The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Review Vol. (is, No. 8 October 198(S T h e A n d e rse n -Jo rcla n A p p ro a c h a f te r Y e a rly 2 0 Y ears 9 I he M onetary-Fiscal Policy Debate and the Andersen-Jorclan Equation 18 A M onetarist Model for E con om ic Stabilization: Review and Update 29 M onetary and Fiscal Actions: A Tfest of Their Relative Im p o rtan ce in Econom ic Stabilization 4.") A M onetarist Model for E con om ic Stabilization T h e Review is p u b lish ed 10 tim es p e r y ea r by th e R esea rch a n d Public Info rm ation D e p a rtm en t o f th e F ed era l R eserv e Rank o f St. Louis. Single-copy su bscrip tio n s a re available to th e p u b lic f r e e o f ch a rg e. Mail re q u e sts f o r su bscrip tio ns, back issues, o r a d d res s ch a n g e s to: R esea rch a n d Public Inform ation D ep a rtm en t, F ed era l R eserve Rank o f St. Louis, P.O. Rope 442, St. Louis, M issouri 63166. T h e views e x p r e s s e d a re th o se o f th e individual a uthors a n d d o not n ecessa rily re fle c t official positions o f th e F ed era l R eserv e Rank o f St. Louis o r th e F ed era l R eserv e System . A rticles h e r e in may b e re p rin t e d p ro v id ed th e s o u r c e is cred ite d . Please p ro v id e th e Rank's R esea rch a n d Public Inform ation D ep a rtm en t with a copy o f re p rin te d material. F e d e ra l R e se rv e B ank of St. L ouis Review O ctob er 1986 In This Issue . . . In M em oriam Leonall C. Andersen 1924-1985 T h is issu e o f th e Review is d ed ica ted to th e m em ory o f L eonall C. A nd ersen, w ho d ied on O cto b er 2 7 ,1 9 8 5 . A ndy’s involvem ent w ith an d co n trib u tio n to ec o n o m ics sp an n ed a perio d o f th ree d ecad es, divided alm o st equally b etw een aca d em ia an d th e F ed eral Reserve Bank o f St. Louis. He began h is ca re e r in 1955 as an assistan t p ro fesso r o f e c o n o m ics at St. O laf C ollege in N orthfield, M in n eso ta; at th e tim e o f h is d eath , h e w as P rofessor o f E co n o m ics at G ustavus A dolphus College in St. Peter, M in n esota. F o r 16 y ears, how ever, from August 1962 to August 1978, h e w as a m e m b er o f th e R esearch D ep artm en t at th is Bank, w h ere h e co n trib u te d sign ifican tly to re se a rc h in m o n etary issu es. A ndy’s best-kn ow n co n trib u tio n s, b o th to th e e c o n o m ics p ro fessio n at large an d to th e rep u tatio n o f th e F ed eral Reserve B ank o f St. Louis in m o n etary p o licy m atters, ap p eared in tw o articles in th is Review. “M onetary an d F isca l A ction s: A T est o f T h e ir Relative Im p o rta n ce in E co n o m ic Stab ilizatio n ,” co -a u th o red w ith Jerry L. Jo rd an , w as p u b lish ed in N ovem ber 1968; it gave rise b o th to th e w ellknow n A n d ersen -Jo rd an (or St. Louis) eq u atio n an d to th e u n en d in g con troversy abou t its m ethod ology an d resu lts. “A M on etarist M odel for E co n o m ic Stabiliza tio n ,” co -a u th o red w ith K eith M. C arlson, w as p u b lish ed in April 1970; it devel op ed a sm all m o n etarist m a c ro e co n o m ic m od el that h as b ee n u sed for policy p u rp o ses at th is Bank an d elsew h ere. To h o n o r Andy, w e are rep rin tin g th e se original articles, along w ith ev aluations by h is co -a u th o rs an d o th ers, o f th e ir co n tin u in g im p o rta n ce an d relev ance. T h e se w ill b e follow ed by a b rief b io g ra p h i cal sk etch an d a se le c te d b ibliograph y o f A ndy’s im p o rtan t co n trib u tio n to eco n o m ics. * * * In th e first article in th is Review, “T h e A n d ersen -Jo rd a n A p p ro ach after Nearly 20 Y ears,” Jerry L. Jo rd a n d escrib es th e th eo retica l an d p o licy -related issu e s o f th e late 1960s th at led h e an d Andy to develop w h at is n ow called th e A n d ersen Jo rd a n ap p ro ach . First, th e y view ed th e ir re sea rch as a seq u el to w ork by Friedm an-M eiselm an , B ru n n er-M eltzer an d o th ers w ho h ad d em o n stra ted th e gen eral p o te n ce o f m o n etary p o licy a ctio n s. In ad d ition , th e ir re se a rc h attem p ted to an sw er a p o licy issu e o f vital c o n c e rn in 1968: w h eth er th e expansive m o n etary stim u lu s o r th e restrictive fiscal p o licy a ctio n s th e n in p la ce w ou ld have th e p red o m in an t im p act on sp en d in g. Jo rd a n th e n d iscu sse s w hy, in h is o p in io n , th e A n d ersen -Jo rd an m e th o d an d resu lts w ere co n sid ere d so con trov ersial. Finally, after review ing so m e rea so n s often cite d for th e alleged d em ise o f m o n eta rism in re cen t y ears, Jo rd a n co n c lu d e s th at th e A n d ersen -Jo rd an a p p ro a ch en d u res; 3 In This Issue . . . specifically, th e sing le-equ ation , red u ced -fo rm m eth o d olo gy em p loy ed to asse ss th e effects o f different p o licy a ctio n s on th e eco n o m y rem a in s u sefu l today. In the se c o n d article in this issue, D allas S. B atten an d D aniel L. T h o rn to n ex am in e th e con troversy th at em erged follow ing th e p u b lica tio n o f th e 1968 A n d ersen -Jo rd an article (see rep rin t o f th e original article b eg in n in g on p. 29). A nd ersen an d Jo r d a n ’s b a sic finding — th at m o n etary a ctio n s have a lastin g im p act on ec o n o m ic activity w hile fiscal a ctio n s do n o t — g en era ted im m ed iate criticism , m o st o f it eco n o m etric in n atu re, th e a u th o rs n o te. Using th e sam e data set as A n d ersen an d Jo rd a n did, B atten an d T h o rn to n investigate th e validity o f th e m a jo r criticism s o f th e St. Louis eq u atio n . In particu lar, th ey te st for b ia s due to m issp ecificatio n , th e p re se n c e o f en d o g en o u s variables on th e rig h t-h an d side o f th e eq u atio n , an d th e o m issio n o f ce rta in righ t-h an d -sid e variables. M oreover, th ey rep o rt find ings p u b lish ed elsew h ere abou t th e d yn am ic sp e cifica tio n o f th e St. Louis eq u atio n an d th e im p o sitio n o f p olyn om ial re strictio n s. In n o in sta n ce co u ld th e a u th o rs find statistical su p p o rt for th e alleged, b u t previously u n tested , im p ro p rieties involving th e estim a tio n o f th e St. Louis eq u atio n . In th e th ird article, K eith M. C arlson reviews th e d ev elop m en t o f th e m o n eta rist e co n o m e tric m od el th at h e an d Andy p u b lish ed in th e April 1970 Review article (reprinted b eg in n in g on p. 45) an d p re sen ts an u p d a ted version, co m p arin g it w ith th e original m od el. T h e gen eral form o f th e m od el, th e a u th o r rep o rts, rem ains u n ch an g e d ; it h a s b e e n sim plified to a rate-o f-ch an g e form , how ever, an d som e su p p ly -sid e variables have b ee n add ed . C arlson ex am in es th e p ro p erties o f th e u p d a ted m od el an d su m m arizes its sim u lation p erfo rm an ce. T h e p ro p erties o f th e m o d el are little ch a n g ed from th e original A nd ersen -C ariso n a rticle: m o n etary a ctio n s have su b sta n tia l sh o rt-ru n effects o n total spen din g, o u tp u t an d u n em p loy m en t, b u t over th e lo n g run, the effect o n total sp en d in g is reflected alm ost en tirely in th e p rice level. A lthough th e m o d el's sim u lation p erfo rm an ce a p p ears satisfactory, a m o re a cc u ra te evaluation aw aits co m p ariso n w ith sim ilar re su lts from o th e r m od els. 4 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 The A ndersen-Jordan A pproach after Nearly 30 Years Jerry L. Jordan T J L O m y know ledge, th e label "m o n etarist' and its essen tial p ro p o sitio n s w ere first p u t forth in an article in th e Ju ly 1968 Review o f th e F ed eral Reserve Bank o f St. Louis en titled “T h e Role o f M oney in M onetary Policy” by Karl B ru nn er. In it, B ru n n e r states that, T h e critiq u e o f e sta b lish e d p o lic y p ro c e d u re s, w h ic h evolved from th is re s e a rc h in to q u e s tio n s c o n c e rn in g th e m o n e ta iy m e c h a n is m , is d erived from a b o d y o f m o n eta ry th e o ry refe rre d to . . . a s th e M onetarist p osi tion [e m p h asis ad d ed ]. T h re e m a jo r c o n c lu s io n s have em erg ed from th e h y p o th e se s p u t fo rth . First, m o n e tary im p u lse s are a m a jo r fa c to r a c c o u n tin g fo r v aria tio n s in o u tp u t, e m p lo y m e n t a n d p ric e s . S e co n d , m o v em en ts in th e m o n e y s to c k are th e m o st reliab le m e a su re o f th e th ru st o f m o n e ta ry im p u lse s. T h ird , th e beh av ior o f th e m o n eta ry a u th o rity d o m in a te s m ove m e n ts o f th e m o n ey sto ck over b u s in e ss cy c le s. T h e p ro c e ss o f defining an d refining w hat w e now m ean by m onetarism grew out o f a con troversy that em erged in th e 1960s regarding th e relative im p o r ta n ce o f m on etary an d fiscal im p u lses. T h e d om in an t eco n o m ic policy fram ew ork o f th e day w as an o u t grow th o f th e thin king o f th e 1930s, w h ich b eca m e know n as K eynesianism . T h e rapid grow th o f govern m ent sp end in g asso ciated w ith th e V ietnam War, th e Great Society program s an d relatively large d eficits in th e federal govern m en t’s bu d g et w ere a sso ciate d w ith co n d itio n s o f rapid e c o n o m ic grow th, rising inflation an d a low u n em p loy m en t rate. T h e p o litical p re scrip tion for th e problem o f inflation w as a su rtax on person al an d co rp o rate in co m es to restrain aggregate d em an d an d re d u ce inflation. Jerry L Jordan is a senior vice president and chief economist at First Interstate Bancorp, Los Angeles, California, and a former director of research at the Federal Reserve Bank of St. Louis. A d herents to th e p ro p o sitio n s th at b eca m e know n as m o n etarism q u estio n ed w h eth er su c h fiscal a ctio n s w ould, in fact, restrain aggregate d em a n d an d red u ce inflation if m o n etaiy grow th rem ain ed as rapid as it h ad b e e n previously. S in ce th e d o m in an t p o sitio n o f th e tim e w as know n as th e K eyn esian revolution, th e critics o f th at view w ere said to have m o u n ted a m o n etarist co u n terrev olu tion . At tim es, th e quality of th e d iscu ssio n w as silly, in clu d in g su ch in sigh ts as “you can pull on a string, bu t y ou c a n ’t p u sh on a strin g” an d “y ou ca n lead a h o rse to w ater, b u t y ou c a n ’t m ake him d rink.” At o th e r an d m ore u seful tim es, state-of-th e-art eco n o m etric te ch n iq u e s w ere u sed to test rival c o n je c tu re s ab o u t m o n etary an d fiscal im pulses. T h e s k ir m is h e s o f th e p e rio d in c lu d e d th e F riedm an-H eller d ebate, th e “B attle o f th e Radio Sta tio n s” — w h ich referred to th e re sea rch d o n e by A ndo-M odigliani (AM) an d F ried m an -M eiselm an (FM) — an d th e a sso cia te d co n trib u tio n s by D ePrano, H ester an d M ayer an d by B ru n n e r an d M eltzer. T h e role o f w hat b eca m e know n as th e gadfly o r m averick resea rch d ep artm en t o f th e Fed eral Reserve Bank o f St. Louis is n ow w ell-know n in ec o n o m ics circle s; al th ou gh it h ad started a few y ea rs earlier, it w as given greatly h eig h ten ed visibility w ith th e p u b lica tio n o f th e A n d ersen -Jo rd an (AJ) a rticle in 1968. T h e 1966 cred it cru n c h an d su b seq u en t “m inire c e s s io n ” h ad d em o n stra ted th e p o ten tial for a re strictive m o n eta iy policy, m easu red in term s o f a d e ce le ra tio n o f m o n e ta iy grow th, to d o m in a te an expansive fiscal im p u lse. In 1968, th e issu e w as w h eth er m o n etary stim u lu s — as in d ica ted by c o n tin u ed rapid grow th o f m o n ey — co u ld d o m in ate a restrictive fiscal im p u lse as m easu red by a tax in crease, re d u ctio n in th e high em p lo y m en t d eficit or 5 FEDERAL RESERVE BANK OF ST. LOUIS som e o th e r su m m ary variable. T h e re se a rc h u n d erly ing th e AJ article w as m otivated by tw o events: th e Jo h n so n a d m in istratio n ’s an ti-in flatio n su rch arg e on p erso n al an d co rp o rate in c o