to 111(' oi l :tnd g;ls l.'xtr.ll"t ion indll"'l!), in the
St:ltC. hut gro\\·,h III LOll i,'ii:l na ':- :ll-Q-trl·/-::ltl· lahor
fOfl'(' indiGlitl["') \\"illl)!,;: relalively ... Io\\". l{c;d personal
in('ollll..· will :-bow IllO<..it-r:.II1.: incr!.::l:-<:'" Tilt' upturn
in oil pri('es d uring 1990 doe:. not ... ignal a 1)(IOm
for I.Ollbi:II1:1 in 199 1. hilt t'xp;l!l ... ion will he more
r:l pid Ihan for mO....1 rece nt Yl.'ars

Appendix A
Definitions of Variables in Forecasting Model Equations
(Quarterly data; monetary values expressed in
1982 dollars)

CPIMSA :::: U.S. consumer price index for all
urban consumers.

State variables

AAABOND :E Moody's Triple-Acorporate bond
interest rate.
INDUSPROD ::: U.S. index of industrial production.
OILPRICE = U.S. producer price index for
crude petroleum.
PPI FGA ::II U.S. producer price index for finished goods.
M1 = M1.
UNEMPl OY:: U.S. civilian unemployment rate.
TBllL == 3-month Treasury-bill interest rate.
PRIME ::: Prime interest rate.
CIVEMPl = U.S. civilian employment .
INCOME", U.S. disposable personal income.
EXPORTS = U.S. expo<1S of goods and seMces.

USNONAG • U.S. nonagricultural employment.

lARIG ::: Hughes rotary rig count for Louisiana.
LACIVLF::: Louisiana civilian labor force from
the U.S. Bureau of Labor Statistics
household series .
LANONAG ::: Louisiana nonagricultural employment .
LAHOUSE ::: louisiana housing permits.
LA INCOME ::: louisiana personal income.
LAMINE ::: Louisiana mining employment.
LADUR ::: Louisiana durable goods manufacturing employment.
l ANDUR :::: lou isiana nondurable goods
manufacturing employment.
LASAlES :::: louisiana sales tax revenue.

National variables
REALGNP:::: U.S. gross national product.
GNPDEFLATOR "," Implicit GNP price deflator.

f.":"ll.,lIIic Rev ie w _ ,\l arch 1\)9 1

fMPORTS = U.S. irrjlOrts of goods and seMces.
PERSCONSUM:::: U.S. personal consumption
expenditures.
NONRES :: U.S. nonresidential fixed investment.
RES = U.S. residential investment.

"

Appendix B

1991 Forecasts of Growth Rates of National Variables

Forecast rates
(Percent)

Variable

1.0

REALGNP
GNPDEFLA TOR
CPIMSA
USNONAG
AAA80ND
INDUSPROD
OILPRICE
PPIFGA
UNEMPLOY
T81LL
PRIME
INCOME
EXPORTS
NONRES
RES
NOTE Appendox A (iel,nes the vilflables used

Il

4.3
4.7
1.0

-.3
1.2

13.8
4.0
.5
-.5
-.5
1.3

2.7
.4

-.7
~

Appendox B

ft-deral Rescryc Bank ofOaUa...

Appendi x C
Features of the Forecasting Model Equations
State variables
DEPENDENT VARIABLE : LARIG

Variable
Constant
LAMINE
LADUR
PRIME
OI LPRICE
R2 = .4 180 ,

,

R = .3857,

Lag

Coefficient

Standard
error

t statistic

0

-.0129
- .7086
2.2273
-.0199
.6390

.0094
.4185
.5615
.0080
.1161

- 1.375
-1.692
3.967
-2 .474
5.502

SSR = .4695,

SEE = ,0808,

DW = 2.031

DEPENDENT VARIABLE : LACIVLF

Variable
Constant
LACIVLF
LAINCOME
REALGNP
GNPDEFLA TOR
PRIME
EXPORTS
R2=.3813 ,

Fi2

Lag

Coefficient

Standard
error

t statistic

0
1
2
1
2
2
2

- .0062
-,2452
,2049
,1722
.7593
-,0024
-.0529

.0029
.0999
.0916
.0963
.1729
.0008
.0312

-2, 156
-2,455
2.236
1.789
4,391
- 2,914
- 1,697

= .3283, SSR = .0048, SEE = ,0083, DW = 1.966

DEPENDENT VARIABLE: LANONAG

Variable
Constant
LANONAG
LANONAG
LAC IVLF
LADUR
LASALES
LASALES
OILPRICE
EXPORTS
NON RES
RES
R2 = .6608,

Lag

Coefficient

Standard
error

t statistic

0
1
2
2

,0034
,2888
.1121
,1693
.2141
-.0390
.0322
.0214
-.0601
- .0908
.0329

.00 11
.1135
.1004
.0786
.0527
.0156
.0141
.0083
.0251
.0358
,0128

3.041
2.545
1, 11 7
2,153
4,064
- 2,499
2.279
2,578
-2 ,399
- 2,535
2,583

1
2

-,
R = ,6094 ,

1
2
2
SSR = .0028,

SEE = .0065,

DW = 1.968
(Continued on the next page)

13

Features of the Forecasting Model Equations-Continued

DEPENDENT VARIABLE: LAHOUSE
Variable

Lag

Coefficient

Constant

0
1
2

-.0157
-.3074
.1009
.3905
3.8016
3.9341
- .1348

LAHOUSE
LAHOUSE
LARIG
LACIVLF
LANDUR
AAABOND
UNEMPLOY
INDUSPROD
PPIFGA
R2 ", .5376,

1
2
2

-,
R :

.4755, SSR "" , .6242,

2 .8222

3.4075
-3 .5255
SEE: .1557,

Standard
error

t statistic
-.484
- 3.247
1.158
2.093
2.033
2.117

.0324
.0947
.0871
.1865
, .8704

1.8587
.0391
.6115
1.8770
1.6764

-3.444
4.615

1.815
- 2.103

DW: 1.99

DEPENDENT VARIABLE : LAINCOME
Variable

Lag

Coefficient

Constant

0
1

.0017
- .2213

LAINCOME
LANONAG
LADUR
CPIMSA
AAABOND
INCOME
NONRES
RES
R2 ", .5746 ,

.6689
.1697

,

1
2
1
2

.3318
- .0086
-.2'50

.0855
-.0956

R : .5246, SSR: .0041 , SEE : .0078,

Standard
error

t statistic
.713

.0023
.1021
.1361
.0647
.1283
.0022
.0890
.0442
.0427

-2.168
4.9'6

2.623
2.586
- 3.950
-2.416
1.937
- 2.242

DW : 1.889

DEPENDENT VARIABLE : LAM INE
Variable

Lag

Coefficient

Constant

0
1
2
1
1

-.0'47

LAMINE
LAMINE
GNPDEFLATOR
LADUR
TBILL
OILPRICE
R2 ", .7612 .

,.

,

.3887
- .1163
.9695
.5025
- .0039
.1207

R : .7407 , SSR: .0197, SEE: .0168,

Standard
error

.0053
.1258
.1005
.3533
.1183
.0019
.0262

t statistic
-2.779
3.090
-1.158
2.744

4.247
-1 .980
4.607

DW.2.095
(Continued on the next page)

.'...Ito ..... l .h'.!oo ....... .. Ib ll k of Dallas

Features of the Forecasti ng Model Equation s-Continued

DEPENDENT VARIABLE : LADUR

Variable

Lag

Coefficient

Standard
error

t statistic

Constant

0
1
1
1

-.0014
.7816
-.0651
.3548

.0019
.0729
.0299
.1536

- .740
10.727
-2.179
2.310

LADUR
LASALES
REALGNP

,

R2 = .6788.

R = .6656,

SSR = .013,

SEE = .0133,

DW = 1.899

DEPENDENT VARIABLE : LANDUR

Standard
Variable

Lag

Coefficient

error

t statistic

Constant

0
1
1
2
1
2

- .0005
.2060
.0095
.0032
.2226
- .1286

.0012
.0574
.0056
.0011
.0638
.0507

-.456
3.587
1.689
2.931
3.489
-2.536

LADUR
LAHOUSE
TBILL
INDUS PROD
NONRES

R2::: .4508,

-A, ::: .4121 ,

SSR = .006 , SEE = .0092,

DW = 2.146

DEPENDENT VARIABLE: LASALES

Variable

Lag

Coefficient

Standard
error

t statistic

Constant

0
1
1
2
1

.0315
1.0839
-4.6418
.6855
.3028

.0083
.3018
1.3394
.2781
.1017

3.815
3.591
-3.466
2.465
2.977

LADUR
USNONAG
NON RES
RES

R2::: .2585,

,

R = .2173 ,

SSR = .1621,

SEE = .0475 ,

DW = 1.92

NOTE : Append'x A de/lnes the variables used tn Appendix C
SEE .. Stal'ldard error of estImate
SSA .. Sum 01 squared fesiduals
OW .. Durbin-Watson stati sti c.

IO ... " .. "" k Rt'vicw -

\bn:h

I ~~ I

"

References
Brar, j agjit, P. Pete ChonR. and Ralph W. Lang~
0989a), "UsinR Lahor ~b rket Dala in Tr.u:k ing
a State Economy," Delta Busilless Review I ,
FallJ\Vimer. 17-25.

- - - . William C G ruben , and Thomas O.
Fomby (1984). "Timc-s<;rics ForcC'.!sting
Models of the Texas Economy: A Compa.rison,FL'<ieral Reserve Bank of D:llbs EcOllomic
Reviell', :\l ay. 11- 23.