m e tax an d th e FOM C’s d ecisio n to ease m o n etary p o licy to cu sh io n th e p re su m ed highly restrictive effects o f th e tax su rch arg e. W e co n sid ere d th e AJ a rticle to b e a seq u el to th e FM article. O ur p u rp o se w as to rigorously form u late p o ten tially falsifiable h y p o th eses abo u t various m a c ro e c o n o m ic p o licy a ctio n s. T h e article also w as an ex ercise in applying w h at w as th e n state-o f-th e-art co m p u te r ized regressio n p rogram s u sin g th e A lm on d istribu ted lag for testin g h y p o th eses. I believe that th e v eh e m en ce o f th e attack s o n th e AJ a rticle aro se from tw o so u rces. First, th e resu lts o f th e study sharply co n tra d icted th e in h erited w isd om of th e tim es and raised serio u s d o u bts about o u r ability to u se activist/discretionary fiscal p o licy to in flu en ce th e eco n o m y in p red ictab le w ays. A se c o n d re a so n fo r th e attack s w as sim ply th at th e red u ced -fo rm ap p ro ach u sed in th e AJ study re p re sen ted a th reat to ec o n o m e tric m od el bu ild ers; it provided a lo w -co st alternative to th e expensive efforts at th e tim e to bu ild large-scale stru ctu ral m o d els of th e U.S. eco n o m y . F o r exam p le, aro u n d th at tim e, th e F ed eral R eserve B oard h ad e n tere d in to a co n tra ct to sp en d ab o u t $1 m illion to bu ild w hat b eca m e know n as th e FRB-M IT e c o n o m e tric m o d el. In m y ju d g m en t, th e stru ctu ral m o d el-b u ild ers o f th e tim es sim ply co u ld n o t afford to leave u n ch allen g ed th e c o m p e ti tio n th a t th is relatively ch e ap ap p ro ach p re sen ted in evaluating p o licy effects on th e eco n om y. In an y event, th o se th rea te n ed by its ch allen g e b o th to e co n o m ic orth od oxy an d to th e u sefu ln ess o f large-scale m od els h a d great in cen tiv e to seek to d iscred it th e AJ m e th o d ology. T hu s, as th e d ecad e o f th e 1960s en d ed , th e lin es h ad b e e n draw n for a p rolon g ed in telle ctu al battle. T h e K eynesian revolution w as still d om in an t, b u t th e ch allen g e o f th e m o n etarist co u n terrev o lu tio n h ad b ee n initiated . T h e 1970s w as a d ecad e o f form ulation, reform u lation an d em p irical testin g o f th e alternative views o f th e m ajo r m a cro e co n o m ic in flu en ces on the eco n om y. By th e en d o f th e 1970s and in to th e early 1980s, fu rth er testin g o f m o n etarist p ro p o sitio n s by a ctu al im p lem en tatio n w as attem p ted , at least in nam e, if n o t in fact. R ecently, th ere have b een n u m ero u s claim s that m o n etarism h as failed. C ertainly th e ca se against m o n etarism h as b ee n tried in th e p ress w ith jo u rn a l ists actin g as b o th p ro se cu to rs and jury. E co n o m ists, 6 OCTOBER 1986 how ever, have y et to co m p lete th e ir d elib eratio n s. B eca u se th e “breakd o w n ” o f th e AJ re su lts is often cited as evid ence against m o n etarism , I w ou ld like to co m m en t briefly on th e cu rre n t controversy. ON T H E RECEN T FA ILU R E’ O F MONETARISM: IS ANDERSEN-JORDAN’ P A S S E ? T h e failure o f m o n eta rism h a s b e e n a sse rte d an d reasserted often d uring th e p a st few y ears. T h is failure is b ased on th e co n ten tio n th at th e relatio n sh ip b e tw een m o n ey grow th an d GNP, o r m o n ey grow th an d inflation, h as bro k en dow n. In p articu lar, th e beh avior of th e in co m e velocity o f m o n ey over th e last few y ears, esp ecially in 1985, h as raised q u estio n s a b o u t o n e of th e cen tral p ro p o sition s o f th e qu an tity th eo ry of m oney, o r m o n etarism — nam ely, th e stability o f the dem an d fu n ctio n for m on ey. T h e sh arp d e c re a se s and in crea ses in co n v en tion ally m e a su red velocity have led to a ssertio n s th at th e d em a n d for m o n ey is u n s ta ble and, th erefore, th e m o n ey su p p ly (M l) is n o lo n g er a reliable in d ica to r o f th e th ru st o f m o n etary policy im pulses. Growth rates o f M l over sh o rt intervals d urin g th e last few y ears have b een highly volatile, an d th e c o n tem p o ran eo u sly m e a su red ratio o f GNP/M1 also h as flu ctu ated over a w id e range. U nfortunately, b o th m o n etarists an d th e ir critics seem to a c c e p t th e view th at th e p u b lic p o licy relev an ce o f m o n eta rism d e p en d s on th e sh o rt-ru n stability o f th e fu n ctio n al relatio n sh ip b etw een d o m estic in c o m e and/or ou tp u t an d grow th o f th e m o n ey supply. T h e sh o rt-ru n vola tility o f a d ata series su ch as th e ratio GNP/M1, h ow ever, d oes n o t y ield an y definitive in fo rm atio n about th e stability o f th e u nd erlying functioned relatio n sh ip b etw een m o n ey an d ec o n o m ic activity. T h ere are th ree im p o rtan t a sp e cts o f v elocity beh av io r in th e re ce n t p ast th at m u st b e exam in ed : th e role o f in stitu tio n al ch an g es, th e im p lica tio n s o f lags and th e ap p ro p riaten ess o f th e n u m era to r in co m p u tin g th e velocity ratio. Institutional Changes Som e an alysts a ssert th a t d eregu latio n o f th e fin an cial system , starting w ith th e D epository In stitu tio n s D eregulation an d M on etary C ontrol Act o f 1980 and co n tin u ed w ith th e G arn-St. G erm ain legislation in 1982, h as a ltered th e beh av io r o f M l and , in tu rn, h as ca u sed th e volatile beh avior o f v elocity in th e last two y ears. T h e em p irical issu e, o n w h ic h th e re is n o c o n FEDERAL RESERVE BANK OF ST. LOUIS sen su s at th is tim e, is th e degree to w h ich th e "d e m an d for m o n ey b a la n c e s ” has shifted upw ard rela tive to in c o m e o r w ea lth and , c o n se q u e n tly , a d ow nw ard shift in at least th e level, an d po ssibly th e tren d grow th rate, o f M l velocity. It h as b ee n argued th at th e rem oval o f R egulation Q in tere st ceilin g s on ban k d ep o sits an d th e innovation of n ew types o f d ep osit in stru m en ts has resu lted in M l co n tain in g a large savings co m p o n e n t. Therefore, faster grow th o f th e m o n ey supply, su ch as w hat o cc u rre d in 1985-86, sh o u ld not be taken as an in d ica to r o f as m u ch stim u lu s as befo re sin c e th e d em an d for m o n ey b a la n ces is also rising. W ith out an e x cess su p p ly co n d itio n , it is argued, th e re is no re a so n to ex p e ct n om inal in co m e grow th to accelera te. M o n etarists generally a c c e p t th e view th at d eregu la tio n an d fin an cial inn ovation have m o st likely resu lted in a red u ctio n in th e tren d rate o f grow th o f M l velocity. T h ere is no reason , how ever, to believe th at th e level o f velocity w ou ld b e ex p e cted to d eclin e. T h e tren d rate o f in cre a se o f h isto ric M2 velocity h as b ee n betw een 0 an d 1 p e rce n t. M2 h as alw ays co n tain e d a relatively large savings co m p o n e n t. D eregu lation an d innovation m ay have resu lted in M l taking on the ch a ra cte ristics o f h isto ric M2. At th is p oin t, how ever, w e do n o t have su fficient evid ence to draw firm co n clu sio n s. Even if M l is n ow like old M2, th e tren d rate o f in crea se o f M l velocity w ould have d eclin ed from th e 3 p e rce n t rate o f th e p o st-w ar period to so m eth in g m ore sim ilar to h isto ric M2 velocity grow th. T h e re is no th e o retical reaso n an d no evid ence to suggest that th e tren d o f M l velocity w ould be negative as a resu lt o f d eregu lation and innovation. The Role o f Lags T h e e x iste n ce o f lags in an en viro n m en t o f highly volatile sh o rt-ru n m o n ey grow th m ust p ro d u ce a highly volatile d ata series for velocity. Volatility o f the d ata series, how ever, d o es n o t y ield u seful in form ation abou t th e stability o f th e u nd erlyin g fu n ctio n al re la tion sh ip . T h is p o in t ca n be illu strated w ith a sim ple exam ple. Su p p o se it w ere know n w ith certain ty th at th e lag b etw een ch an g es in m oney and ch an g es in n om in al in co m e w as exactly 180 days. Su p p o se fu rth er th at th e grow th o f M l a cc e le ra te d an d d ecelera ted sharp ly over intervals lasting exactly six m o n th s. Starting from any initial con d itio n , a sharp accelera tio n in M l grow th for six m o n th s w ould n o t be m a tch ed by an a ccelera tio n in th e grow th o f th e n u m era to r o f th e OCTOBER 1986 v elo city ra tio . C o n se q u e n tly , c o n te m p o ra n e o u s ly m easu red velocity (V) w ou ld d eclin e d urin g th e interval. At th e en d o f six m o n th s, th e n u m era to r w ould begin to rise m ore rapidly, w hile th e d en o m in a to r d ecelerated , ca u sin g a sh arp in cre a s e in th e V-ratio. Six m o n th s later, o n c e again th e d en o m in a to r w ould a ccelera te w hile th e n u m era to r d ecelera ted , ca u sin g a plu n ge in th e V-ratio. O bserving th e b eh av io r o f th e Vratio over several s u c h cy cles co u ld easily lead an u nd ergrad u ate m on ey -an d -b an kin g stu d en t to c o n clu d e th at velocity, o r th e d em an d for m oney, w as not stab le. It w as a p erfectly stab le a n d p re d icta b le fu n c tion al relatio n sh ip , how ever, that p ro d u ced th e vola tility o f th e d ata series. T h e relevance o f th e p o in t is that, over th e last several y ears, w e have observed in creasin g ly sharp a ccelera tio n s an d d ecelera tio n s o f M l grow th, w ith e a ch m ovem ent in th e rate o f ch a n g e ten d in g to last tw o to th ree qu arters follow ed b y a sh arp reversal. Sin ce th e real-w orld lag is n o t d iscrete, bu t ra th er is d istribu ted and variable, th e ch a llen g e to em p irical re sea rch is to develop te ch n iq u e s to id entify th e a ctu al lag stru ctu re. Appropriate Numerator It is co m m o n p ra ctice to co m p u te th e velocity of m o n ey as th e ratio o f GNP divided by M l o r a b ro ad er m o n ey m easu re. T h e original qu an tity th eo ry eq u a tion, how ever, w as MV = PT, w h ere T is tra n sa ctio n s. C hanges in GNP reflect prim arily ch a n g es in d o m estic output at prevailing p rices, n o t total tra n sa ctio n s at prevailing p rices. T h e u se o f GNP in th e velocity ratio im plies b o th a clo sed eco n o m y a n d th e stability o f su ch co m p o n e n ts as b u sin ess in ven tories. S in ce th o se assu m p tio n s are not a good rep resen ta tio n o f th e real w orld, th e u se o f GNP for co m p u tin g velocity ca u ses several p roblem s. In casu al conversation, it is co m m o n to refer to GNP as a m easu re o f aggregate d em a n d o r to tal sp en d in g in th e eco n om y. It is n o t. A rigorous form u lation o f th e qu an tity theory o f m oney, o r m o n etarism , involves a sta tem en t abou t th e d em a n d fo r m o n ey b a la n c e s re la tive to w ealth o r p e rm a n en t in co m e. Sin ce qu arterly flu ctu atio n s in GNP as co m p iled by th e C om m erce D epartm en t are not a good proxy for ch a n g es in w ealth o r p erm a n en t in co m e, th e u se o f GNP to co m pu te th e inverse o f th e d em an d for m o n ey — velocity — ca u ses sign ifican t m e a su re m en t p roblem s. T h e th eory im plies th at an a ccelera tio n in m o n etary grow th resu lts in in crea sed sp en d in g grow th. In a 7 FEDERAL RESERVE BANK OF ST. LOUIS clo sed eco n om y, a sh o rt-ru n m an ifestatio n o f th e in crea se d sp en d in g w ould be an involuntary an d u n a n ticip ated re d u ctio n in b u sin ess inventories, in crea sed ord ering and in creased p ro d u ctio n . T h e longer-term effect is a re d u ctio n in th e p u rch asin g p o w er o f n o m i nal m o n ey u n its — inflation. In an o p en econom y, an in cre a se in m o n ey grow th m ay b e acco m p a n ie d by an in cre a se in d em an d for im p orted good s as w ell as d om estically p ro d u ced output. A sh arp an d su stain ed a ccelera tio n o f m on ey grow th th at is acco m p a n ie d by a large in crea se in im p orts su ggests a d eclin e in th e GNP/M1 ratio, at least for a w hile. O th er a d ju stm en ts, how ever, grad u ally do take p la ce — su ch as in th e foreign ex ch an g e value o f th e d o m estic cu rre n cy w h ich ch an g es th e relative p rices o f in tern atio n ally trad ed o u tp u t; even tually m ore co stly im p orts and m ore com p etitive e x p o rts will reverse th e situ ation. T h o se lags can be veiy long an d are difficult to p red ict, in tro d u cin g fu rth er u n certain ty into th e relation b etw een m o n ey an d GNP grow th. T h is p h e n o m e n o n lim its th e u sefu ln ess o f m o n etarism in co n d u ctin g sh o rt-ru n -o rien ted d iscre tionary p o licies sin c e th e u sefu ln ess o f m on ey grow th as a n in d icato r o f th e th ru st o f m o n etary p o licy is usually gauged in term s o f its reliability in forecastin g GNP grow th. WHAT REMAINS O F TH E ANDERSEN-JORDAN APPROACH? O ne ce n tra l m o n etarist p ro p o sition h as alw ays b een th at activist, d iscretio n ary p o licies are n e ith e r n e c e s sary n o r d esirable. T h erefore, it is iro n ic that th e "St. Louis eq u atio n ” u n in ten tio n ally stren g th en ed th e view s o f th e p u b lic p o licym akers w h o w an ted to “m a n ag e” m o n e ta iy policy to ach ieve d ifferent e c o n o m ic resu lts. T h e em p irical relatio n b etw een m o n ey grow th an d n o m in al in co m e w as u sed as a ratio n a liz a tion for an activist, d iscretio n ary policy u n d e r w h ich fa ste r o r slow er target grow th rates for m o n ey w ere ad o p ted to achieve faster o r slow er grow th rates o f n o m in al GNP and, in tu rn, m ore o r less inflation, o u tp u t an d em p loym ent. 8 OCTOBER 1986 T h e u se o f th e St. Louis eq u atio n to engage in “finetu n in g ” w as n e ith e r in ten d e d n o r a n ticip a te d by us. As I n o ted earlier, o u r in ten t w as to d em o n stra te th at expansive m o n eta iy fin e-tun in g, in ten d ed to offset a p resu m ed co n tra ctio n a ry fiscal im p act, w as n eith e r n e c e s s a iy n o r d esirab le. M u ch to o u r su rp rise and chagrin, o u r resu lts w ere u sed by so m e to d em o n strate th e efficacy o f m o n etary fin e-tu n in g . T h e b a sic p ro b lem w ith activist p o licies is th at n u m erou s factors affect e c o n o m ic p e rfo rm a n ce; in th e p a st two d eca d es, th e re have b e e n am p le o p p o rtu n i ties to a ccu m u la te d ata abou t th e effects o f b o th p o licy an d n o n -p o licy d ev elo p m en ts on e c o n o m ic activity. Yet, very little, if anything, h a s b e e n le a rn ed from all th is a ccu m u la ted ex p e rie n ce . T h e lags in th e effects of policy a ctio n s are ju s t as variable an d ju st as u n c e r tain, an d p o licy a ctio n s still a c c o u n t fo r le ss th a n h alf th e variability o f ec o n o m ic variables. T h e ap p en d ix to th e A n d ersen -Jo rd a n a rticle e m p h a siz es th e im p o rta n ce o f th e “Z -factor,” a variable su m m arizin g all th e o th e r fo rces th a t in flu en ce to tal sp en d in g in th e eco n o m y . W hile th e tex t o f th e article co n clu d e d th at m o n eta ry p o licy a ctio n s w ere re la tively m o re im p o rtan t th an fiscal a ctio n s, th e analysis in th e ap p en d ix su ggests th a t a m o re co m p lete c o n clu sio n w ould have b e e n “an d n o n -p o licy fa cto rs are even m o re im p o rta n t.” T h e a rtic le ’s im p a ct o n e c o n o m ic policym aking w ou ld have b e e n m o re favorable had it n o t led to an in c re a s e d re lia n ce on m o n etary over fiscal policy, b u t h ad it in stea d co n trib u te d to a gen eral d e-em p h asis o f fin e-tu n in g a ttem p ts b y p o li cym akers. Som e o f th e fru stratio n an d d isa p p o in t m ent ex p ressed by m o n eta iy p olicym akers in re cen t y ea rs m ay have resu lted from th e u n satisfacto ry resu lts th ey observed from th is m isu se o f th e theory. In m y ju d g m en t, th e en d u rin g co n trib u tio n o f th e AJ a p p ro ach is th e m eth od ology em p loy ed to a sse ss th e d ifferential im p a cts o f p o licy a ctio n s o n th e eco n om y, n ot th e sp ecific resu lts offered at th e tim e. W hile in stitu tio n al an d tech n o lo g ica l ch a n g es over tim e m ay alter em p irical results, stu d en ts o f p u b lic p o licy d e b a te s c a n still u sefu lly ap p ly to d a y th e sin g le eq u ation , red u ced -fo rm a p p ro a ch u sed by th e AJ study 18 y ears ago. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 The M onetary-Fiscal Policy Debate and the A ndersen-Jordan Equation Dallas S. Batten and Daniel L. Thornton p ERHAPS Andy's m o st im p o rtan t an d lastin g c o n trib u tion to th e ec o n o m ics p ro fessio n w as h is re sea rch w ith Jerry Jo rd a n th at resu lted in th e p u b lica tio n o f th e A n d ersen -Jo rd an (A-J) eq u atio n or, as it is m ore w idely know n, th e St. Louis eq u atio n . Alm ost im m ed iately, th e tw o fou n d th eir w ork th e su b je ct o f in ten se criticism an d controversy — m u ch o f w h ich co n tin u es, thou gh in to n e s th at are significantly m u ted .' W hile th e criticism s o f A n d ersen -Jo rd an w ere fo cu sed on various te ch n ica l and ap p lied eco n o m etric a sp e cts o f th e ir work, th ey w ere m otivated, in large part, by A -J’s co n clu sio n th at m o n etary p o licy h as a significant and lasting effect on n o m in al GNP and th at fiscal p o licy h as n o lastin g effect. T h e se re su lts c o n flicted sh arp ly n o t only w ith th e co n v en tion al w isd om abou t th e relative effects o f m o n etary an d fiscal p o licy a ctio n s bu t w ith th e resu lts o f large-scale eco n o m e tric m od els o f th e tim e. T h e p u rp o se o f th is p ap e r is to review th e criticism s that em erged follow ing th e p u blicatio n o f th e A-J equation.- W e n o te that m any, if n ot all, o f th e criti cism s o f th e A-J p ap e r apply equally w ell to th e vast m ajority o f p u blish ed re search , th en and now . M ore im portantly, u sin g th e original A-J data, w e find no evid ence to su p p o rt th e se criticism s. Dallas S. Batten and Daniel L. Thornton are research officers at the Federal Reserve Bank of St. Louis. Rosemarie V. Mueller provided research assistance. 'The monetary-fiscal policy debate was actually initiated prior to Andersen-Jordan (1968) by Friedman and Meiselman (1963). Just as the ensuing debate surrounding Friedman and Meiselman’s results was waning, however, Andersen and Jordan appeared, rekindling and intensifying the disagreement over the relative effi cacy of monetary and fiscal policies. 2While our review differs from recent ones by McCallum (1986) and Meyer and Rasche (1980), it is necessary to traverse some of the ground they covered. TH E ST. LOUIS EQUATION AMD ITS CRITICS R ecently, C ooley an d LeRoy (1981) have argued th at a clo se co rre sp o n d e n c e te n d s to exist b etw een th e advocacy o f a th eo ry an d th e re su lts o f scien tific inves tigation. It is not surprising, th erefo re, th at w h en two know n an d vocal p ro p o n en ts o f m o n etarism rep o rted em p irical resu lts th at strongly su p p o rted m o n etarist pro p o sition s, th e resu lts w ere received w ith sk ep ti cism , w h ich w as in ten sified by th e ir u se o f a single, "re d u ce d -fo rm ” eq u atio n . C ritics w ere su sp icio u s that A-J inadvertently h ad eith e r m issp ecified th e m od el or used faulty e co n o m etric te ch n iq u e s to ob tain th eir results/' T h ree m a jo r criticism s em erg ed follow ing th e p u b li catio n o f th e A-J eq u atio n . First, it w as argued th at th e eq u ation w as m issp ecified b ec a u se im p o rtan t exo g e nous, righ t-h an d -sid e variables had b een exclu d ed . Secon d , critics claim ed th at A -J’s u se o f ord in ary least sq u ares (OLS) h ad resu lted in sim u lta n eo u s eq u atio n bias. Finally, it w as asse rte d th at A-J h ad failed to identify th e relevant exo g en o u s in d ica to rs o f m o n etaiy an d fiscal policy a ctio n s. In ad d ition , critics w ere co n cern e d th at th e A-J resu lts w ere sa m p le-sp ecific or n ot robu st to various ec o n o m e tric m o d ification s, in clu d in g th e ir u se o f A lm on’s (1965) poly n om ial d istrib uted lag estim atio n tech n iq u e. T h e p erce p tio n th at A-J had so m eh o w erred w as e n h a n c e d w h en d e Leeuw and K a lch b ren n er (1969), Silb er (1971) an d Sch m id t an d W aud (1973) tried u n su ccessfu lly to re p lica te th e 3A number of critiques appeared very shortly after the publication of the A-J paper, e.g., de Leeuw and Kalchbrenner (1969), Davis (1969), Corrigan (1970), and Goldfeld and Blinder (1972). 9 FEDERAL RESERVE BANK OF ST. LOUIS A-J resu lts.4 T h e follow ing se c tio n s ex am in e th e se c rit icism s. M ISSPECIFICATION T h e ch arge th at A-J h ad m issp ecified th e ir eq u atio n by om ittin g im p o rtan t variables, o th e r th a n m on etary an d fiscal p o licy variables, w as leveled by n u m ero u s co m m en tato rs. To u n d e rstan d th is argum ent, c o n sid e r th e original A-J eq u atio n : 3 ( 1 ) AY, = a + 2 i= 0 3 p, A M ,_, + 2 i= 0 -y, AE , , + u, , w h ere Y, M an d E d en o te n o m in al GNP, th e m on ey stock (M l) and n o m in al h ig h -em p loy m ent govern m e n t exp en d itu res, respectively, an d u, d en o te s th e u su al rand om d istu rb an ce term .3 T h is eq u atio n ca n be w ritten m o re co m p actly as: (2) AY, = a + (J (L )A M , + 7 ILIA K , + u, , w h ere (3(L) an d 7(L) are p olynom ials in th e lag o p era to r L, su c h th a t L"x, = Lx,_„ an d w h ere p(L)Ax, are d istribu ted lags o f a finite o rd er k.“ A-J c h o se k = 3. If a relevant ex o g en o u s p o licy variable, Z„ is om itted , th e tru e sp ecificatio n is n ot eq u atio n 2, but 131 AY, = a + p iL lA M , + 7 (L )A E , + 5 (L )A Z , + e, , in w h ich c a s e th e erro r term in eq u atio n 2 is u, = 8(L)AZ, + £,. Fu rth erm o re, estim ates o f th e m o n etary a n d fiscal p o licy re sp o n ses from eq u atio n 2 w ill be b ia sed if AZ, is co rrela ted w ith AM, o r AE,. T h is criticism o f th e A-J eq u atio n , w hile poten tially d am aging if valid, ap p lies equally w ell to virtually all 4lt now appears that these differences resulted from differences in programming or in the imposition of polynomial restrictions. Batten and Thornton (1985) have replicated the A-J results to the second or third decimal place. Even though other researchers may have been unable to replicate the A-J results exactly, their studies generally supported the qualitative findings of A-J. 5The original A-J paper also contained specifications with the ad justed monetary base as the indicator of monetary policy actions and a distributed lag of high-employment government revenues as an additional right-hand-side variable. Equation 1 is the most com monly estimated form of the equation, however. Furthermore, following an exchange between Friedman (1977) and Carlson (1978), the equation was specified in growth rates of the variables. It is interesting to note that A-J also estimated a growth-rate specification, but only reported the first-difference results. For the most part, the issues discussed below are indepen dent of the specification. 6The notation used here is the same as employed by McCallum (1986). Digitized for10 FRASER OCTOBER 1986 applied e co n o m etric re sea rch , in clu d in g m o st largescale, sim u lta n eo u s-eq u a tio n e c o n o m e tric m o d els o f the A-J vintage.7 M oreover, alth o u g h it w as co m m o n ly argued th at th e A-J eq u a tio n w as poten tially m issp e c i fied, e co n o m etric th eo ry d o es n o t su ggest th a t it is m ore su sce p tib le to th e resu ltin g b ias th a n o th e r esti m ated eq u atio n s. In d eed , th e re w as n o ev id en ce th at th e ir resu lts w ere b iased sin c e n o te sts for m issp ecifica tio n w ere p erform ed . W hile th e ir resu lts provided n o ev id en ce th a t th e A-J eq u atio n is m issp ecified , M odigliani an d A ndo (1976) p re sen ted evid en ce from a M o n te C arlo-style ex p eri m en t th at led som e to d o u b t th e validity o f th e A-J re su lts * Using artificial d ata g en era ted by th e MPS eco n o m etric m odel, th ey u sed a St. L ouis-style eq u a tio n to estim ate th e red u ced -fo rm p a ra m eters. T h e resu lts in d ica ted th at th e St. L ouis-style eq u atio n p ro d u ced p o o r estim a tes o f th e “tru e ” m o n eta ry and fiscal m u ltipliers, serio u sly overstating th e size o f th e m o n etary in flu en ce an d u n d erestim a tin g th e m agn i tu d e o f th e fiscal policy effect. T h ey co n c lu d e d that th e A-J red u ced -fo rm estim a tio n te ch n iq u e y ield ed u nreliab le estim ates. T h is co n clu sio n , how ever, is u nw arranted .'1 If a 7For example, Duesenberry et al. (1965). 