~GS P

Analy.
sis: Louisiana Economy lluough a New
Lookjng Glas.'i,~ Delta BIiSilless Review 1, FalV
Winter, 31-41.

- - , - - . ;md - - (1989b).

Brown, Stephen P.A., a nd John K. Ifill (988).
"Lower Oil Pric~s a nd S I :It~ Em ployment ,"
C0 l1temfJ0 rtlry Policy Issues VI, july,

60--68.

Gruben, WiIli:Ull c., :md Willi:lln T. l ong (l981:1a),
~Thc New Mexico Economy: Outlook for
1989: Fcderal lkscrvt: B;lnk of Dallas
t.COIlO lIIlc Review, November, 21-36.
- - -, and - -- 0988b). "Fort."C:lSting the
TcxlIs Economy: Applic;ttions and Evaluation
of a Systematic Multiv:lriate Time-Series
Model,- F(.-der.!1 Reserve B:mk of Dallas
Eco llo mic Rc!view, january, 11 - 28.

Hoehn , j ames G . and James J Bala:rsy, Jr. (1985),
"The Ohio Economy: Using Time-Series
Characteristics in FOTl.'CJsting, - Federal Reserve
Bank of Clc\'dand Working Paper No. 8508
(Cleveland, December).

t6

l\kNcL'S, Stephen K. (1987). "Consensus Forecasts:
Tymnny of the ~'Ia ;ori t y?" Federal Reserye
Bank of Boston New England Eco llomic
Rel/iew. November/ Decemher. 15-2 1.

Rosenblum, l!:trvey, Stephen Bro wn , Evan
Koenig, Keith Phillips, and Mine YikcJ (1990),
"Oil Prices ;lI1d the Econom ic O utlook,"
l:cd cr:ll lks(:rvc Uank of Dallas 17Je SOllthwest
Economy, November.
Schmidt, Ro nald II (1989), "N:lIural Rc:.:SOllrccs
and Regio nal Growth," Federal Reserve Ban k
o f San Francisco Ecol1omic Review, FlIll , 3- 19.

Sherwood'<:lll , C:lrolyn (1990), vAsscssing
Regio nal Economk Stahil ity: A Portfolio
Approach," Feder.!1 Reserve Bank of San
Fmncisco Economic Reviell', Wime r. 17- 26.
- -- ( 1988). "Explo ring the Rebt ionsbi ps
Between National and Regional Economic
Fluctuations, " Feder.!1 Reserve B:mk of San
Franci:.co EcO ll om ic Review, Summer, 15-25.

Ft.-deral Reserve Bank of Dallas

Kenneth M. Emery
Economist
Federal Reserve Bank of Datlas

Modeling the Effects of Inflation on the Demand for Money
I.·Clll ... l· ,hI.' r . :dcral Ik..l,:rw..... mo~t import:lnl
1'l·~po n ... ih i1it )' I... ('onll'olli ng thl.· \ ~t1 UI.· o f
Illonq'. pol iq'l1lakcl"! arlO "CI)' conl.·l.·rnt:d wilh
forl.·cl ... ting t he public's dcmand for money.
Durinj.: thL· pa~t fiftccn }'L·:lr... however. mtxlding
11lL' tkl11:tlld for money ha ... 1)L'cumc \'cry prohk'll1atic 1'('1':1 numhcr of rl.·a .~(l!lS I'riltlar}, among tiles..:
l'e:ISOlh is ,hI.' d)'ct'i intl:i tio n Ii:1.'> Ii:Kl. tx)t h
din.'l·tl} and indiret'ilr. on thl' dL'Ill:tnd for money
Tlt t .. :llliclc focu:...:.... 011 thL' qUL·..IIOIl of how
intlation .. I!ould IX::.I he ir1l'Orr)(U:II..:d into t:mpiric:t!
modd .. of the dl.'m:lnd 1'01' mOIlL·Y. To model the
dfL"l'b of intblion, 1'I,;'!'>Carclll'J"o h:I\l,: traditionally
:t.'NIIllI."(1 [Iial including a nominal interbl r.lte on
.. holl-II.:nn bond.. would hc :'UrriI.K:nl .\Ion: ft'1.'I.,tlIl~, Il(I\\I.·\l'r . :-.t:)IllI.' (I.·x·an:ht:r- ( l bIXi. Hendry.
:md :'Iarr 11}HH. and ilelll.'! and !\k:hr.l 19$9) ha\'l'
Ix:gull 10 diTL't:ll} indudl.· Ihl.· inl1:lIion r.ue- in
addilioll to the nomin:tl inll.·ll.,...1 r.IIL' -in their
l.·mplrR,:tlmodd:-. \X/hill;' IheN,; :-.llIdiL.... :iuppiy
i1l1por1:tnl l.·rnpiril':11 finding~. Ihq' providl;' "1..'1')' linle
IhL'ofelil':rl Ju ...llfienion for indudltlj.: th~ inf1:rtion
r.HL· ."l·P:Ir.11c1r in m(xlcl~ 0 1 11]oney demand
In 1111... :1I1idc. I re, 'iL·\\ t hL· Ir.ldllion;rl 0pp0r1Unit~ {'O.~I :If).,:ument for mdudill~ ~hot1-t~rm intL'I'L'SI
r:III.'." in lIlone},-dL'm:rnd IlHxlcis :!nd lhcn prt.:.x·11I
li1eorL'licl1 r.llion:lks for : II.~() dilt'a~1' induding thc
inllalion l':r1l.· T hi:i d i;>.(.'\I....~i()n is l'ollO\\'L'{1 by :rn
L'xamin:ll ioll of how Ihe CllrL'I').,:":I1(.:!.: of intcrt..... t1X::lrinJ.( Illorll'), Illay h:I\'e ~11tl.'r'(:d Iht:: dfL-'Cls of
intl:ilion 011 lilt.· d!.:lIIan<i for I\IOllq'. In thl' b.~t p:111
of Ill..: all ide. I ex:ulIinL' \\'hL'IIlL'T inf1:ltion indl..'(xl i.~.
or 11:1 .. I)(:t·n. ;1 ... i~nificlill ,kIL'rrllin:tnt of tht:
deIHand for Illoney

B

The importance of inflation in
explaining money demand

liL's L'conomists h:I\'L' had in mexlelinj.: Ih~ demand
for mone}, durin~ Ihe pa:-.t t\\'O dccadc:i, During
thl.· 1970.~. high mtbtiorl r.lIc~ lr.m~l:rtcd inlo high
nominal iOlL·fL·... I r. lles :Ind. ('onsequl;'nll y. a hig h
opp0r1unily Co.,1 of holding moncy. Whl.'n (.:O1llhi ned \\'il h kg:!1 impt:tiim..:nts 10 thL' payment of
inle re~t on monL')'. this high opporlunity l'O.~1
induced finand:11 innm'a tions thaI :I1IO\\'L,d firrll.~
I() m:rn:lg<.' 1l(JIlintl.·rL·s'-I)I.'~lring rnonq' m()1'1.'
d()~dy I In p:lrt l)(:cau:-.e 01 tbe~L' innov:Hio lb.
~Iandard moncy-tlL'ma nd models I X'g~tn to ()\'l'Tpnxlicl Ihe amount o f mone}' b:ll:tncl.·s Ille public
desiTl.'d to hold In till.' I~s, hO\\,t·\·t:r. IIK'se
lIIodding I;:rror- Wl·(I.· (I.·'"crscd wben tr.lditional
model:i hl;:~:m 10 umIL'rpredicl thL' :lInounl of
money ha];tn('L·... Ihe p"hli(' desired [0 hold . .tll
1'I!/oCJ~),. thl.' mtlo of Wos.... n:lIional pnxllll't (GNP)
to the ,\II money "'Io(:k, hegan to f:tlt. tlil!'" turning
from the upward lrend it had displayed throllgh011\ the po~lw:l r pl.'rioG (Fi}{urt.' I ).
Two h}]X11IlL':-.e:-. l.·tI\crgcd to L'xpl:tin Ihis
more rl;'('L'nt phL·llolllcnon. The fir:-I hyp()the:ii.~
maintained Ihal Ihe Inm)\':ttions thaI IX'j.::m in the
1970.... :md ('onlinlll.·d \\ illl Ihe inlroduct ion of
nc~()tiabie ()rdl'r~ (II' withdr.twa] (NOW) :1l'c()lInt:i
(i nd uded in ,\ ] I J. ;t IOIl~ wilh t he simllltam:()us
phascout of inl efl's/·r;t lr.: {'dling rq.:lIl:uioIlS, !t:d to
;tn inCfe;lSL: in thl.' (IL'rn:lrld for 1\'11 money h:tbnccs

I WIsh ro thank VI MIChael ColI. Evan F /{oerng and Joseph
N Ha slaglothelpiuleotrmenlS A/letrOts. however. 8re my

~"

_arO'l$ IIIC/uded f!e'N COfpcxate caslr managerCC/lfllQIJCS such 8S repurchase agreements oerte.-

rhese
ment

..tormall()'l systems and f!l!.t/II ms/lumen/s SUCIl as lock
~es

1nl1.I\iorl h.l:-. hL·I.·n an illl1x,rt:ull lactor
nllltril)lliing Ix)th dirL'('liy :llld indifL'clly to ditTinll-

coo/rol dlSlXltsemefl/ and payable thtough drafts

SeeHcslet ( 1ge1lOt POIret, $Impson and Mavsk.opi(l 919)
lot a d,scUSSJ(ln aI /heS(! and O/het .nnovaIO'lS

17

Figure 1
Levels of M1 Velocity

",
•

,

(Cox and Rosenblum 1989). The second hypothesis stated that falli ng inflation rates in the ea rly
1980s increased the demand fo r money, leading to
a fa ll in MI velocity (Gavin and Dewald 1989,
Rasche 1989). Both hypotheses assign infla tion a
major C'dusal role in money-demand behavior
during the past two decades.

Modeling the opportunity cost
of holding money
In the empirical literature on the demand for
money, a short-term nominal interest idte, R, has
commonly served as a proxy for the opportunity
cost of holding money. : Because money has not
traditio nally ea rned inte rcst. the real return o n
money has been cast as the negative of the inOa-

, Recent e)(cspllOO$ mclvde Gokifeld and Sichel (981).
8aba. Hendry. and SIIJ" (1988). and Het~eI and Mehra
(1989) &lverat studieS have 10000 that inflal/Ofl. In adell/I()f)

to a nommal interest rate. IS a sign/lteant determinant of
fi"OQ"Iey demand durlflg pe:r1Ods 01 hypennf/a/1On These
sIuCIeSIflCVdeCagan( 1958). FltJt!ktJI(I971). andHartJerger
(1963)
• The IdenbtyA •

I .. 1\

rs known as /he Frshef eqvatlOfl

Ex

ante. infl8tlOl1lS rep/a(;ed IJy exp«ted inIIatlOfl BehaVlOfally. /tte Idea IS thai /efIder$ will set the Iendong rare. R WIth
/he fft/JlIZallOfll1llJllflfltJllOfI erodes /he f)<JfCfla5Jf)fJ power 01
money /flal is I~ Therel()(e. II lICluallllflatlOt'r equals
expected inf/arJOfl. the ~ eams a raal rate 01 return r

.8

tion rate. The real return on honds, on the other
hand. equals the nominal return on bonds minus
inflation (r - R - It). The refore. the opportunity
cost of holding money is R-equal to r + n-thc
difference betwcen the real return on bonds and
the rea l return on mo ney.J Infl ation is clearly onc
comJXlnent of the opportunity COSt of holding
money. Inflation diminishes the purchasing JXlwcr
of money. Therefore. when inflation in<.-reases,
the demand for Illoney falls.
O n the basis of this simple analysis, one
might think that empirical models using a shorttem) nomi nal interest rate to capture the opportunity costs of holding money due to infla tion
would be adequate. However, there are reasons
to suspect that models that incorporate only a
short-tenn nominal interest rate may fail to
capture the full cffects of inflation on the demand
for money.
The inflalion rate as a direct, significant
determinant of the demand for money
Alternatives to holding money_ One way
inflation may din::ctly cnter empirical models of
money demand concerns precisely the issue of
money substitutes and opJXlnunity cost. By including only a short-tern) inte rest rdte in money-demand
models, the implicit assumption is that the only
alternative to holding money is to hold short-teml
honds. Several economists (Caga n 1956, Friedman
1956, Frenkel 1977) have JXlintcd out that during
pt.·riods of vel)' higb inflation rates, asseL'i other
than short-tenn bond,; may scrvc as substitutes for
money.
Considcr individua ls' behavior toward real
goods and services during periods of high inflation. As inflmion rises and becomes particularly
t:.Lxing. individu:.ls may acceler::l1e their purchases
of real goods :.L nd services rather than hold
money, o r they may choose to store their wealth
in physical assets, such as land or gold. The idea
is that these physical asSClS are immune to the
devaluing effects o f inflation, whereas noninterestbearing money i.s not. Even during periods o f
moderate inflat ion. such as thc 1960s and 1970s,
assets other than s hon-tcnn bonds may have
served 3S substitutes for money holdings. If asselS
other than short-term bonds serve as substitutes
for money, the implicatio n for empirical models o f

Federat Reserve Bank of Dallas

nlonl:Y d~m:tnd is that t h~ nominal r;.l(I:S of H:turn
on tll\..'.'>C ()dl~r as.~ets should he indu<k'd in thl'
modd. along with [h~ nominal return on shorttl'nn honlb.' And, ;I.~ Fril'{lman ( 19'>6) pointl'd
out. if L"~r1ain phy:-.ie.ll goods chac yidd chei r
return in kind scrve as Mlhsticut~s for mont:y
holdings, th!.:n th~ nomin;ll r~[Urn on cheSL'
physiGd g(xKb-in chis (';.lSC tIll' innation r.Ltl'should Ix: induded as ;1 det ~rminant of money
dl.:'lll:lnd ,..elxlr.lIdy from the nominal interl'st r.lIL'
on IXlOds.
AmJtILL'r illustr.lti(m ((m('erns l(lO~ -l ~n n l)(lOd..
SUPI)( )......... that I)()tll ...hoI1- :md I()n~-tenn I)(mds ;lrc
sllh~titlltes for money holdings ;Lnd that inflation
incrcasl·s. If a given llloney-{km:md l1l(xiel ~x­
du(ks the mte of return on lonK-term bonds. the
m( x ld may tt il to det~L"t the substitution of money
inlo lo ng-tl:l111 honds (';I\lsl:d hy the higher infl:ltion
er(x ling the value of money. In this case, the r.lte of
rctum on shon-tenn l)()n(is will fail to Clpture tIll:
full opronunity ('ost of holdinK money. and
int1:tlion may have :m independent effL'{'t ..
Many L"(·onomi:-.b would :LrAlle th:lt even if
phr:--.i(:l l or othcr as.~ets an: l-ouh '>titutes for moncy.
va riation.~ in nomin;ll int~rest r.th:," o n shon-term
IXlIlds brgd)' reflL·t:t vari;lIions in inflalion ;md
long-tl'rm IXlOd r.lh::S so that induding proxiL'S for
the:-.c tcrm:--. would hL' fL"(lun d ant \Vhik it is [me
th;lI inflation and nomin:l l interest r.lIes l el1(J IO
1ll()\'~ t0t-:e(h~r CFigll1't: 2). the ('{)rrebtion is Ell'
from pcrfL'Cl. Thcr(;.{ore. \lsing IhL' only nominal
r.lIe of return on shon-term honds to measuf(;.· lht'
opronllnity ('()....( of holding money dl11.: to inflation may not hc sufficient
Money-demand adjustment behavior, Another
explan:ltion for why inflalion llla y din.:ctly t'ntL'r
empi rical1llodels of money (!L'l11a nd concerns the
L"ostlint'ss of :H.ljllsting cash hal:mcl's, A class of
thcoretical m(xlds known as (111'('11/01'), modf.>!" of
1II01l~J' demaud suggests thaI. in tht: shon run.
individuals 1ll;IY not fully L'Ol'sid~r the effeL"fs o f
inflation In lliL'Se models, ad justing clsh balanL"cs
involvc" L"cn;lin fixC(1 co . . t . . SUL"h :lS [r,mspolt:.!tion
costs. information (.:osts, and tr.lns:lction ('o... ts:
cons..'qllcntly. individlmb estah lish uppcr and
lower (;Irget l)(llll1ds for thdr nominal h:l bnct's
and adju:-.t thl'ir :K'11I;t1 holdings only when thL'
bounds arc viol:IIL"<.1. '111e,'>(;: IX)lmds can lx.'
thought of as lX'int-: M:t in the short 11m, ,'i() th:lt
Economlc Review _ March 199 1

Figure 2

Three-Month Treasury Bill Rate and Growth
in the Consumer Price Index
Treasury bill (p&rcent)

r
"

"
"

"

individuals arc dema nding nominal bahmces and
passively allowing inn;ttion to erodc the value of
their hal:mL"~s in the shon run, In the longer run,
however, individll:t b adjust their targel ho unds 10
rt'fle(.·1 tht: co.~t of innation and (){her del enninants
of the dem;l nd for mon~ y
1lle impoltant point IO nolC here is that if
Ihe inventory-type hehavior is prescnt, then
individuab' demand for money in the shon run is

• ThIs result can be seen by using me approach K1 whICh
moneylS allowed 10 en/ef a repteS6f1tativc IndiVIdual's utility
funcllOl1 (see POflerba 6nd Rorcmoerg r987) If the onty
arguments of /he indlvldual's ullllly func/1OI1 6,e money and
consumptlOfl. rIlen //le mtI'glnel lale of subsr./ulion be!Ween money and consumptlOfl WIll equal rile nomInal rate
of return on a benchm6rk asset rhet serves as a store 01
value and does noI6flter th6 ur.llly funcrl(J(l In //l,s case. the
demand for money is a /(In(;tion 01 consumptlOf'l and the
flCll7l<nal r6tcof return on the benctmarlr asset However."
there IS an alternatIve svt>slllure fat moocy besides con,
SumpllOl1 th8tlS also en ergument oflhe /IldivIduaJ's utili/y
tvocllOl1, /hen ~ dtlmand funcbonS have as thetr 8Igu·
f1ICf1lS IIOl only con5umpllOl1 6fld the rate 01 rerum on the

/.>enCIVJJarlr asset. bul alS() /lie rare 01 relum on the a/lema/We subsllMe for money
~

dlscvsseo on fOOlOOIe 4.
lonQ-tent1 bonds would now be the benchmark 115ser and
shotI-lerms bonds would dlfflCt/y enter the uldsty fUflCbOn

In rCfnl$ 01 the ul/ll/y frtJfl'tC'Nt7k

DeC9L1S6 they PflJVIdo 1IqtNd"y servrces

'9

Figure 3

Transactions Balances
BiIboos 01 dOIlaNi

·.......""'"'"•

~
MIIMC DIpoIil Ac:c:cura
~ ~ Mo.aIIII F.....