8See McCallum (1986), p. 17 and footnote 16. 9McCallum (1986) criticized the Modigliani-Ando results by arguing that they failed to distinguish between reduced-form and “finalform” multipliers. He considers the case where AZ, = a0 + a, AY,_, + a2AM,_, + a3AE,_, + e,. Substituting this expression into equa tion 3 yields the following: AY, = a + 0'(L)AM, + -y'(L)AE, + E; where the coefficients are defined to conform, e.g., 0'(L) = [1 -La,8(L)]“ 1[0(L) + La;,5(L)]. In contrast to the finite order distrib uted lags of the A-J equation 2, the distributed lags on this final-form equation are of an infinite order. Also, the error term of the A-J equation, u„ is hypothesized to be white noise, while that of the above equation, e is an infinite order AR process under the as sumption about u,. The distinction between reduced- and final-form equations may not be important, however, because if the lags of the final-form equation are truncated to match those of equation 2, these equations are indistinguishable save the error structure. While this difference will allow one to distinguish between the two equa tions, it will only do so if one is willing to make strong claims about the underlying distribution of u,. (McCallum notes this; see p. 24, footnote 8). It is interesting to note, however, that our results obtained by adding a distributed lag of AY to the A-J equation support McCallum’s idea that the A-J results reflect ail of the direct and indirect effects via lagged values of nominal GNP. A-J and Darby (1976) argued that the equation captured direct and indirect effects via other contemporaneous endogenous variables. McCallum also argues correctly that “it is hard to imagine any important macroeconomic variable that is truly exogenous . . (p. 13). If there are really no exogenous variables, however, then the true reduced form would be a Sims-type VAR model where the only exogenous variables would be the policy and, perhaps, other innovations. FEDERAL RESERVE BANK OF ST. LOUIS stru ctu ral m odel is well defined w ith additive, n o r m ally d istribu ted errors, co n siste n t estim ates o f the red u ced -fo rm p aram eters can be o btain ed by th e u se o f in d irect least squ ares, a la A -J.l(l B e ca u se th e MPS m od el d oes n ot n ecessarily reflect th e tru e stru ctu re o f th e U.S. eco n o m y (for exam ple, it ignores poten tially im p o rtan t so u rces o f crow ding out throu gh w ealth effects an d R icardian equivalence), th e M odiglianiAndo exp erim en t can n o t b e a criticism o f th e A-J resu lts o r o f th e A-J m eth o d olo gy ." C on sequen tly, th e M odigliani-A ndo evid ence is p red om in an tly a sta te m ent about K eynesian vs. m o n etarist views o f the w o rld .1 Fu rth erm ore, they provide no general infor m atio n co n cern in g th e u sefu ln ess o f th e red u ced form estim atio n . By design, th e A-J eq u atio n did not co n fo rm to th e re d u ce d form o f th e MPS m od el; so it is n o t su rp risin g th at th e p aram e te r estim ates w ere poor. T h e exp erim en t m erely rem in d s u s that, if o n e estim ates a red u ced form th at is know n to be m issp ecified, th e resu lts m ay be biased . Gordon’s Evidence E xcep t for th e u sual ch e ck s for serial co rrelatio n an d h etero sk ed asticity , th e A-J eq u atio n w as n o t su b je c te d to form al te sts o f m od el sp ecificatio n . G ordon (1976) cam e clo sest to testin g th e A-J eq u atio n for m issp ecificatio n . He ad ded a set o f “o m itted variables,” Z, to th e St. Louis eq u ation. C laim ing that th e se variables w ere n o n sto ch a stic, h e te ste d for th e ir statistical sig n ifica n ce and m easu red th e im p act o f th e se variables on th e A-J eq u atio n sim ply by observing w h eth er th ey affected th e size and statistical sig n ifican ce o f th e estim ated long-run m o n eta iy and fiscal m u ltipliers. Unfortunately, the Z-variable h e co n stru c te d — the sum o f n et exports, co n su m e r ex p en d itu res on new au to m o biles and n o n resid en tial fixed investm ent — w as arguably m o re en d o g en o u s th a n th e m o n ey an d '“Unique estimates of the structural parameters cannot be obtained, however, unless the system is exactly identified. "Klein (1976), p. 50, noted in his discussion of the Modigliani-Ando paper, “If the world were constructed along lines portrayed by the MPS model, St. Louis conclusions could have been innocently obtained by one who did not bother to estimate the structure. This is the strongest statement that can be made.” 12Gordon (1976) chides Schwartz (1976) for missing the point of the Modigliani-Ando critique because she criticizes the specification of the MPS model. But this is exactly the point. Gordon later states incorrectly that “the major contribution of the paper is its demonstra tion that the correlation between included policy variables and other excluded variables severely biases the estimated St. Louis multipli ers and renders useless the reduced form technique” (p. 60). OCTOBER 1986 ex p en d itu re variables th at A-J h ad u sed . H ence, Gor d o n ’s results, w hile by an d large favorable to A-J, say little abou t w h eth er A -J's resu lts w ere affected by sp ecificatio n erro r.1 3 R E S E T T e st R e su lts Ideally, o n e sh o u ld test th e sp e cifica tio n o f a m odel by co m p arin g it w ith a w ell-sp ecified alternative. Sin ce the red u ced form o f th e MPS m o d el (or an y o th e r large-scale K eyn esian m odel) is w ell sp ecified , it cou ld , in prin cip le, be u sed as th e alternative in a test o f the A-J eq u atio n . U nfortunately, m o st large-scale m od els have to o m any ex o g en o u s variables for th e red u ced form to b e estim a ted d irectly. Even if it cou ld b e estim ated directly, how ever, it w ou ld b e difficult to o btain a data set th at is co m p arab ly d ated w ith the original A-J data. T h is h a s p ro m p ted u s to u se a g en eral te st o f m issp ecification , th e RESET test o f R am sey an d Sch m id t (1976), w h ich requ ires n o ad d ition al data. T h e RESET test is a gen eral d iag n o stic test for various types o f m issp ecificatio n s, in clu d in g om itted variables, w h ere the alternative h y p oth esis is n o t w ell sp e cified .1 A pplied 4 to eq u ation 2, th e F -sta tistic ca lcu la ted a cco rd in g to th e R am sey-Sch m id t version o f th e R ESET test is .52, w h ich is n o t sign ifican t at th e 5 p e rce n t level.1 H ence, 3 the RESET test provides n o su p p o rt for claim s th at the original A-J eq u ation w as m issp ecified b ec a u se A-J h ad om itted sign ifican t ex o g en o u s variables from th e ir analysis. 13Gordon performed no formal tests. He noted merely that, when his Z-variable was included, the sum of coefficients on AM became smaller and, during one short period, was insignificant. (This period is the one for which the correlation between AM, and AM,_, and his composite variable is the highest.) There was no discussion, how ever, of the problem of multicollinearity or possible bias induced by including variables that are clearly endogenous. (If these extrane ous variables do not belong in the model, the estimates are consist ent but may be biased in small samples.) ,4ln general, if an equation is misspecified, the residuals will have a non-zero mean. The RESET test is designed to detect a non-zero mean of the residuals. The test is performed by adding AY2, AY3, ..., AYhas additional regressors to equation 2 and testing the hypothe ses that these regressors have no joint effect on the dependent variable. The test here was performed for h = 2, 3, 4; the result with the lowest significance level (in this case, h = 3) is reported. See Fomby, Hill and Johnson (1984), pp. 411-12, for a discussion of the RESET test. 15When A-J originally estimated equation 2, they used restricted least squares in the form of Almon’s (1965) polynomial distributed lag estimation technique. We have recently shown, however, that none of the important conclusions of A-J depend on these restrictions [Batten and Thornton (1985)]. Consequently, all of the empirical results reported here are obtained with OLS. 11 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 SIMULTANEOUS EQUATION BIAS A n u m b er o f critics argued th at th e A-J resu lts w ere u n reliab le b e c a u se th e ir p o licy variables w ere not strictly exo g en o u s. B e ca u se o f th e ir know led ge o f th e issu es su rro u n d in g targets and in d icato rs o f m o n eta iy policy, A -J w ere acu tely aw are o f th e n ee d to select ex o g en o u s in d icato rs o f policy. In d eed , they c o n sid ered a b ro ad ran ge o f m e asu res o f m o n eta iy a n d fiscal a ctio n s th at had b ee n cite d frequ en tly in th e litera ture."* In th e ir analysis, th ey a ssu m ed th at all ex clu d ed variables eith e r w ere in d e p en d e n t o f m o n eta iy an d fiscal a ctio n s o r w ere in flu en ced by th em , so th at m o n e ta iy an d fiscal p o licies ex erted an in d irect effect o n th e eco n o m y th rou gh th e se fa cto rs.'7 A-J reaso n ed th at if m o n eta iy an d fiscal in flu en ces w ere n o t in d e p e n d e n t o f o th e r factors, th e c o n sta n t term , w h ich th ey argued su m m arized th e im p act o f th e se factors, w ou ld have ch an g ed as th e se variables ch an g ed . Using a C how test to test w h eth er th e p aram eters o f th e ir eq u atio n w ere tem p orally stable, th ey fou nd n o evi d e n c e o f instability. Given th e a tten tio n th at A-J gave to th is issu e, it is odd th a t th e ir w ork w as singled ou t as s u b je c t to sim u ltan eo u s eq u atio n bias, w h en a n u m b er o f w orks o f a p p lied e c o n o m ics o f th is vintage w ere n o t criti ciz ed for applying OLS to eq u atio n s w ith righ t-h an d sid e variables th a t w ere m o re clearly en d o g en o u s."1 Wu Test Results Again, d esp ite claim s th at th e A-J resu lts w ere q u e s tio n able on g ro u n d s o f sim u ltaneity, sy stem atic te s t ing for sim u ltan eo u s eq u atio n bias h as b ee n sparse. M cC allum (1986) co m p ared OLS an d in stru m en tal variables (IV) estim ates o f th e A-J eq u atio n , bu t p e r form ed n o form al te sts. E xten d in g M cC allu m ’s an aly sis, w e perform a W u (1973) test u sin g th e original A-J data. Like M cC allum , w e u sed th ree lags o f AM, AE an d th e th ree-m o n th T re a su iy bill rate as in stru m en ts for AM, an d AE,. T h e resu lts are rep o rted in table 1. 16Both Andersen and Jordan participated in a Conference on Indica tors and Targets of Monetary Policy held at UCLA in 1966. Andersen contributed to the conference proceedings; see Andersen (1969). 17This possibility was also considered by McCallum (1986) and Darby (1976), though McCallum included a lagged dependent variable to obtain his distinction between the reduced form and the final form; see footnote 9 above. ,80ne of the most important of these was Chow’s (1966) pathbreaking work on money demand, in which current values of real GNP and a nominal interest rate appeared on the right-hand side of the equation. 