• OIW' 0 * * - DI!XIIb

..
NOTE: Transactions balances are M2less savings and small iln18
dEtPQs,rs.

not purdy :1 dt..'I11:II,d for re:l1 bal:mo,..s " And, :IS
tht.' Appendix :-ho \\"s, if individu:lis bil 10 :HTOUIli
fm tht.., fulleO't..'("( o f inl1mion on Iht.' dt.,tn:lI1d for
lIlurK'}' in tht.., :-.l1on 11111 (that is, if individu:.ls do
not exhihit pllrel)' :1 ( klll:md for re:l1 1110nt.'Y
h:d:lI1ce., l, lllt.'n the infl alion r:lI e \\ill be:l ...ep:rratt.'
:md :-.r~n i fk:.tnl detenn in:rlll in empiri c:d m()(kt...
o f the dt.'mand for mo ney.

The impaci of inflation in a deregulaled
financial environmenl
The IWills bro ught a prolifl:r.llion of rnoneylike ass\.:ls sllch :l.~ NOW accounts that, unlike

Another

'cason IndlVld ua!s may be concerned wIth f)()(TII'

nal. 'ather than ,eal money balances IS /hal ban/< fee
schedvles /lfe based on f)()(TIIna! accOUtlt balances

The 8PfXCIP',aIO fICIffllI!al talC 01 rcrvrn

on tn(ltI(.'y '5 USIJ8IIy

COflSfflJClcd as a we<ghted meaSLWe oIlheddfefem nomtnal
re/ums pard on the IfIdIvrdual COfT¥)(JIlf1il/S 01 me monetary
Bggfeg;lle UIldef consldera/1OtI

rhos ass..mes /hal changes If! III/iatlOfl do flO! af/ocl the
rcal rateS 01 rtHUm en varIOUS assets dlfferenlly fhal IS. a

'·percent cllaf"98lf1/he 1fI/ia/1OtI rate fCSvl/s III a r ,percenr
c/laflfJ6lf1l11 f/OtTIII1a/ ralllS 01 relUm

zo

t."lr rrl'nq and demand (kposib. t';l rn a nominal
1:ltt.· o f rt:tu m and ...till pro\'ide t '~l n,.rcti()n :-.crvices
Fi~ lII"t..' j pre:-.cllls [he compo... ition o f till.: ;\12
aRi!lq.:ate, \\ h id l d t':lrly :.ho\\"... thai the intt..'rt..·:-tIX':l ring (·ompont..' nt o f ,\12 h:l:-. hl'l"olUe \'t.'I)'
:-uhslalllial .
\'\ 'hal dOt.'... the int« )(lul"tlon and proliter.ltion
0 1 Ihe:-t.· a...:-t.'t:-. Imply for in natlon ':- dl"t.'"(.'t on the
demand for monl;'Y! To ;rnS\\l'r this qUl;'stion, mos[
t.'l"OnOI11l ... t!> I,.,\"c,.' 1ll(){lifit.-d tht: rdl.'\";rnl nominal
i ntt.'re... t r.lIt.' II1dlldl;'d in the l.'lllprri(:al Ill()(kl :-on
(hal it b now tho.: diffc renct.' I)o.!t\\ cen Ihe aprropri;lle nomi n:d rl;'turn on mo nt: y :rnd Iht' nOlllin;r!
n:lllrn Oil :. shor1 -1I.:I"I)) ho ndo Thi." difference
rqm..'sents I hI.' oppo rtllnity cost of holding mOlley
if holdi ng a short -It..·lm bond is Ihe aitl;'rnativc to
holdill~ money W hat if ;lssctS other lh:rl) shol1 -t..:rm bonds
:1I"t.' serving as sllhstilutes to holding Ill()ney~ T o
thl." l."xtent th:lI (:h:..nges in inllatio n art.' rdlet.kd in
thl." nominal relllnl!> o n both mont..'}' and ass,,:ls
tlr:.t :-t.· ....·e a .~ :-ouh... tillrt e.~ for mo ney. <:hang..:s in
i nll:rtio n sho uld nOI afft..'Ct thc n:I:lli\'t.· rt.·:.1 rt..·turns
lX'tween monev and its ...uh..slitult.·:-o [nlbtion.
tht.'fd'on:. sho uld not afft..-'CI tht.· dCI11:.rnd for
lllo nt.'Y regardle ......... o f whether o r nOl the nl()(k-l
indutle:-. tht.., ".rtt..'S o f return of all :rppropriat..:
mo ney substi tutes ~
111C impliGuions ()f fin:.Jn ciai dl:rt..·gul:nion
for the in\'c nlol)'-t)'Pc behavior o f l11ont..'}' ckmand
:1ft..' also illlponalll !\gain, in completely derq.(lIlalt..'d finandal markets. the nominal returns on
both 11101lt.·y and honds fully n.:nel"t :.. l"han~t..· in
the intbtion r.lIe , and the ch:mgl;.' ([OI;.'S not alter
R·btin:' rl;.';rl rt.:IUfIls between monl;.'y and honds
Basically. i"n:lIion d ocs not "ffl,;'ct ;rdjustment
hdl;rvior hec<luse innation d(X's not altl;.'r TI..'l:rtivt..'
re:rl returns hl..'twl;.'l..'n mon..:y and bonds So. t:rrg..:!
hounds arl." set in Ihe invcl1tolY lllodcl.~ ;IS usual.
but ch:lI1ges in inflation no longcr requirt..· an
:ldjuStIHl'nt in either nominal bal:rnct..'S or Ihl,;'
target hounds bcc:..u:.e tht.., "III,;' of rl."turn pai<.l on
the llomiTl;l1 h:.l;mce!> will comllt:matl,;' for the
changl." in infbtion (:I:' delllon .~tr.I1t.-d in Ihl;'
Appendixl, 1I0wl;'vcr, to Ihe extent th:.11 inflation
en)(le:-. the v;llut.' o f the Cllnr.?llC), component of
Illont')'. the in flmion r:rtt..' rn;.y still be:t ~i~nifiGlnt
ne~;lIi ... e dt.'lennin;mt o f the dl'rn:.rnd fo r money.
In sUl1ullary. sho n -tenn nornin:..1 intcrt'sl
r.rtl..'S on hond s m ar fail 10 fully capture thl;' t'fft..-'CL'i
Foo.,nal R"""n'e BaRk of Dallas

of inlbtiun on till.' d . ... I11;md lor 11101l ...:}' :l11d may
tllu.... 1,.·( llll1il,.·t \\itll til ..... \ iI,.'\\ ()f tr.ldili(lIlallIHI11I.'Ylkl11,md . . . tlldi ...·.... TIll' r...·;t-.(ub fOI Ihi." l':1ilul'l.'
indud ...• 1111..' ol11i.... sion 01 r.tlo..· . . of n:turn on a,·;,,:I ..
(oth...·r IIl:tn .... hOI1 -1...'rm hO!l(t....) Ihat 1Il:1} ....1...1'\..1... as
Mlh~tlt\ll ...•.... lilr mlml'Y 1Hlltlin).::. alld thl'

pl"l..·. . I..·1\Cl·

of Ilxl.."{1 \,."IN .... in .ldlll . . lillJ-t cl ....h habnn· ..... Om.:
\\ (lu ld Ilunk. :t pno!'l. I h:1l thl' in;lhdit~· 01 a
nomi n;11 r:1\ .... of n:lurn Ol) .... !loll-lI,.·nH huntt.... 10
Gtplllr...· l ull} th ...· df...·I,.·h of inlblioll i .... mo.. . t likdy
to h.t\l·II(.·I,.·wTI.."{1 dll1'in).: 11K' 19..0 ..... \\Ill'n high
inl1;1lion 1~1I ...·:- 111ad...· :h:-...·l..... n1iK'r Ih:Hl bonds likdy
:-ul).. . tit1.I1 ...·.... lllr t1llll1l'Y In Ihl.' 1):I:.t d l'cld ..... 1H)\\ ",·\·",·r. Ih ...· l·m ...·q.\...·I1l'1.' of inll'l'l....... I-lx·:lr;n,l.( nlOl1..:}
:111(1111"" r...·l;! li\l·ll)\\ Inll:tlilltl. \\llidllll:ld ...· 111:lny

k'... :. a!lr;Kli\l' nUlI""'Y Mlil.. .lillll..:...... Sll,l.(,I.(..:sl
th:11 Ih",· ...·11;.:1.."[:- of intl:tlion on IlHlIlo.,.·} d""I11:lI1d

:I:-:-l'I ....

lIlay h:t\l' Ill't'um ...· !L· . . s Importanl during th ...· 19XOs
th:tn for Ilw 1........ 1 01 th ...· P0:-I\\:lr p ...·riod Th ...·
an;tly . . . I..... tlll·r..:ftll ...... lu1'l1 .... tll:lll ...·m]1;l'i ...·al l';';;llllill:I'
!Ion of 1111 .... 1......... \11.'

Some cmpi.rical results
In t hl .... :'l.'t·lion . I ...·":l1l1inl· \\ hl·lh...·1' 11ldmling
Ihl..· 1Il11u;on r.tl ...·. in additIon to :t j)ol1llllal inllT...·:.1
r.lll'. add:. l':-..plan:llory po\\ .... r to mOIl ...·~-d ...·1l1:tnd
IUIKId,.....
TIK' l'llljllrtt;11 \\01'1.: nm ... i ...h of ...·...lim;lling
mOlll'y-dl'Ill,lnd r...·grl·,.... io!l. . . of tilt: fonn

\\ I1l·t....' Cb:l ...·011 . . 1:1111. III I .~ lilt: I'v:11 mon.... y .~IOl'k.
is 1'L':tl (,,\,1'. f( i . . lilt: IIl1'l'l"IllOlllh Trl..':I:-ury hill
r:1l",'. IOf i.. . Ill ...· O\\1l I~tll' (If t'l'llInl 011 IllOIlL'y. Jt is
Illl' inll.llillil 1"1,,,' (lll'sl dilli.·t'",·nCt· III' 111<': l()~ (II IhL'
priel,.· k·\ l·ll. and IJ I . . an t'rror Il'fIll Th..: dala :11-",'
quant·dr. ;Ind tilt: 0\\ 11 r:111..· of I'l'llIrn on 11l()11L'~· i:.
ITlllll tilt: Ik,;lld I,f (;11\ l·I·I\I,I':. (If III",' F.... tkr:~1
Ik........ rh· !'>\:.ll·1ll Allo\II ...·r <.1:11:1 af.... fTlltll Ciliha:.",·.
\ltlll\,,'\ ·tkluand Illt Kk·l . . l':-I illl:II ..."{1 \\ lth

y

<tU:trll' r1~ d,II.1 Iml-.\ iIKorpor:II ...· thl' app;lr...·lll

b)ZI.!I..'d n· . . potl ....l' III' th ...· d ...·Ill.md lill" mOll..:} to
... h;I!l~"""" III tlw lll-ll·l·lllin.ltlh "f th ...· d ...·Ill;tnd for
Ill()nl..·~ In tlK" P;1-.1. 1110:.1 mOll ...·) -til'm:md mlxkb
I:con o miC Rt" 'icw- ,\ I arc h 199 1

:Kcompli.. . h\",d Illi:. hy 1I .~ing r..:gr ....:.slotl... of Iltl..·
litrln dl·."lTlbl"{1 hy L'quation A" in Ilil' Appendi;.;.
in \\ hicll onlr tlK" (!L-p\:mk'nt \·;tr;:lhk 1..·!l1l·fS ;tS :t
bAAl'd \.';.;p1:tn;lIory \·ar;:II1I..:. Ikn'nll)" hO\\·e\·(·r.
11ll' rl.' ....lr;l·t;\'I.· n;lIlIn.: of Ihi.... "'"{Iu:llion. in Il.'rms of
Ihe bAAI..'<.l re . . . I>0n........•. . :Illo\\(·d. h:t~ come umkr
('filid~Ill :' In r..::.porl~l·. mOT": g..:n":~~11 modds..~uch
:I~ ""qu:lI;on J. 11;1\''': h...'I.'Ollll· pn..·\·;tk'nI h ....'(~llISI..·
th ..:y all .. l\\ :t mot'...• 11I..-;.;ihk rc . . . IXltlSl' pallt;'m 10
:Iri .......· from dl:mg...·:. in Ih .... d..:t .... nnin:ltlh of mon ...·)'
d ...·ll1;lnd ,.,

til\.'

Ikenl .. . ",· Ill..: .lbm·1.' :1I1:tly:.b .~lIAA""'~t..... Ihal
of i nll:tt ion Illay h ..:nlllie ks ... ;mIX)rtanl

drl.."{·\.~

a:- mon..:} bl..·u)I11I.'.... mlWl' int":I'\"'.~I-lx·arillg. 1ll0n ...'Yd ...·mand 1l10{kl.~ :11'<..' ..::-lim:III.."{1 ti)!' I Ill' :lggrL'g:tll.'. . .
il.I IA. 1\'11. ;Ind 1\12 T:thk I (ks(Tilll's Ill .......",· :lggl''':'
g:ttl'S As Ihl' ;lggt'I.:).::Il ...·s Ix,com .... hI'O;Kkr. m()ving
from 1\1 lA 10 1\12. a blgl.'r proportion of I hI.'
aggr .... gall' i.. . int..:l'l·... t·h....aring Addilionally. lilt·
:,:ullpk pl'l'i(xl umkr ...·on .... kkralion. ll)'ij-W), is
tli\'Kkd inlo 1\\0 :.uilpnilxls. Tlli .... ....I .... p i... donI.' for
1\\0 "l'a"'JIl~ I'ir~t. di;lgIlO:.lil· :.1;lIbtit.·s. sucll :IS
11.::.1 .... for . . t:thilily of tltl.' 1.."{lll:llion .... · (,·I ICIl"il·i .... nt:..
uldic;lIl' Ih:1I :I ~lIlglo: t1l1Kkl for th..: ...·ntir.... s:tnlpk
perilKI is nol :Ippfopnatl.: for 1\1 IA . MI. or il.12 "
Scnllld. hl't.~tll:''''· Illl' r.lngl.' of inll·rl':-.I-Ix:aring
:t:':-l'b h":gall to I..·"p:tnd gre:tlly in 19HO. IWII
inll..·1'l· ....lin}.t <]u...•...lion .... :lrl' \\'1ll'IIlL'f Ih..: dkt.·I:. of
inll;tlilJ!l diff...·r :I('R)~:' IllIJllt'l:tl)' :lJ.Q.:TI..·g:ttl·'" :tnd
\\ 1I1..·,ItI..·r thl'}" d ill".... r :to'O.....:. lil11": ]'x:rilxt:.. Til ....
rl·grl.' ....Mlln:- :tl'l' :.rl(.:dlkd In hoth log-I..:\·..:! :lnd
fil,:,Ol-dilr...·I'l·IK..: r(Jrlll lx·ctu . . ..: Ill' spuriml:' l'orr..... I:Ilion.... Ihal m;IY ;lrisl..· \\ 11":11 \ ':Iri:thks :Ir..: slrongly

. For onstance Ihe cqmlllon reQuires thai real money oal.
/lnces adJUst /0 cllangcs

In

caell 01 tile money· demand

(l()1()fmmllnlS ....111 rIle same gcomcmc lag

Til,S /rcndClOOS nollfTlPfy tMr models of the IOfm of CQual/fXl
1 overcome 0I118f orOO/Cms such

as Idcrm/;ca r!()(l IssueS.

IIlIle/enl III eS/1fTI1l11fl!) moncy-dcmand equal!()(lS See

Roley (1965) lor a C1tSCUSSIOfl 01 some oIlhesc ptoblems
ACkUlOIlIJ1/y IfItemalellwoacheslocquatOOfl 1 exlS/. such

as error COf/CC/1QIl models ac/vocalcd I:1y Henaty (1919)
TheSe IfICIuOO Chow ICSlS antJ 8f0Wl'l. Or.Kbtn and £\I3fl S
CUSUMSO leSI s181,SJICS

lor

paramel()f SJilb<h1y TeslS lot

senaicorrt!lallQll. heleroscedaSJlCl1y and ARCH reSIduals
WIJPO(IOO /fle lmal

model SPOClllCiltlOflS

"

Table 1

Components and Definttions of Money
Bank reserves
Currency'
M1A

Travelers checks 01 nonbank issuers

Ml

Demand deposits2
Other checkable deposits

(CX:;Ds)~

M2

Money market deposit accounts (MMDAs)
Money market mutual fund (MMMF) balances'
Savings and small-<lenomination time deposits'
Overnight repurchase agreemenls and overnight Eurodollars'

, Currerocy 0UISida the Treasuty , Fedlfal AHeM! Ba.nks, and the vaulls of depository IfIStrtutions.

• Oemand deposrts at aI corrvneIdal batiks othef than those due!(l depository insbtul.ons. the U.S. government, and fOfelOn banks
and otfioaIln$litu!ions less cash itt<'nS N'1 the process ot coIectIon and Federal F\eserIte lbat.
> Consist 01 n&gobable order ot Wlthdtawal (NOW) and aUIOma\1e lI'anster servlc:e (ATS) accounts at depository inslltutoons, credit
U!'IIOI'1 share dra" accounts, and demand depoSits at Ihrifl lnstilutions.

• Time deposits, including retail 'epl.ln::hase agr~nts tRPs).

N'1

amounts less Itlan 5100,000.

• 0vern0Qht (and c:ontinulIIg conlr8Ct) repurchase agreements ISsued by a. oomtTl8fcial banks and overnight

Eur~ IuufId 10

U .S. rvsderIts by foreign bfanchvs of U.S. banks WOfIdwida.

NOTE: M2 excludes individual retirement accounts (IRAs) and Keogh balances held at deposOOry nstitubons and money market
funds and aI balances held by money mal1tet funds (e~cepl insbtullOfl'OI'I/y 1utIds), U.S and foreign c:ommet'cral bilrIq, and
the U.S. and foreign gove«menlS.