12 Table 1 OLS and IV Estimates of the Andersen-Jordan Equation Variable OLS Constant 2.311* (2.82) AM, 2.121* (2.87) AM, , 0.312 (0.32) am ,_2 2.696* (2.69) AM, 3 0.671 (0.87) SAM, 5.800* (7.34) 0.379 AE, (1.40) AE,_, 0.523 (188) ae ,_2 0.022 (0.08) -0.763* AE,_3 (2.95) 2AE, 0.161 (0.52) Joint F-test, AM 15.84* 3.17* Joint F-test, AE R2 0.61 DW 1.747 SE 3.96 IV 2.548* (2.45) 0.676 (0.33) 1.652 (0.84) 2.005 (1.56) 0.452 (0.51) 4.785* (3.68) 1.300 (1.46) 0.315 (0.81) -0.217 (0.54) -0.832* (2.81) 0.566 (1.17) 8.65* 2.75* 0.54 2.010 4.42 'Indicates statistical significance at the 5 percent level. Absolute value of t-ratio in parentheses A co m p a riso n o f OLS an d IV estim a tes sh ow s som e large d ifferences, p articu larly for th e co efficien ts on AM, an d AE,. T h e IV estim a tes sh o w a sm a lle r initial effect o f m o n ey an d a larger initial effect o f govern m ent ex p en d itu res relative to th e OLS estim a tes. Nev erth eless, th e Wu test ch i-sq u a re sta tistic is .20, not statistically sign ifican t at th e 5 p e rce n t level. It is n ot too su rp risin g th at th e IV estim a tes are relatively im p recise. T h e first-stage R-s w ere .54 an d .38 for AM, an d AE,, respectively. M oreover, th e fact th at th ree lags o f AM, an d AE, are u sed as in stru m en ts m ean s th at AM, an d AE, are likely to b e highly c o rre lated w ith th e o th e r reg resso rs o f th e A-J eq u atio n . W hile the test co u ld b e ca rried out w ith alternative in stru m en ts, th ere is n o obvious g uide to th e ir s e le c tion. In any event, it is unlikely th at th e resu lts will be FEDERAL RESERVE BANK OF ST. LOUIS Table 2 Estimates of an Autoregressive Version of the Andersen-Jordan Equation Coefficient Variable t-ratio’ Constant 2.464* (2.58) 2.049* AM, (2.84) AM,., -0.206 (0.21) 2.971* am ,_2 (2.98) AM, _3 0.399 (0.45) 2AM, 5.213* (5.11) 0.277 AE, (0.97) AE,_, 0.638* (2.27) AE, 2 0.025 (0.09) AE, 3 -0.709* (2.80) SAE, 0.231 (0.66) AY, , 0.250 (1.86) -0.194 AY, 2 (1.52) A Y ,, -0.030 (0.26) SAY, 0.026 (0.15) Joint F-test, AY = 1.91 Joint F-test, AM = 9.59* Joint F-test, AE = 3.00* R2 = 0.628 DW = 2.146 SE = 3.87 'Absolute value of f-ratio 'Indicates significance at the 5 percent level con v in cing u n less they are ro bu st over a b ro ad ch o ice o f in stru m en ts. It can only be said that, based on th e in stru m en ts used, th ere is n o evid ence o f sim u lta n e o us eq u ation bias in th e A-J eq uation. Granger Causality Results T h ere is ad ditional evid en ce th at th e A-J results w ere n o t affected by sim u ltan eo u s eq u atio n bias. Sim ultaneity requ ires tem p o ral feed back b etw een m o n ey and in co m e. T hu s, th e lack o f G ranger (1969) cau sality from in co m e to m oney is a n ecessa iy , th ou gh n o t sufficient, co n d itio n for statistical exogen eity .l!l W hen Elliott (1975) perfo rm ed tests o f G ranger cau sality betw een m oney, in co m e and governm ent exp en d itu res, he found u n id irectio n al cau sality ru n 19See Wu (1983) for a discussion of these issues. OCTOBER 1986 nin g from m on ey to in co m e an d b id irectio n a l ca u sa l ity b etw een ex p en d itu res an d in co m e.2 M ore recently, " using th e original A-J data, B atten an d T h o rn to n (1985) found u nid irectio n al cau sality ru n n in g from m o n ey to in co m e an d no cau sal ord erin g b etw een in co m e and exp en d itu res. T h e fact th at in co m e d o es n o t G ran ger-cau se m on ey im plies that the co efficien ts on th e d istribu ted lag o f AM, do n ot reflect th e feed back o f in co m e on itself via m oney; in stead , th e se co efficien ts m easu re th e direct, an d po ssib ly in d irect, effects o f m o n ey on th e eco n om y. T o verify th is in terp retation , a th reeq u arter d istribu ted lag o f AY w as in clu d ed in th e A-J eq u ation as sep arate regresso rs an d th e sig n ifican ce of th e se co efficien ts w as tested . T h e resu lts are rep o rted in table 2. T h e co efficien ts on th e lags o f th e d e p e n dent variable are not significantly different from zero — individually or jointly. F u rth erm o re, th e co e f ficien ts on th e m oney an d ex p en d itu re variables differ little from th e OLS resu lts o f table 1. The Sims Evidence A lthough h is criticism w as n o t d irected exp licitly at th e A-J equation, Sim s (1980,1982) h as argued recen tly that th e im p act o f m o n etary p o licy a ctio n s is very sm all if in terest rates are in clu d ed in the sam e eq u a tion .3 T o investigate S im s’ co n je c tu re , w e ad d ed a 1 co n tem p o ra n eo u s an d th ree-q u a rte r d istrib u ted lag o f th e ch an g e in th e th ree-m o n th T reasu ry bill rate (ATB) to th e A-J e q u a tio n .- T h e results, rep o rted in table 3, show that only the co n tem p o ra n e o u s co efficien t on ATB is significant. M oreover, th e co efficien ts on th e m o n ey an d ex p en d itu re variables are little ch a n g ed from th o se in table 1, an d n o n e o f th e qualitative co n clu sio n s abou t th e effectiv en ess o f m o n eta iy or fiscal p o licy a ctio n s is altered . T h u s, as w as th e ca s e for allegation s o f m issp ecifica tion, th ere is co n sid era b le d isp arity b etw ee n th e c o n ventional w isdom an d th e em p irical resu lts c o n c e rn ing th e issu e o f sim ultan eity. N evertheless, th e claim th at sim u ltan eity is a serio u s p rob lem for th e A-J eq u ation is a d eeply en tre n c h e d an d w idely a ccep ted “ Elliott used Sims’ (1972) procedure which requires that the data be filtered, a process that can affect the test results. See Feige and Pearce (1979). 2,McCallum (1983, 1986) has critiqued Sims’ results on theoretical grounds. “ The equation was also estimated with the level of the Treasury bill rate; however, none of the qualitative conclusions were changed. 13 FEDERAL RESERVE BANK OF ST. LOUIS criticism o f th e ir w ork.23 T h e evid en ce ex am in ed in th is sectio n , how ever, su ggests that estim atio n o f the original A-J eq u atio n w as n o t affected by sim ultaneity. INAPPROPRIATE INDICATORS O F PO LICY ACTIONS A third m ajo r criticism o f th e St. Louis eq u atio n w as th at A -J’s in d icators o f p o licy actio n s m ay be in ap p ro priate. Failu re to u se ap p rop riate in d icato rs co u ld bias th e estim ated param eter's, p erh ap s by d istorting the relative im p o rtan ce o f m o n eta iy an d fiscal action s.-4 In a sen se, th is arg u m ent is an ex ten sio n o f the policy en d o g en eity argum ent sin ce its p ro p o n en ts co n ten d ed th at th e ap p rop riate in d icato r o f m on etary p o licy sh o u ld n o t resp o n d en d o g en o u sly to forces o u tsid e o f th e F e d ’s co n tro l. F or exam ple, in th e first p u b lish ed criticism o f A-J, de Leeuw an d K alch b ren n e r (19G9) criticiz ed th e u se o f th e m o n eta iy b ase (and im plicitly M l) as an in d ica to r o f m o n eta iy policy a ctio n s on th e g rou n d s th at so m e o f its co m p o n e n ts (particularly, cu rre n cy an d borrow ed reserves) w ere en d o g en o u s an d n o t co n tro lled by th e F ed d irectly In stead , de Leeuw and K alch b ren n er offered an alter native exo g en o u s policy m easu re th at th ey o btain ed by su b tractin g cu rre n cy an d borrow ings from th e a d ju sted m o n eta iy base. W hen they estim ated an A-J type eq u atio n u sing th e ir m easu re o f m o n eta iy policy actio n s, th ey fou nd th e cu m u lative m o n e ta iy policy m u ltip lier w as m u ch sm aller th a n th a t o f th e A-J eq u atio n an d n ot significantly d ifferent from zero. On th e o th e r hand , th e ir estim ated cu m u lative govern m e n t sp en d in g m u ltip lier w as su bstan tially larger and w as statistically significant.2’ 1 In th e ir reply, A-J (1969) p o in ted o u t th a t d e Leeuw an d K a lch b ren n er’s fo cu s o n th e u ses o f th e m o n eta iy b a se w as in ap p ro p riate. A lthough th e banks an d the p u b lic d eterm in e th e u ses o f th e base, th e Fed co n tro ls th e size o f th e m o n eta iy b a se th rou gh its in flu en ce over th e s o u rc e s o f th e base, th e largest co m p o n e n t of ?3While Andersen and Jordan acknowledged that money could be endogenously related to income and expenditure variables via a “ Fed reaction function,” they considered this to be of little practical significance. See Andersen and Jordan (1969), p. 16. 24For some, the concern was that some of the effect of fiscal policy might be incorrectly attributed to monetary policy. See Blinder and Solow (1974). 25This line of argument was also taken by Gramlich (1971). ^Government receipts were also included; the estimated cumulative multiplier of government receipts also increased but was statistically significant only with longer lags. 14 OCTOBER 1986 Table 3 Estimates of the Andersen-Jordan Equation with a Distributed Lag of Interest Rates Lags AM ATB 0 2.409* 4.216* (3.07) (2.37) 1 -0.633 0.122 (0.61) (0.06) 2 2.124 0.199 (2.00) (0.10) 3 0.737 -Q122 (0.79) (0.07) Sum 4.637* 4.415 (4.95) (1.39) Constant 2.910* (3.47) Joint F-test, AM = 7.65* Joint F-test, ATB = 1.96 Joint F-test, AE =3.10* R2 = .635 DW = 1.78 SE = 3.83 AE 0.313 (1.17) 0.639* (2.32) 0.002 (0.01) -Q666* (2.47) 0.228 (0.94) 'Indicates statistical significance at the 5 percent level Absolute value of the t-ratio in parentheses w h ich is th e F e d ’s h old in g s o f U.S. gov ern m en t se c u ri ties. T h u s, th e F ed d eterm in es th e size o f th e m o n eta iy b ase throu gh its sales o r p u rch a se s o f governm ent secu rities. Fu rth erm ore, A-J n o ted th at ch a n g es in th e M l m o n ey stock d uring th e ir estim a tio n period w ere d o m in ated by ch a n g es in th e m o n etary b a se. H ence, th e Fed ex e rcised co n tro l over M l th rou g h its co n tro l o f th e so u rces o f th e m o n etary b ase. S in ce th is e x ch an ge, th e d isag reem en t over th e m e a su re m en t of m o n etary p o licy a ctio n s h a s su b sid ed , a n d th e m o n e tary b a se an d M l (and, at tim es, b ro a d e r m o n e ta iy aggregates) are gen erally a ccep ted , an d co m m o n ly u sed , as in d ica to rs o f p o licy actio n s. A -J’s m ea su rem en t o f fiscal p o licy a ctio n s w as criti cized m ore th an th e ir m easu re o f m o n e ta iy policy a ctio n s. R ecogn izin g th at ce rta in co m p o n e n ts o f b o th fed eral govern m en t ex p en d itu res an d reven u es re sp o n d en d o g en o u sly to th e level o f e c o n o m ic activity, A-J u tilized h ig h -em p loy m en t m easu res, w h ich w ere a d ju sted for th e se in flu en ces. De L eeuw a n d K alch - FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 b re n n e r co n ten d ed that th is a d ju stm en t w as in c o m plete b eca u se it failed to elim in ate th e in flu en ce of inflation. T h e su b stitu tio n o f in flatio n -ad ju sted , highem p loy m en t govern m ent ex p en d itu res an d revenues, how ever, h ad little im p act on th e estim ated p ara m e ters o f th e eq uation. Subsequently, how ever, Sch m id t an d W aud (1973) found th at C orrigan’s resu lts d ep en d ed critically on th e polynom ial re strictio n s h e im p o sed .2 W h en th ese " restrictio n s, w h ich a p p eared to b e re je cte d by th e data, w ere relaxed, Sch m id t an d W aud obtain ed resu lts w ith th e IS m easu res th at w ere sim ilar to A -J’s. G ram lich (1971) felt th at the n o n -m o n e ta iy "exo g e n o u s ” in flu en ces w ere too narrow ly d efined . C o n se quently, h e co n stru cte d two b ro ad er co m p o site m ea su res. His ex p en d itu re m easu re w as governm ent p u rch ase s plus exports, g ran ts-in -aid and an invento iy a d ju stm en t for d efen se p u rch a se s. His revenue m easu re in clu d ed hig h -em p loy m ent p erso n al taxes plus in terest pay m en ts and social secu rity co n trib u tio n s le ss exo g en o u s tran sfers (that is, all transfers excep t u n em p loy m en t co m p en satio n ). W hile th ese ch a n g es did result in larger (and m ore nearly sta tisti cally significant) su m s o f estim ated co efficien ts for the n o n -m o n e ta iy in flu en ces, th e gen eral resu lts o f A-J rem ain ed in tact. T h e evid ence suggests th at A -J’s resu lts co n cern in g th e effect o f fiscal p o licy w ere n o t critically d ep e n d en t on th e ir m ea su rem en t o f m o n etary o r fiscal policy a ctio n s. M eyer an d R a sch e (1980) su m m arized th eir investigation o f th is issu e by n o tin g that, “th e m od i fication s suggested . . . have n o t g en erally resu lted in d ram atic ch an g es in th e estim a ted m u ltip liers in sim ple red u ced -fo rm eq u a tio n s.”2 ” C orrigan (1970) offered w hat ap p eared to be th e m ost dam aging criticism o f th e h igh -em ploym ent m easu res o f fiscal policy actio n s. He argued th at they did n ot rep resen t ap p rop riate in d icato rs o f d isc re tionary fiscal policy actio n s, sin c e hig h -em p loy m en t m easu res (especially revenues) w ou ld ch an g e w ith h ig h -em p loy m en t in co m e. In th e ir p lace, h e offered h is initial stim u lu s (IS) m easu re o f d iscretio n a iy ch a n g es in fiscal policy. T h e IS m easu re o f governm ent ex p en d itu res did not differ significantly from th e highem p lo y m en t m easu re. T h e IS m easu re o f revenu es, on th e o th e r han d , differed co n sid erab ly from its high em p lo y m en t co u n terp art. In p articu lar, th e IS m ea su re o f a ch an g e in governm ent revenu es w as n o n z ero only in qu arters in w h ich a tax w as in tro d u ced , m o d i fied, su sp en d ed o r elim in ated . W h en IS m easu res w ere su b stitu ted for highem p loy m ent m easu res in an A-J type eq u ation , th e resu lts w ere startling: the estim ated cu m u lative im pact o f ch an g es in M l d eclin ed , w hile th o se o f b o th ch an g es in governm ent ex p en d itu res and o f ch an g es in governm ent reven u es ro se significantly and, m ore im portantly, w ere ap p aren tly statistically sign ifican t.2 7 T h u s, Corrigan co n clu d e d th at fiscal p o licy actio n s had a m eaningful im p act on n om in al e co n o m ic activity. 27Corrigan did not report t-statistics or standard errors for the summed coefficients. Assuming that the estimated coefficients are uncorre lated, one obtains a t-statistic of 3.01 for testing the hypothesis that the XE = 0 and a t-statistic of 9.46 for testing that XT = 0. Both of those are statistically significant at the 5 percent level. INAPPROPRIATE RESTRICTIO N S To estim ate th e ir d yn am ic sp ecificatio n , A-J u sed A lm on’s (1965) polyn om ial d istrib u ted lag estim atio n te ch n iq u e th at w as d esign ed to im prove th e p recisio n o f th e estim ated p aram eters o f a d istribu ted -lag m odel. T h e te ch n iq u e co n stra in s th e p aram eters of e a ch d istribu ted lag to lie on a polynom ial o f a given d egree. P erh ap s b ec a u se relatively little w as know n abou t th e p ro ced u re w h en A-J p u b lish ed th e ir paper, critics co n ten d ed th at th e A-J resu lts m igh t b e d ep e n d en t u pon, o r at least sensitive to, th e ir c h o ic e o f lag length or polynom ial d egree."1 T h ere have b ee n relatively few investigations o f th is a sp e ct o f th e A-J eq u atio n . T h e best-kn o w n stu d y by Sch m id t an d W aud (1973), as w ell as o th ers by C orri gan, de Leeuw and K a lch b ren n er, an d Silber, focu sed prim arily on th e selectio n o f th e lag length. B eca u se th e se stud ies h eld th e polyn om ial degree fixed, how - 28The restrictions forced the estimated parameters of each distributed lag to lie on a second degree polynomial. ^Meyer and Rasche (1980), p. 59. McCallum (1986), p. 14, simply notes that “if there is a fiscal policy measure that carries a strongly significant sum of coefficients in an equation of the St. Louis form, its existence has not been well publicized.” “ Specifically, if the lag length is too long or the polynomial degree too high, estimated parameters are unbiased but inefficient. Alterna tively, if the lag length is too short or the polynomial degree is too low, the estimates are biased. Therefore, it is important that the appropriate lag length and polynomial degree be determined. The parameters will also be biased if the chosen lag is too long and exceeds the true lag by more than the true polynomial degree and may be biased even if it exceeds the true lag by an amount less than or equal to the true polynomial degree. See Batten and Thornton (1983) for a discussion of this and other issues, and for other references. 15 FEDERAL RESERVE BANK OF ST. LOUIS ever, th ey did n o t analyze co m p letely th e restrictio n s im p o sed by th e A-J sp ecificatio n .3 1 W h en Elliot (1975) exam in ed th e lag stru ctu re and th e polyn om ial restrictio n s separately, h e co n clu d e d th at A-J resu lts w ere n o t p articu larly sensitive to lag stru ctu re o r to th e polynom ial restrictio n s. His c o n clu sion, how ever, w as n ot b ased on statistical tests. He m erely co m p ared p aram e te r estim ates for different lag stru ctu res an d polynom ial degrees. M ore recently, B atten an d T h o rn to n (1983) perform ed a system atic exam in ation o f th e sp ecificatio n o f the A-J equation u sin g re ce n t data, an d B atten an d T h o rn to n (1985) perform ed a sim ilar analysis u sin g th e original A-J data. T h ey co n clu d e d th at th e p olicy-relevant resu lts of A-J do not d ep en d on th e ir c h o ice o f lag length or polynom ial degree. SUMMARY AND CONCLUSIONS L eonall C. A n d ersen ’s b est k now n an d m o st signifi ca n t co n trib u tio n to e c o n o m ics is h is collaborative re sea rch w ith Jerry L. Jo rd an , w h ich resu lted in p u bli ca tio n o f th e A-J eq u atio n . F o r a perio d o f nearly 20 y ears, it h as b e e n th e su b je ct o f m u ch in tere st an d co n sid erab le critic ism .1 Few o th e r p ie ce s o f applied eco n o m ics, if any, have b een so th orou ghly d iscu ssed , analyzed an d investigated. OCTOBER 1986 u nknow n o r unavailable w h en A n d ersen an d Jo rd a n 's critics w ere m o st vocal. F u rth erm o re, so m e o f the criticism s are valid w h en ap p lied to sam p le p erio d s b eyond that exam in ed by A n d ersen an d Jo rd a n .*1 T h e se facts n otw ith stan d in g, th is review v in d icates A n d ersen an d Jo rd a n o f an y serio u s b re a c h o f th e stan d ard s o f e c o n o m e tric p ra c tice an d su ggests that, in reality, it w as n ot th e ir a p p lica tio n o f e co n o m etric m eth o d s th at w as co n troversial, b u t th e ir resu lts.34 A n d ersen an d Jo rd a n sh o u ld b e co n g ra tu la ted for providing o n e o f th e m o st stable, lastin g an d robu st eq u atio n s in ap p lied e c o n o m ics. In o u r o p in ion , h o w ever, th e ir m o st im p o rtan t co n trib u tio n is th at they sh ook th e fou n d ation s o f co n v en tion al ec o n o m ic thou ght an d su b je cte d th e resu lts o f stan d ard applied ec o n o m ics to c lo se r scru tin y. T h is fo rced e co n o m ists an d p olicym akers to take a c lo se r look at th e issu e of th e efficacy o f m o n eta iy an d fiscal policy. R EFE R E N C E S O ur review o f th e original A n d ersen -Jo rd an study an d th e criticism th at em erg ed follow ing its p u b lica tion p o in ts out th e obvious, bu t seld om articu lated , fact th at all o f th e criticism s o f A nd ersen an d Jo r d a n ’s w ork apply equally w ell to m u ch o f th e ap p lied e c o n o m ic re se a rc h o f th at tim e, an d even today. W e also n o te th at A n d ersen and Jo rd a n w ere aw are o f m an y of th e caveats o f th e ir w ork an d took p re ca u tio n s against th em . M ost im portantly, u sin g th e ir original data, w e te ste d th e A n d ersen -Jo rd an eq u atio n for m issp ecificatio n an d sim u ltan eo u s-eq u atio n bias. W e find th at n o n e o f th e o ft-cited criticism s o f th e ir eq u atio n is (or co u ld have been) su b stan tiated by th e se statistical tests. G ranted, som e o f th e te ch n iq u e s u sed w ere Almon, Shirley. “The Distributed Lag Between Capital Appropria tions and Expenditures,” Econometrica (January 1965), pp. 17896. Andersen, Leonall C. “Money Market Conditions as a Guide for Monetary Management,’’ in Karl Brunner, ed., Targets and Indica tors of Monetary Policy (Chandler, 1969). Andersen, Leonall C., and Jerry L. Jordan. “Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabili zation— Reply,” this Review (April 1969), pp. 12-16. _________ “Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization,” this Review (November 1968), pp. 11-24. Batten, Dallas S., and Daniel L. Thornton. “The Andersen-Jordan Equation Revisited,” Journal of Macroeconomics (Summer 1985), pp. 419-32. ________ _ “ Polynomial Distributed Lags and the Estimation of the St. Louis Equation,” this Review (April 1983), pp. 13-25. Blinder, Alan S., and Robert M. Solow. “Analytical Foundations of Fiscal Policy,” in The Economics of Public Finance (Brookings Institution, 1974), pp. 3-115. Carlson, Keith M. “Does the St. Louis Equation Now Believe in Fiscal Policy?” this Review (February 1978), pp. 13-19. 31After the polynomial degree has been chosen, alternative lag speci fications amount to imposing polynomial restrictions on different parameter spaces. Consequently, the restrictions implied by differ ent lag specifications are not nested within each other when the polynomial degree is fixed. 32One of the most recent additions to this literature, Raj and Siklos (1986), applies spectral analysis to the Andersen-Jordan equation for the period 1/1947 to IV/1984. Again the results are consistent with those of A-J. :liFor example, Thornton and Batten (1985) find bidirectional Granger causality between money and income over the period from 11/1962 to 111/1982. 14It is interesting to note the similarities between the A-J equation and Granger’s (1969) and Sims’ (1972) examination of causal ordering. Furthermore, except for the exclusion of the own distributed lag of the dependent variable, the A-J equation closely resembles the now frequently used and commonly accepted vector autoregression model. 16 FEDERAL RESERVE BANK OF ST. LOUIS Chow, Gregory C. “On the Long-Run and Short-Run Demand For Money,” Journal of Political Economy (April 1966), pp. 111-31. Cooley, Thomas F., and Stephen F. LeRoy. “ Identification and Estimation of Money Demand,” American Economic Review (De cember 1981), pp. 825-44. Corrigan, E. Gerald. “The Measurement and Importance of Fiscal Policy Changes,” Federal Reserve Bank of New York Monthly flewew (June 1970), pp. 133— 45. Darby, Michael R. “Comments on Modigliani and Ando,” in Jerome L. Stein, ed., Monetarism, vol. 1, Studies in Monetary Economics (North-Holland, 1976), pp. 67-68. Davis, Richard G. “How Much Does Money Matter? A Look at Some Recent Evidence,” Federal Reserve Bank of New York Monthly Review (June 1969), pp. 119-31. de Leeuw, Frank, and John Kalchbrenner. “Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabili zation — Comment,” this Review (April 1969), pp. 6-11. Duesenberry, J. S., G. Fromm, L. R. Klein, and E. Kuh, eds. The Brookings Quarterly Econometric Model of the United States (RandMcNally, 1965). Elliott, J. W. “The Influence of Monetary and Fiscal Actions on Total Spending,” Journal of Money, Credit, and Banking (May 1975), pp. 181-92. Feige, Edgar L., and Douglas K. Pearce. “The Casual Causal Relationship Between Money and Income: Some Caveats for Time Series Analysis,” Review of Economics and Statistics (No vember 1979), pp. 521-33. Friedman, Benjamin M. “ Even the St. Louis Model Now Believes in Fiscal Policy,” Journal of Money, Credit, and Banking (May 1977), pp. 365-67. Friedman, Milton, and David Meiselman. “The Relative Stability of Monetary Velocity and the Investment Multiplier in the United States, 1897-1958,” in The Commission on Money and Credit, Stabilization Policies (Prentice-Hall, 1963), pp. 165-268. Fomby, Thomas B., R. Carter Hill, and Stanley R. Johnson. Ad vanced Econometric Methods (Springer-Verlag, 1984). Gordon, Robert J. “Comments on Modigliani and Ando," in Jerome L. Stein, ed., Monetarism, vol. 1, Studies in Monetary Economics (North-Holland, 1976), pp. 52-66. Gramlich, Edward M. “The Usefulness of Monetary and Fiscal Policy as Discretionary Stabilization Tools,” Journal of Money, Credit, and Banking (May 1971), pp. 506-32. Granger, Clive W. J. “Investigating Causal Relations by Econo metric Models and Cross-Spectral Methods,” Econometrica (July 1969), pp. 424-38. Klein, L. R. “Comments on Modigliani and Ando,” in Jerome L. Stein, ed., Monetarism, vol. 1, Studies in