"

tn.:nded I! The diagnostic:; indicatt:d that suitahle
MIA and M I m(xI~1:; wuld be specified by
hre:Lking the s:.11llple period:u 1979:.3 for both Ihe
k'\'d and first -d i ff~rence :.rx-cifications Ii For M2.
par..uneter stahility t"ould not he obtained withOlJl
dr()ppin~ the 19,3-,H pcri(KI Once those years
wen.: dropJX."'<I. howen:r. the test statistics indi<:~Ih..--d weil-beh:l\"Cd Illodeb for the 1959: 1- 1979:.3
:lnd the 1979:4- 19H9:4 period:. I'
One Ia:;t issue I:; the :Lppropriate lag icn!-,oth
to lL.~e for the right-h:md-.;;ide va riables in C<.juation L After some l;:xperirnent;lIion. [ followed
Goklfdd and Siebel (19H7) and lietzel and Mdml
( 1989) hy u~ing fo ur lags o f each v:lri:lble. 11
A:; Goldfekl and Sichel (1987) point OUL
equation I C:1n be u:;ed to di:'lingu ish lx-'\ween
ShOl1- and long·run ellens of inll:llion on money
tk-rn:lnd. Testing d! - 0 for any j::iven i amounts to
testing the hypothesis th:lt realllloney balances
adjust without :1 I:Ig to ch'lnges in infl ation, or in
other words, tll:!t the!\: an.: no shon-run effe«s of
inflation on the dell\and for money. I" A second
hypothesis is that infl'ltion m:.y have a short-nm.
but not a long-nm, efft."t.l on n.:al halano..""S . In teon"
of I;.'qu.ltion I. this hypothesis is

It.'nt with the hypothe:.i .~ that inflation effects are
import:lnt for money dem:lOd and even more
imporlant fo r those m o n~tary aggreg-.ltes th:ll have
sm;lll interest-bc:Lring componenlS.
The analysiS :llxwc :LIso :.uggested that the
(:"mergence of intcresl-1X::Lring money should
diminish the effect o f infl:Llion on the demand for

OZ

Thls finding occurs when the '8Dres$(Xs are not slatoona/)'
OIfferenclf/(J the vsnalJles usually e/lffIII'lat6S any nonsla-

llOfIllfiry III tll6 dala In /116 analysis herc. estllllal!Ofl and
dIagnostic tcsrs are done lor oeth rile level lind first ..
d,l/erCflCo Spccmcaroons ThIS IS done as a lest 01 rObust..
ness of t/lolcsulrslor lt1ee/lllCtsol,nflallOfllfimoney·demand
myresslOlls
IJ DlagnosllC SlatlstlCs wcrcr;cneralcdwhcri 1/10 whole sample

was IJsed in the eSllmation and lor varIOUs subsampleS
Th6se slallSIICS tndicated that a well-t)chaved model could

be spcciflCd /or Ih8 1.953. 1 1979 3 period and lor the
1979.4 1989 4 perJOCi ArlemtJlmfJ 10 lengthen the sample
periods /or Clthcr rrodeI resutlOO If! pal8I1I81er Ins/abillry. as
IIldlCSrcd by the CUSUMSO tests AddillOf)al/y. /or all tile
IIggregare!l. thetJotnoQcnelryofrne demand for tnOne)' with
respf1CllO the pnce level. cmbcddcd In the level spect(/C81lCn$. was tcsted BOd IICceplcd

.. With M2. hOwever. Itle SlatistlCS onchcated /lnll a stable
rrodeI could t:le SDecilled beyond the 1979 3 bleak U$JIlQ
etth8t the IlIs/-{;ilffercr1C8 Of Jellfl/ spccificatoon Ht;mever.

Tahles 2 :md 3 present the Muns of the
distrihuted \:L~ cocm("ielll.~ ;lnd the I statistics for
Ihe test that the Slun of the d;s ~qu;ll zero, llsing
Ihe log-level :md fiN-diffl;"rl;"nce specificnions 1Both the first-differt.' lKt: ;lnd log· level specifica tions yield the s'l1nl;.' (ondusions. First , the inflation
rate cntcrs with a neg:lIivl' sWtistically significant
sign for MIA :Ind MI during the 1953-79 period. l~
The coefficient on tht: intlation term Is more
negntivc than tlwt on thl! interest-rate term and
has a larger I st;tlistic In tenns of explanatory
power. for both MIA and M I, including the
inflation r.L1C improvcs the explan:nory power of
the first-dilTcrence model hy 35 to 40 perc~nt. ''/
For M2. innation imposes no statistically signifi ..
canl direct shon - or I()n~ .. rlln effects on the
dem;md for mone)'. '111C intcrest .. rate effects,
ho,,"c\'er, are important. The resullS indiC'.ltc that
the inflation rate should he dif't."Ctly includl;.-d in
MIA and M1 monc}'-dcm:ltld models for the
1953--79 period. Thl.!rcforc, the resullS are consisEcon o mic Rcvlt.-w- M an:h 199 1

parameter stlllJi/!ly could not be oolalfl8d for !he efJt!re
1959-89 period
CholeiltO 8S/lI1I8tC a different ~

me

afrer 1979 3 was made on the baSIS aI the c/1ange in
oporatmg ploclldures by /116 Federal Reserve at //la /lime
and because stablliry could not be oblSmed unll/ 1982. the
end of the period lor the parhculllr operallng ptoccdures
" lags lon9er than lour were sta/lstlcally insl{Jnificanr and did

not add explanatory power 10 thc rQf}resslons
Of

For th61lfst-dd'erence speclflca/lOIl. tile hypothesis IS /hal
the first difference of tnlllllion does not a/leel !he first
d,/lerence

01 '681 balances

" If thel statlst,c is greater Itlan 2. then the sumo( coeffic/8llts
1$ dccmcd 10 b4! slatlstlcally SlgrullCant Ihis les/IS for the
less restrlCll" hypo/tlcsls tna/lnflatoon does have Iong-rvn
effllCts on Ih6 dem8na fOf mot'Iey

.. Very lew cNfcrences 8fJ(J881 In the fes..Dts between MI and
MIA 101 the 19S3-79f)11nOd becaus801tlC1 checkable dePOSits IS a very smaR component through 1979
Of

11lB adJUSled Ws Ifl(;rCIISC 10 0 59 and 0 56 from 0 43 and
040 /or MIA and MI. respectMJiy

"

Table 2

First-Difference Specification of Money-Demand Regression
MIA

ra
ro.

lb,

rd,

1953:1- 1979:3

1979:4-19894

.928
.093
- .0057
-. 021

1.05
-. 130
- ,001
- .075

(7.86)
(1.53)
(2.85)
(3.50)

R'

(5.00)
(0.63)
(0.14)
(2.00)
.7.

.59

Ml

1953:1- 1979:3

ra,

.902
.095
-.004
- .022

lb,
I. c,

rd,

(7.27)
(1 .53)
(1.61)
(3.67)

R'

1979:4- 1989:4
.518
.360
-.013
- .003

(2.41)
(1.30)
(1.94)

(0,10)
.70

.56

M2

1959:1-1979:3

ra,
ro.

.02\)

1.23
- .058
- 034

lb.

rd,

(0. 10)
(2.03)
(2.42)
(0.69)

Fi '

1979:4-1989:4
.389

(1.30)

.450

(1.40)
(0 .78)
(2.00)

- .007
- .035

18

56

NOTE: t stallSIICS we in parentheses

1l10l1l'V T hi:-. I~ till' l·:I.~C lX'C llI:-'C. t() thl' L'x tl:nt that

1)llil11 o.: ..l illl:\I,,:"I>ftllo.: lnlbl;()lll·lkl·t.... rail in till'

thl> p:1I1il"ubr lllunl·t;lry : 1~).:rq~: ltl· i,.. intL'i"l""I-

lir~l-di lrL'rl:l1lL' ~pLTilk' lt ion hu t il1nl"; I~L' III till"
II J,I.!- k\ d ~pl"l'lnl': !I;<)I1 h)f 1ho.: ,\ 11 . \ ~IX·l·ilk:l1i()l1.

hL'; lrin~ . l" h;lngL'~ 111

;ntl;niul1 \\ ilJ III >t ;lffL-cl tht:

op pnrtl1ni[~ lO:-.1 oj' ho\tllll).: 111()IW~

1l1lkL'lL II1\.'

L'ltlp lr ictll'l"~lI h ~ IllllicltL' tl1:l 1 till' dlrcll

dkcb of

11l1l;ilion. :L~ 11ll':I"lIrl'd h~ till: l()l:ffil'il' nh on tilL'
inll;l l!on ! enll~. 'I!'L' no lon).:l:1' .... t:Ll i .... t; ctlly ,..;).:nifi
CUll for :\ 11 in

till' l 'rlJ-l-.'<) rx.:rind Th.: long run

lh l' in!bli()11 uldl'ICll'nh ;lI'L' Illl\\ ()11 1 ~ l1l:trgin:tll~
,ignificln1. .dl!tOllgli 11ll' point l' .. til11:lll·'" l1;1n:
imrl':I,.. ed Tltl' inlll1ion l·( ...·f!llll·nh for Ih\: ,\ 12
~lX·t' i nG l1 ion .... ,II'\' no\\' I1I;Ir!-(in;dl~ .. ignitk:tlli.

\\!tik'lho.: 1)()ll1ll'~tlllt:lIL'~ arlO \lrtU:l II ~, tIK' ~:Ull\" III
tern", of ~horl run l.ffL·l"l'.
\)ll'ur~ IX'\\\l'l'l)

:t

OIlL' 10 onL' n)rI't'I:l1ioll

till' IHL"l'llll' 01 dl!'''·''! long-run

:uld ..hon-nm ..:fln!' In ot lwr \\ord,. \d1\."n lh~fl'

Less

~,""II>O" ,1'1

Sl3tlOil,d .. "<)II'

039""'" 0

24

",f/alOOf!

(X;c;ur

mar

bg 1110' U':JSCfI ~

1arrJC'f

The ~.l"'lnrc 01 wlflJl1(JI) dechncs 10

49 1//>1' fhlr'~ ~',1lI

\\ L'rl" no long rUIl l·fI ..:Ch. !IWfL' ,Iho \\ l'rL' no
.... hoI1-run dl...·ch Tf))..!L'IIll'f, tllL' .......· fl·,I1I! ... ;t1l;;"
('flrhi~! 0.:1l1 ..\ illt till' h\ pOlho.::-., .... Ih,l1 IIln:ujOll h:1.'