Monetary Economics (North-Holland, 1976), pp. 50-51. OCTOBER 1986 McCallum, Bennett T. “Monetary Versus Fiscal Policy Effects: A Review of the Debate,” in R. W. Hafer, ed., The Monetary Versus Fiscal Policy Debate: Lessons From Two Decades, Studies in Monetary and Macroeconomic Theory and Policy (Rowman and Allanheld, 1986), pp. 9-29. _________ “A Reconsideration of Sims' Evidence Concerning Monetarism,” Economic Letters (vol. 13, nos. 2-3,1983), pp. 16771. Meyer, Lawrence H., and Robert H. Rasche. “Empirical Evidence on the Effects of Stabilization Policy,” in Stabilization Policies: Lessons from the '70s and Implications for the ’80s (Washington University in St. Louis, Center for the Study of American Business, Working Paper No. 53, 1980), pp. 41-112. Modigliani, Franco, and Albert Ando. “ Impacts of Fiscal Actions on Aggregate Income and the Monetarist Controversy: Theory and Evidence,” in Jerome L. Stein, ed., Monetarism, vol. 1, Studies in Monetary Economics (North-Holland, 1976), pp. 17-42. Raj, Baldev, and Pierre L. Siklos. “The Role of Fiscal Policy in the St. Louis Model: An Evaluation and Some New Evidence,” Journal of Applied Econometrics (1986), pp. 287-94. Ramsey, James B., and Peter Schmidt. “Some Further Results on the Use of OLS and BLUS Residuals in Specification Error Tests,” Journal of the American Statistical Association (June 1976), pp. 389-90. Schmidt, Peter, and Roger N. Waud. “The Almon Lag Technique and the Monetary Versus Fiscal Policy Debate,” Journal of the American Statistical Association (March 1973), pp. 11-19. Schwartz, Anna J. “Comments on Modigliani and Ando,” in Jerome L. Stein, ed., Monetarism, vol. 1, Studies in Monetary Economics (North-Holland, 1976), pp. 43-49. Silber, William L. "The St. Louis Equation: Democratic' and Re publican’ Versions and Other Experiments,” Review of Economics and Statistics (November 1971), pp. 362-67. Sims, Christopher A. “Policy Analysis with Econometric Models,” Brookings Paper on Economic Activity (1:1982), pp. 107-52. ________ _ “Comparison of Interwar and Postwar Business Cy cles: Monetarism Reconsidered,” American Economic Review (May 1980), pp. 250-57. _________ “Money, Income, and Causality,” American Economic Review (September 1972), pp. 540-52. Thornton, Daniel L., and Dallas S. Batten. “ Lag-Length Selection and Tests of Granger Causality Between Money and Income,” Journal of Money, Credit, and Banking (May 1985), pp. 164-78. Wu, De-Min. “Tests of Causality, Predeterminedness and Exoge neity," International Economic Review (October 1983), pp. 547-58. _________ “Alternative Tests of Independence Between Stochas tic Regression and Disturbances,” Econometrica (July 1973), pp. 733-50. 17 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 A M onetarist Model for Econom ic Stabilization: Review and Update Keith M. Carlson I n APRIL 1970, Leonall A nd ersen and I p u blish ed an article, "A M o n etarist M odel for E co n o m ic Stabiliza tio n ," in th is Review.' In this article, w e developed a sm all m od el o f th e U.S. eco n o m y p u rp o rtin g to explain th e m ovem ents o f ce rta in key e c o n o m ic aggregates, nam ely, n o m in al GNP, o u tp u t (real GNP), p rices, u n em p loy m ent and sh o rt- an d long-term in terest rates. T h e m o d el's fo cu s w as on th e role o f m o n etary aggre gates, in particu lar, M l, in th e d eterm in atio n o f th ese e co n o m ic variables. T h e p u rp o se o f th e p re se n t article is to review th is m od el in light o f d evelop m en ts sin c e 1970. T h is review beg ins w ith a d iscu ssio n o f th e d evelop m en t o f th e original m odel an d is follow ed by an exp lan atio n o f th e key d ifferences betw een it an d th e cu rre n t v ersion o f th e m od el. T h is cu rre n t version is analyzed by d em o n stratin g its re sp o n se to sh o ck s an d its ability to sim ulate, e * post, m ovem ents o f n o m in al GNP, output, p rices, u nem p loy m en t an d in tere st rates. OUR EARLY A TTEM PT AT MODELING In 1970, m a cro e co n o m ic m od el-bu ild in g w as a p o p u lar ex ercise. T h e M ichigan an d W h arton m odels, w h ich h ad existed for a n u m b er o f y ears, w ere c o n tin ually bein g m odified an d updated.- T h e FRB-M IT m odel, first p u b lish ed in 1968, w as still bein g refin ed :' T h e D ata R eso u rces m o d el w as in th e d evelopm ent 'Andersen and Carlson (1970). 2See Klein and Burmeister (1976), pp. 188-210 and pp. 248-70. 3de Leeuw and Gramlich (1968 and 1969). 18 stage.4 E a ch o f th e se m o d els co n ta in e d a large n u m b er o f eq u atio n s an d fo cu sed o n a secto ra l breakd ow n of GNP derived from th e K eynesian a p p ro a ch to GNP d eterm in ation. Andy an d I felt th at th e se m o d els did n o t place p ro p er em p h a sis on th e role o f m o n e ta iy a ctio n s. Furth erm ore, th ey focu sed prim arily on th e sh o rt run — a p ro jectio n h o rizo n of, at m ost, several qu arters. W e w an ted a m o d el th a t m oved from th e sh o rt-ru n to a lon g-run d yn am ic eq u ilibriu m w ith a p p rop riate re c ognition bein g given to initial co n d itio n s in this p ro cess. In addition, w e w an ted a m o d el th at w as sm all en ou gh th at th e in terrela tio n sh ip s am o n g th e vari ables co u ld b e u n d ersto o d easily. M oreover, w e sought to bu ild on existin g re sea rch at th is Bank, com bin in g various resu lts to sh ed light on th e issu e of e co n o m ic stab ilization in a w ay th at w ou ld overcom e som e o f th e sh o rtco m in g s o f larg e-scale K eynesianstyle m odels. O ur co n c e rn s abou t th e sta te o f m od el-bu ild in g strongly in flu en ced o u r efforts to develop an a ltern a tive m a cro e co n o m ic m od el o f th e U.S. eco n o m y . We w ere n o t co n c e rn e d abou t resp ecifyin g behavioral eq u atio n s (for exam ple, co n su m p tio n , in vestm en t, etc.); rath er, w e w an ted to ca p tu re em p irica l re la tio n sh ip s b etw een a relatively few key m a c ro e co n o m ic variables th at w ere im p licitly g ro u n d ed in ec o n o m ic theory. T h e fu n d am en tal b u ild in g block o f o u r m o d el w as th e A n d ersen -Jo rd an (A-J) eq u atio n , w h ich fo cu sed on 4Klein and Burmeister (1976), pp. 211-31. FEDERAL RESERVE BANK OF ST. LOUIS th e two ch ie f arm s o f policym aking, m o n eta iy and fiscal actio n s.5 Although th is eq u atio n did not provide a m od el o f GNP d eterm in atio n , it w as useful in fore castin g and in policy sim u lation s. In th e A-J eq uation, GNP w as “d eterm in ed ” solely by cu rren t and past m o n eta iy an d fiscal policy actio n s; o th e r in flu en ces on GNP w ere fou nd to be ran d o m d uring the sam p le period investigated bv A n d ersen -Jo rd an . A n o th er im p o rtan t bu ilding block in th e c o n s tru c tion o f th e A n d ersen -C arlson (A-C) m o d el w as the in te re st rate e q u a tio n , d ev elo p ed by Y oh e an d Karnosky in 1969, in w h ich in terest rates w ere sy stem atically related to past inflation.1 T h e ir resu lts w ere ' co n siste n t w ith th e F ish erian th e o iy o f in tere st in th at they sh o w ed th at inflation prem ia are in co rp o ra ted into n o m in al in tere st rates. To co m p lete th e m odel, w e n eed ed eq u atio n s for the u nem p loy m en t rate and th e p rice level. T h e m ost fam ous and generally a cc e p te d u n em p loy m en t rate eq uation, developed by A rthur Okun, w as easily m o d i fied for o u r p u rp o ses.7 T h is eq u atio n co m b in es a given p o ten tial GNP w ith actu al GNP to provide an estim ate o f th e u nem p loy m en t rate. Finding an ap p rop riate p rice eq u atio n w as a m ore ch allengin g task. M ost large m o d els u sed a w agem arkup eq u atio n and, in so m e cases, som e type o f Phillips curve eq u atio n . T h e se eq u atio n s did not fulfill o u r req u irem en ts. Instead , w e developed a p rice eq u a tio n th at co m b in ed th e Phillips curve resu lts w ith p rice exp ectation s.'1 W e u sed th e co efficien ts on the inflation term s in th e long-term in tere st rate eq u atio n as o u r m easu re o f p rice ex p ectatio n s. We th o u gh t o u r a p p ro ach w as novel, an d it seem ed to w ork qu ite satisfactorily at th e tim e. In retro sp ect, it seem s ru d i m en tary and has not w orked as well in re ce n t y ears. T H E ORIGINAL SPECIFICATION AND SUBSEQUENT CHANGES OCTOBER 1986 eq u a tio n s an d in clu d ed . th e ex o g en o u s variables th at are T h e original an d cu rre n t v ersion s o f th e A-C m od el are su m m arized in table 1. T h e m od el still has the sam e n u m b er o f key en d o g en o u s variables; however, th e th ree GNP variables — total spen din g, o u tp u t and p rices — are now sp ecified in rates o f ch a n g e in stead of first differences. T h is ch a n g e w as m ad e in th e 1970s, w h en th e first-d ifferen ce form beg an to exhibit h etero skedasticity.-’ In any event, th e rate-of-ch an g e form is ea sier to interp ret, an d th e fu n d am en tal p ro p erties of th e m odel are u n ch an g ed . M o n etaiy a ctio n s have a sh o rt-ru n effect on o u tp u t an d a lon g-run effect on p rices; fiscal a ctio n s have little effect on o u tp u t or p rices in eith e r th e sh o rt- o r long-run. A nother ch an g e w as the ad d itio n o f tw o m o re exog en o u s variables — energy p rice s an d exp orts. T his ch an ge, n ecessita ted by d ev elo p m en ts in th e 1970s, w as a cru d e w ay to in co rp o ra te su ch co m p lex fa c tors."’ N evertheless, it en a b led us to keep th e m odel sm all. Fu rth erm ore, ch a n g es in energy p rice s also e n ter th e cu rren t m od el th rou gh th eir in flu en ce on poten tial GNP. A nother change, n ot sh ow n explicitly in table 1, is the red efin ition o f tw o exo g en o u s variables — p o te n tial GNP and federal ex p en d itu res. P oten tial GNP is now estim ated u sin g p ro d u ctio n -fu n ctio n m eth o d s developed by R asch e an d T a to m ." Fed eral ex p en d i tu res are now cyclically a d ju sted ra th er th an highem p lo y m en t.1 T h e ratio n ale u nd erlyin g th e fiscal 2 m easu re rem ain s th e sam e — to co n stru c t a m easu re o f federal sp en d in g th at ex clu d e s th e cy clical effect of the eco n o m y on th e budget. Finally, in th e cu rre n t version, th e p rice, lon g-term in terest rate an d u n em p loy m en t eq u a tio n s are ad ju sted for au to co rrelatio n to avoid b iasin g th e e sti m ated stand ard erro rs o f th e coefficien ts. T h e original m odel w as recursive, w ith th e p articu lar form o f ea ch eq u atio n d eterm in ed , for the m ost part, by th e data. Sin ce 1970, several ch an g es have b een m ade in th e m od el in term s o f th e form o f th e Although th ese ch a n g es m ake it im p o ssib le to co m pare m eaningfully th e su m m ary sta tistics for th e two versions, the two v ersions sh o w sim ilar estim a tes of th e im p a ct o f m o n e ta iy a n d fisca l a ctio n s . An eq u atio n -b y -equ atio n co m p a riso n is su m m arized in th e sh ad ed in sert on page 21. 5Andersen and Jordan (1968). 6Yohe and Karnosky (1969). 70kun (1962). 8See Considine (1969). 9Carlson (1978). 10Rasche and Tatom (1977b). "Rasche and Tatom (1977a). 12de Leeuw and Holloway (1983). 