I ""'\(>I\\L" a

k" ill1l)('.';ll1llk"IL'nlun.ll11

".."tI.,,..,.. t R.,,,,,,..'"

( )f

\11 \\i th

tSa nk of Da Uas

Table 3

Log-Level SpecHlcatlon of Money-Demand Regression
MIA
1953:1-1979:3

Ea,
H ,
E c,
Ed,

1.046

(61 .53)
(5.07)
(2.59)
(5.39)

.039
-.002
-.014

ii '

1979:4-1989:4
1.146 (10.05)

- .135
.003
- .059

(2.11 )
(0.81 )
(1.90)

.98

.99

Ml

Ea,
H ,
Ec,
Ed,

1953:1-1979:3

1979:4-1989:4

1.020
.032
-.002
-.011

.680
.246
- .013
- .036

(60.10)
(4.05)
(1.62)
(4.15)

R'

.99

(5.62)
(2.05)
(2.20)
(1.56)
.99

M2
1959:1-1979:3

Ea,
H
Ec,
Ed,

(19.27)
(1.45)
(2.87)
(0.71)

906

1."
-.066
-.015

ii '

.98

1979:4-1989:4
.934
- .012
- .007
-.018

(9.83)
(0.11 )
(0.73)
(2.22)
.99

NOTE: I $lallS\o(::$ ale In parenlheses

th ...' ...·m ...·rg...·nn· ill inl ...·r...·,.,1 Il.... aflng 111(lI ..... ~

\X'11~

inl1;!lion h:l.' IWnJTll ...· :1 III0r..., ,ignifinl1l (kllTllli11;11)[llf

[\12 r ...·lll;lln':1 1)U/1-1o...' .

IH'\\ ...·'",")"

.\ddiliun;lll~ . th ...· gl'Ill"!~1i r .... ,.,ulh of Ih .... i"irstditl...'r...·nt ...· !llod .... b ;l rl' 10.. .,." th.!!l 'ati,Etclorv
Although tlw ...·"pLIl1;Hor~ pI ," .... r. :1' m ...·a:-.lITnl hy
1<-"-.. illl"r"".I'"", from Ih ...· ...·arlil·r :-':lIllplo...-. \ .... ~ ,.....,' of
11ll" ...·"pbn;lI(lr~ \.lri,lhlo...-, olh ...·r Ih;1Il bgg...·d rl';l1
h:lblll"l" :11..... 'I:UI,licll ly ,iglllficll11
TIll" UllU 111'101) I dr:I" frol11 Illl" l 'mplrictl
:1I)'I I ~ .. i' IlIl·n. I' Ih,11 lor til<..' 1I)'l.~ -I)
dlll'lll~ tndudillg

p.:riod.

till" inll:ilioll r:1I ..., in empirical
11)(1(10...·1:-. III !lMIIll"} tlo...·lll:tnd ,Ilk" ,uh:-.I,lmi:lll·"pbn:lI ()r~ I)()\' ...·rlor \II.md .\11A Forthl · ItF9-.,",'<)
IXTIIM..I. thl' l" idl·Ill'...• I' mi" ...,<1 :I' tCI \\ hl'[hn

inll:tlion h;l .~ b ...·("olm: k:-.s import:lnt for mon ... y
d ...·mand and \\ hl:lhl:r inll:i(ion 1~11 .... t ...·nll.' ., hould
ho...· dir...·clly indlllk'd in lllont.:y·d",'Ill:lnd model ....

Conc lus io ns
Thi' :trud ...• :Iddr...·.,:-..... d (It.... i:-..~\I ...• of how
inlbwm ,hOlI\d hl'., ( ht.: in(:()rpllr:l1l'd into mont.:}"lit-manti IIII)(k·h AIIIIII\I).:h r...·ct.:l1l ly ~(II1l ...· ...·{"()IlI ...
mhl;'o h.IH· dirn·t l} inn)rpor.ul:d tllo...' intblioo r.l1l:
into IIWlr mOlk'h of Illont') lK:mand . 11)(I,t ,tudi...·...
,1"11111...• 111.11 ,I 'holl-I ...·nn Ilolllmal i tlll·n..·,( r.llt.: i,
'uffiu.... lll 10 captur.... [h ...· dkch o f inthlion In Ihis
:ll1idl". 'I\.·,il·"t.:d till" Ir:ldllilm:l t argum....n t fllr
indudin)-: ()nl~ :1 ,hol1-ll'rl11 iOll'H.·,1 r:11l' and

"

Empirical Implications of Alternative Money Substitutes
The analysis in the main text indicates
that if assets other than short-term bonds
serve as substitutes lor money, then their
rates of return should be included in moneydemand specifications . If these rates of return
are excluded, then inflation may be a significant determinant of money demand , in addition to a nominal rate of return on short-term
bonds.
The analysis suggests , then , that moneydemand models should incorporate rates of
return on all assets that may be serving as
money substitutes. Including such rales of
return would then render the inflation terms
insignificant. To examine this hypothesis,
equation 1 in the main text is modified to
include contemporaneous and lagged values
of long-term Aaa corporate bond rates under
the assumption that long-term bonds or capi-

pft..' ... t.:ntl'{I I\\" rc; ....on ... mllu.on m,.y ;.I-.()

dIH'('/~)'

enter llltKh:J ... ,,( thl' tk'lII,md f"r ll1()n~y TIll' ........

rc:."'()Il.s .... t.'t11 fnUl. t\\(1 (:Io:,:lor ...-Illt.., (,'():-llinl.'''':- (If
(:ontinl.ou ... lr ;Idiu~t i ng moncy habncc.s and Iht.'
exblL'rKl' (.1' :.lIl·I'I1;I\.\ '-' "'lIh:-l iUUl· ... for 1l1011L' Y I
al.'>!) a rg ul..'d Ill:.!. lhco!'l·tictllr. lilt' ;ntrodu('lior1 of
inll're ... l-hl..'al'l ng .nOl'ey .ll:.}' h: ~''''l·n Ih(' ne('d 10
indud(' Ihe ;nl1.lI .o., r: IlL' dir'l:'L'lly ;nlo moneyck'lll;md nHxlvl ...
Enlpirk:.i l'\'ilkon n : indic .tL'cl tha t ... t:l!j~lic:dl \'
s;!-lniti C:ln l .• ll').,::lt;\(' i()1\).,:-l'l. n dll'ct.~ ()f inlbli(lll
on Ihl' cic!ll:m d for r1101K'~ o(:t llrred <illrin)..:l9') j---r)
;md th:!! Ihe"'l' cO ...·u ... Wen: ,'01 fully captlll'l.'d hy
simply a ... !torl lu m imel'\,''''1 r:lIl·. Till' l'e"'ll lh fill'
Iht: 19}{(j.., \\'L'I'L' 1l1iXL'd :1'" 10 \\,hl..·l lIer (krq..:ubt iol1
ha.s led to in[1I1I011 l><-"il11-( k ...., of .111 L'xplan;t lory

factor in tltL' (IL'mantl for monq
I h;l\l' omdudL'd . IIK'n. Ih;1I dunng t ill'

rxN\\ ar p'-'ri()(1 unlil Illl' ~'I rl y 1900:-. \\ Itl..·n .110rll·)
did not l';1fIl inll·l'l· ... I. intbw)!l had an impO!l;u\l .
(hrL'{·t dTI..·(:l on tIll' dem.lnd for lllonL'Y-;Ul l..·ffl..'CI
dl'oWWI [ro lll .1 ... inllul'rlcl' Ihrou).,:h .sho n -It·nli
;ntL'rc...ll'all..·... By L',('\udmg Ih l • din:('1 dh'lh o(

'6

tal are serving as substitutes for money , If inflation is significant because these rates of
return were previously omitted from the moneydemand specifications, we would expect to
find that , once they are included , inflation
would be statistically insignificant. In fact.
when the Aaa corporate bond rates are included in equation 1, inflation remains significant in every specification (for both samples
and in the log-level and first-difference specifications) . Furthermore. the coefficients onthe
other included variables, including inflation,
are unchanged. In addition, the Aaa corporate bond rates are insignificant for atl the
specifications . These results suggest that
inflation enters as a significant determinant
of money demand for reasons other than the
omission of rates of return on long-term bonds.

mlb'''111

f"l.l1 ll le
19-'). ('X-t)llotlll ...I... · lr:.lditl0n;l[
lllotley-tlL·m.Uld 1ll()(ld . . d.d nol pI:donn .h w{'11 a ...
11ll')' ('ollid h:I\": . \hhough "'Ollll' Ilworl·lIc;.11
cl11lll\..' d..:Ill;lnd It,r 1ll11I1l'\

p<. I"'I\\<lr lX'nexl

until

iu ... t if'ic.tI;lln for includi ng lilt, r.lll· o f i l)l1alion

m( mLT-dl'Ill:lIld IlJ(Kk-t...

!1l

Ilw \\ iciL''oprL'ad
II ........ o( ir\ll'rl'sl -hearitlg mOlll'} :1Ilt! Ih L, pre"'l'1lI1y
.lJ(Kk-l~ltl· 1~ 1\ ...· of inlbllo)) ... u1-(ge ... t tiLl! tl lI'L'ct

dYt'l'h 0 1 inl1;1l1on Illay
Lint m t)l\:

p:t ... t dl:.'C. ci<-,

.'\.·.II;1ill ....

h:I\'L'

IX:O'Illt..' k'~", i lllpo r-

Appendix
Adjustment Costs and InflaUon In Money-Demand Models
ThiS Appendix shows how inflation may
enter empirical models of money demand if
the short-run demand for money does not
completely enforce zero erosion of money
due to inflation. The analysis largely borrows
from Goldfeld and Sichel (1987) and Hwang
(1985) . A divergence between short- and
long-run demands lor money typically is motivated by cost-minimizing behavior, where
the costs of disequilibrium are balanced
against adjustment costs. The adjustment
formulation is usually of the form

equation A.2 into an equation suitable for
estimation requires a specification for m; or
M;. Before doing this, however, it is worthwhile to elaborate on the analysis of Hwang
(1985), who shows that specifying partial
adjustment in completely nominal or real
terms is nol necessary. Hwang specifies a
quadratic cost function of the form
(A.3)

C - a ,[ln

M; - In M,J2

+ ~[(In M, - In M,...,)

+o(ln P, -In P/-1 )]2.
(A.l )

Inm, -Inmr-, ""A(ln m;-lnm/-1)'

where m, is the actual stock of real money
balances and m; is the long-run desired stock.
Equation A.1 is known as the real partialadjustment model (RPAM) because itenforces
zero erosion 01 money balances due to inflation . The nominal partial-adjustment model
(NPAM), which enforces complete erosion of
money balances due to inflation , is given by
(A.2)

In M, - In M""j : A(ln M; - In M/-1)'

where M,'" m' Pf M;: m~ P,and P,istheprice
level . The difference between equation A. 1
and equation A.2 can be seen by rewriting
equation A.l as
Illn M, : A(ln m; -In m/-1) + Illn

Ma rc h 19') I

°

In M, -In M,.., '" A(ln M; - In M,.., )
+ "((In PI- In P/-1 )'
where A 3 0.,1(0. , + ~) , and y : Oa/(a\ +~) '"
0(1 - ').
Given astandardspecificationfor

m;, say

~.

Hence, the RPAM implies that an instantaneous adjustment to changes in the price
level does occur. Converting equation A.1 or

F.cono mic Nc vic w -

The first and second terms of equation A.3
correspond to the disequilibrium and adjustment costs, respectively. For 0",0, minimizing
equation A.3 yields the NPAM, while for 0:1 ,
minimizing equatKln A.3 yields the RPAM.
Values of 0 between
and 1 yield hybrid
models. In general, minimizing equation A.3
with respect to M, yields

where y\ is a transactions variable, R, is a
nominal interest rate, and 1t. is the rate of in(Continued on the next page)

n

Adjustment Costs and Inflation In Money-Demand Models-Continued
flation . Combining equation A.3 and equation
A.4 implies an estimating equation of the form
(A.S)

In m, = ).90 + A6,ln Y, + ).621n R,

+ (1 - A)ln m"" + pin (P,tP,) ,
where p= H1 -A) + A63 + S(l -A)] . Two points
need to be made from this analysis. First. as
long as Sdoes not equal 0, then inflation [In(p,1
P,.)) will enter as a significant variable even if
inflation is not a determinant of
The sec~
and point is that estimation of equation A.S will
provide estimates of A and p, but these esti~
mates are consistent with an infinite set of
underlying valuesof 63 and S. This result means
that there is a problem in identifying the source
of the inflation effects. One cannot say whether
inflation enters the model because of adjust~
ment behavior or because inflation is a determinant of the real desired long-run level of real
balances.
Th> emergeroe of interest-bearing mooey
has important implications for inventory~type

m;.

money-demand models and the real and nominal partial-adjustment mechanisms that
under1ie these models. The idea under1ying
inventory-type money-demand models is that
in the short run individuals monitor their nominal balances and will allow inflation to erode the
real value of their money holdings. In the long
run, however, individuals take into account the
devaluing effects of inflation on their money
holdings. With the emergence of interestbearing money, inflation no longer erodes the
value of money. As long as the nominal rate of
return on money fully reflects the effects of
inflation, an increase in the inflation rate will
not increase the opportunity cost of holding
money. Therefore, in terms of inventory moneydemand models , an increase in inflation does
not entail individuals adjusting target bounds
or long-run desired money balances. In fact,
the distinction between the real and nominal
partial-adjustment mechanisms no longer
exists because inflation no longer affects the
value of money.

I'cdcr:'" Mcscn'" Bank of OaU:l$

References
(bhl!. Yo.. . hihi.....l, David F J!endry, and Ross i'vl
Starr (988). ~ U.S. M oney Ot:mand, 1 %O- 1 9R4. ~
Discussion P:lpcr no. 88-8. J.jnivcr..ity of
Cllifomi:l. $;.1Il Diego (Ix'Cember)

!:kIr..ky, RoIx:n U. 0987>. ~Tb(;' Fisher lIyp()tht.'Sis
and the rort..'<.':ISI.:lbility of Innlltion, ~ jOllnwl oj
MOllelal)' EcOllOmics 19 (3-24).

The Tr.msaclions Ot:mand for Money, ~ in
ECOIlOmic MOliellillg. ed. P OmlCrocI (London:
Ileinclll:lIln Press). 217~2
Hester, Donald D (1981), -Innovations and Monclary Control,- Brookings Papers on £COl/omic
AClilriO', no. I : 141-89.

lIetzd, L Hoben. and Yash P. MehTa (989), "The
Phillip (1956). ~Thc Monetary Dynamic.... of
HYI'X'rin!larion: in SII/dies illihe QUlIIIW)I
7bl.'OIY ojMOlley, (..'(1. Milton Friedman
(Chicago: Universiry of Chit:ago Press), 25--117.

c'1g-1Il.

Behavior of Monc), Demand in the 198Qs,ft
jOl/mal of Mom:y, Credit, and &mkil/g 2 1
(Novcmlx:r): 455-63.
J-IW:ln~,

Cox, W. Michael, and IIarvey HoscnblurIl <I9H9)'
"~'I one}' and Inflation in II Deregulated Environment: An Q\'crvicw," Fcderal HeSL''''c B:lIlk of
D:lllas /Xollomic Review, ~'ia y, 1- 19
Frenkel, j acoh A. (1977). "'1"',: rom'art! E.xch:Hl,!.(e
Rate, Exp<.'Clations :lIld thl.! Dcm:md for Monq':
lh: Genn:m JI)'perinfbtion: Amerit'flJl
Ecollomic Re/'ie/l' 67 (Scptemhcr): 653-70.
Frit.'-'dm..'ln, ~lihon (956), "111C Quantity TIle<II)' of
!\'I oney-A Ik'Statement,~ in Studies ill fix'
Qllallli~I' J7x"OryojMolley. t.-'cl Milton Frit..-'dman
{Cllicago: University of Chicago PR'S,,.>. 3-21.
Gavin, William T , :Ind William G. Dewald (19H9).
"'I11C EfTt..X't of Disinf1:lIionary Policies on
Monetary Velocir)', ~ Cmo JVI/l'llal9 (Spring!
Summer): 149-63
Goldfdt!, Stl.!phcn M., and Daniel E. Sichel (19H7),
"]\'Ioney Demand: The Effects of Inf1mion and
Alternative AdjL1stlTJl.!nt Mcch:lIlisms," Rel1iew (if
Ecollomics (fl/d Sialislics 69 (Allgu,~t): 511- 15
1-larlX-'fgt..'I', C. Amoid 09(3), ·...'11C Dynamics of Inf1:Ition in Chile: in Ml!lwlIT!melll ill 1::.'c:0I10mics
Essays ill Mallx' lIwlica/t'collomics (lmlteol/omelrics ill .HemOl)'of )'ehllda GIl!,ifeld, ed. Carl
F Christ (Stanford: Stanford L:niwrsity ]>n:ss)
IIcndl)'. Dj\'id F (979), " P~d i<.1i\'e Failure :lnd
Econo metric ~Iodelling in ~1:Knlt.."CO!l()mil'S:

Economic Review -

March 199 1

H:lc-$hin (1985), ;Test of the Adjustment
Process .. nd Unear J lomogeneily in a Stock
Adjustment McxlcJ of Money Dcmand," Review
oj EcOl/omies lind 5~tlliSfics 67 (November):
689-92.

Mishkin, Frc.'oeric S (198 1), "-111e Reailbte: An
Empiril':l llnvesligalion," Canwgie-Rocbesler
COlifen'l1ceSeries Oil Public Policy IS: 519-41.
Poner, I) Richa rd. '!1lOm:1S D. Simpson, and Eik'Cn
MlIlI:-.kopf ( 1979), -Fin:lnci31 Innovations and
Monetary Agg~gat(.'S ," Brookings Papers 0"
Ecollomic ACfivify, no 2
!'oncl'lla , M. james, .tnd j ulio J R()(cmberg (1987).
-Money in the Utility Function: An Empirical
Impit..·mcnt:llion, " in New Approaches 10 Monelmy EcOllomics, ed. William A. Barnett and
Kennl.!lhj . Singleton CCunbridge; Camhridge
University Press), 219-40.
Hasche, nohcn II. ( 1989), "Some Evidence on the
Elusivc 1982 Shift in Vclociry Drift," in Shadow
0l~n j\'!:u'kc( CommiWo.,'e, "Policy Statement and
Position Papt..'rs," Public Policy Working Paper
~rics, no, PI'S 89-01, Univcrsity of Hocbester
(Rochester, N,Y,; University of Roche;ter, BrJdley
Policy He:o.c:lrch Cemer, 19-20 M:m.:h), 37-,m
Roley, v, Vance (985). "Money Dcm;md Predictability, " jOIln1t11 ofMOII~J" Credil, alld IJa"ki"g
17 (November, pt. 2): 6 11-41.