19 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 1 St. Louis Model: Original vs. Current Version Original version Sample period: 1/1953-IV/1969 (1) Total spending equation 4 4 AY, 2.67 2 mAMt_, 2 e,AE,,, i= 0 i 0 = + + = Current version Sample period: 1/1960-1V/1984 (1) Total spending equation 3 Y, = 3.08 - 4.28Z1, + 2 m,M, i= 0 4 -i 3 + 2 e,E,_i 2 ex,EX,,, i= 0 i= 0 2m, .96 2e, .07 2ex, .00 R2 .32 SE 3.91 DW 2.22 (2) Price equation 3 P, .96 2 pe,PE,_, i= 0 5 2 d,D,_i 1.68PA, i= 0 - .96Z2, .80Z3, 2pei .09 2d, = 08 p -.01 R2 = .77 SE = 1.48 DW = 1.98 + 2m, 5.57 2e, .05 R2 .66 SE 3.84 DW 1.75 (2) Price equation 5 AP, = 2.70 + 2 dp,., + .86 APA, i= 0 = = = = = = = = = = = = + + + + 2d, .09 R2 .87 SE 1.07 DW 1.41 (3) Demand pressure identity D, = AY, - (XF, - X,.,) (4) Total spending identity AY, = AP, + AX, (5) Long-term interest rate equation RL, = 1.28 - .06M, + 1 42Z, 16 16 p + 2 X , * ,. , + 2 p ,( ' ) i= 0 i = 0 U '- '/4 2x, = .20 2pi = .96 R2 = .92 SE = .28 DW = .69 (6) Anticipated price definition 17 p A P A ,-Y ,_2([( 2 p ).01 + 1 ]'« -1 ) i = 1 U'-'/4 2p, = .96 = = = = = = (3) Demand pressure identity D, = X, - ([(XF,/X,_,)< - 1]100) (4) Total spending identity Y, = P, + X, (5) Long-term interest rate equation 20 RL, = 5.66 + 2 p,P,_, i= 0 2p, = .55 R2 = .07 p, = 1.15 p2 = -.1 6 SE = .42 DW - 1.96 (6) Anticipated price definition 21 PA, = 2 p,P,_, i= 1 2p, = .55 (7) Unemployment rate equation U, = 3.90 + .04G, + 28G,_, R2 = .92 SE = .30 DW = .60 (8) GNP gap identity XF, - X, G'= XF, -1 0 0 20 (7) Unemployment rate equation U, - UF, = .29G, + 16G,_, p, = 1.20 P2 = 0.33 R2 = .71 SE = .21 DW = 1.95 (8) GNP gap identity XF, - X, G'= XF, -10° FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 The St. Louis Model: Original vs. C urrent V ersions T h e total sp en d in g eq u atio n h as b ee n ch an g ed from a first-difference to a rate-of-ch an g e form , and the rate o f ch an g e o f exp orts and a dum m y variable, d esigned to cap tu re in a cru d e w ay th e velocity shift after 1981, have b ee n ad d ed. T h e lag o n m o n ey w as also sh o rten ed by o ne quarter. In th e cu rren t version, th e p rice eq u atio n h as th ree additional variables and is now in rate-ofch an g e form. Energy p rices are now in clu d ed as an in d ep en d en t variable, and d um m y variables re p re sen tin g p rice co n tro l an d d eco n tro l in th e early 1970s are also inclu d ed . T h e d em an d p ressu re vari able (equation 3) has b een redefined to avoid m ix ing n om in al an d real variables. D em and p ressu re is defined n ow as actu al o u tp u t grow th relative to its poten tial to expand. T h e co efficien t on an ticip ated prices, 1.68, now appeal's out o f line w ith both th eo retical and past em p irical estim ates. However, w h en th e sum for P in the long-term in terest eq u a tion, .55, is taken into acco u n t, th e p ro d u ct o f the two co efficients, (1.68 X .55), is m ore in lin e w ith th eo retical exp ectation . Unlike th e rest o f th e eq u atio n s, th e long-term in terest rate eq u atio n is essen tially u n ch an g ed . Sin ce its sole fu n ctio n is to g en erate th e w eigh ts on past p rices to u se in th e m easu re o f an ticip ated p rices, th e resu lts for th e early version h ad c o n sid erable appeal b ec a u se 2 p , w as clo se to o n e. T his result is n ot observed in th e cu rren t version. Al though th e form o f th e eq u atio n is n ow m u ch sim pler, th e fit o f th e eq uation, w h ich in clu d es an a u to co rrelatio n a d ju stm en t, is veiy poor. T h e good fit o f th e original resu lt w as m islead in g b e c a u se th e resid u als w ere highly co rrelated . T h e seem ingly m a jo r ch a n g e in th e form o f eq u a tion 6, w h ich d efines a n ticip a te d inflation, resu lts from th e ch an g e from first-d ifferen ce to rate-ofch an g e form . T h e cu rren t version is m u ch ea sier to in terp ret, bein g sim ply a w eigh ted sum o f past rates o f p rice ch an g es. F in a lly , th e u n e m p lo y m e n t e q u a t io n h a s ch an g ed only slightly. E ssen tially, th e co n sta n t in th e original version h as b ee n rep laced by th e fullem ploym ent u nem p loy m en t rate. T h is rate is in ten d ed to b e co n siste n t w ith the estim a te o f p o te n tial GNP. Table 1 Continued ... Symbols: A = change in dollar values (note: AP, = X,_, (P, - P,_,) and AX, = P,_,(X, - X,.,)) • = annual rate of change in variable Y, = total spending (GNP in current prices) M, = money stock (M1) E, = federal expenditures (high-employment in original model and cyclically adjusted in current model) EX, = Exports of goods and services (in current prices) P, = GNP deflator (1958 = 1.00 in original model and 1982 = 1.00 in current model) D, = demand pressure PA, = anticipated price level XF, = potential output (Council of Economic Advisers’ estimate in 1958 prices in original model and Rasche-Tatom estimate in 1982 prices in current model) X, = output (GNP in 1958 prices in original model and GNP in 1982 prices in current model) PE, = relative price of energy RL, = corporate Aaa bond rate U, = civilian unemployment rate G, = GNP gap as a percent of potential output Z, = dummy variable (1/1955-IV/1960 = 0; 1/1961-IV/1969 = 1) Z1, = dummy variable (1/1960-IV/1981 = 0; 1/1982-IV/1984 = 1) Z2, = price control dummy (1/1960-11/1971,11/1973-IV/1984 = 0; 111/1971-1/1973 = 1) Z3, = post-price control dummy (1/1960-1/1973, 11/1975-IV/1984 = 0; 11/1973-1/1975 = 1) 21 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 2 The Current Model’s Response to a Fiscal Shock (shocked values denoted by prime) Quarters elapsed 1 2 3 4 5 6 7 8 12 16 20 24 28 32 36 40 Exogenous variable EVE 1.0454 1.0440 1.0444 1.0445 1.0457 1.0461 1.0454 1.0455 1.0458 1.0479 1.0467 1.0436 1.0424 1.0391 1.0422 1.0412 Y/Y 1.0025 1.0035 1.0032 1.0030 1.0036 1.0036 1.0036 1.0035 1.0036 1.0037 1.0036 1.0034 1.0033 1.0031 1.0032 1.0032 x/x 1.0022 1.0027 1.0035 1.0017 1.0028 1.0029 1.0025 1.0020 1.0000 .9998 .9983 .9988 .9977 .9984 .9990 1.0002 Endogenous variables P/P 1.0001 1.0002 1.0005 1.0008 1.0011 1.0014 1.0018 1.0022 1.0033 1.0041 1.0047 1.0050 1.0051 1.0046 1.0039 1.0033 U'-U -.061 -.107 -.137 -.099 -.105 -.125 -.116 -.095 -.004 .020 .067 .049 .082 .067 .046 -.008 RL'-RL .001 .004 .009 .015 .020 .026 .033 .038 .053 .056 .053 .040 .025 .003 -.015 -.026 NOTE: To calculate percent change tor E, Y, X and P, subtract 1 and multiply by 100. U'-U and RL'-R are differences of percents. P R O P E R T IE S O F TH E CURRENT MODEL To d em o n strate th e p ro p erties o f th e cu rren t m odel, it w as su b je cte d to th ree different “sh o ck s.” In ea ch case, th e sh o ck began in 1/1975, an d th e sim u lated re sp o n se w as calcu lated th rou gh IV/1984. T h e th ree sh o ck s are:1 3 1. F isca l sh o ck . An in crease in cy clically ad ju sted ex p en d itu res equal to 1 p e rce n t o f GNP. 2. M onetary sh ock . A gradual in crea se in M l over a y e a r to 3 p e rce n t above th e b ase path. 3. Supply-side sh o ck . A low ering o f th e w orld oil p rice by 20 p ercen t. T h e resu lts o f sim ulating th e m o d el w ith e a c h sh o ck are sh ow n in tables 2-A. T h e se resu lts are su m m arized in table 5. 13These are the shocks simulated for Professor Klein's model com parison seminar, which was reorganized in 1985. For results of the earlier seminar in the 1970s, see Klein and Burmeister (1976). 22 Fiscal Shock Results T h e in crea se in cy clically a d ju sted ex p en d itu res quickly in flu en ces total sp en d in g. T h e total effect, however, is at m o st a .37 p e rce n t in cre a s e o r a m e a su red elasticity o f .08. T h e fiscal m u ltiplier, AY/AE, using average values for 1 9 7 8 -7 9 (the m idd le o f the sam ple period), is .38. T h is is m u ch low er th a n o th e r eco n o m etric m o d els.1 4 T h e d yn am ics o f the m odel in d ica te that th e initial in crea se in total sp en d in g is tra n sm itted first an d tem porarily to real GNP, th e n fully to th e p rice level. In fact, it ap p ears th at th e p rice level ov ersh oots its final equilibrium , im plying an u n d e rsh o o tin g o f real GNP. O utput an d th e p rice level co n tin u e to o scilla te after 40 qu arters, b u t th e fiscal sh o ck h a s essen tially no effect on ou tp u t in th e long run. C on sequ en tly, th e effect on th e u n em p lo y m en t rate is sm all, w ith th e o scillation o f th e u n em p lo y m en t rate sy n ch ro n o u s w ith output. Sim ilarly, th e effect o n th e lon g -term in terest rate is negligible even fou r o r five y ea rs after th e shock, as in tere st ra tes rise w ith in flatio n an d fall w h en th e rate o f in flation d eclin es. 14See Klein and Burmeister (1976), p. 338. OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 3 The Current Model’s Response to a Monetary Shock (shocked values denoted by prime) Quarters elapsed 1 2 3 4 5 6 7 8 12 16 20 24 28 32 36 40 Exogenous variable M/M 1.0014 1.0073 1.0188 1.0280 1.0300 1.0300 1.0300 1.0300 1.0300 1.0300 1.0300 1.0300 1.0300 1.0300 1.0300 1.0300 Y'/Y 1.0004 1.0026 1.0081 1.0163 1.0231 1.0265 1.0272 1.0272 1.0272 1.0272 1.0272 1.0272 1.0272 1.0272 1.0272 1.0272 X/X 1.0003 1.0019 1.0084 1.0145 1.0208 1.0226 1.0212 1.0186 1.0087 1.0016 .9948 .9919 .9891 .9904 .9930 .9965 Endogenous variables P/P 1.0000 1.0001 1.0004 1.0012 1.0026 1.0045 1.0067 1.0090 1.0181 1.0258 1.0319 1.0361 1.0380 1.0370 1.0341 1.0311 U'-U -.007 -.057 -.262 -.532 -.810 -.961 -.946 -.851 -.421 -.087 .202 .339 .450 .402 .311 .164 RL'-RL .000 .002 .008 .024 .051 .086 .125 .167 .308 .390 .414 .371 .272 .140 .014 -.076 NOTE: To calculate percent change for M, Y, X and P, subtract 1 and multiply by 100. U'-U and RL'-R are differences of percents. Monetary Shock Results A m o n etary sh o ck w orks th rou gh th e m od el in th e sam e w ay th e fiscal sh o ck d o es — via total sp en d in g. T h e d ifference is that th e effect is m u ch faster and larger. Normally, a m o n etary sh o ck is fully reflected in total sp en d in g after fou r quarter’s (see eq u atio n 1 in table 1). W ith th e ex p erim en t rep o rted h ere, M l bu ild s up over a y e a r’s tim e to 3 p e rce n t above th e b ase path. C on sequen tly, th e full effect on to tal sp en d in g is not registered until the seventh quarter. T h e d yn am ics o f th e m od el take over qu ite quickly w ith re sp ect to o u tp u t an d th e p rice level. O utput initially rises, bu t after fou r y ears retu rn s clo se to its b ase path level; it th e n falls below th e b ase level as the inflation rate co n tin u e s to in crease. In fact, th e e la stic ity o f th e p rice level peaks at 1.27 after seven y ears. T h e 4 0-q u arter sim u lation is n o t long en o u gh to d eterm in e th e n atu re o f the lon g-run equilibriu m . T h e m o n etary sh o ck p ro d u ces a stron g o scillato ry m ovem ent in th e u n em p loy m en t rate. Initially, th is rate drops quickly, falling to alm ost o n e p ercen tag e poin t below its base p ath after only six qu arters. After fou r y ears, U m oves back to its base path an d th en in crea ses above it, staying th ere for th e rem ain d er o f the sim u lation period. T h e effect o f th e m o n etary sh o ck o n in terest d e p en d s d irectly on the p rice level resp o n se. Inflation an d in terest rates resp o n d slow ly to th e sh o ck . As long as inflation in crea ses, in tere st rates rise above th eir b ase path. W h en in flation slow s after abou t seven years, in terest rates m ove b a ck tow ard th e ir b a se p ath . As w ith several o th e r variables, th e sim u lation period is n o t lon g en o u gh to d eterm in e th e n atu re o f th e final equilibrium . Supply-Side Shock Results To sim u late th e effect o f a su p p ly shock, th e p rice o f oil p e r barrel w as a ssu m ed to d rop 20 p ercen t, w h ich red u ces th e relative p rice o f energy by 8 p e rce n t. T h is variable d irectly affects th e p rice eq u ation an d in d i rectly affects th e p rice level b ec a u se th e d rop in th e p rice o f oil is a ssu m ed to in sta n ta n e o u sly in crea se p o ten tial ou tp u t by .4 p e rc e n t.1 By assu m p tio n , total ’ sp en d in g is n ot affected by th e su p p ly sh ock , th at is, th e relative p rice o f energy is n o t in clu d ed in th e total sp en d in g eq u atio n . T h is a ssu m p tio n is in d isp ute, how ever, as T ato m argues th at to tal sp en d in g is te m porarily affected by su ch a sh o ck ."1 l5Rasche and Tatom (June 1977). 16Tatom (1981). His argument rests on the significance of only one of the lagged values of PE. For this reason, this variation has not been introduced into the version of the model summarized in table 1. 23 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 4 The Current Model’s Response to a Supply-Side Shock (shocked values denoted by prime)_________________ Endogenous variables Exogenous variables Quarters P'/P XF/XF Y'/Y X'/X PE/PE elapsed .9998 1.0040 1.0000 1.0000 1 .9200 1.0000 .9999 .9996 1.0040 .9200 2 .9992 1.0040 1.0000 1.0015 .9200 3 1.0000 1.0006 .9988 1.0040 4 .9200 .9984 1.0040 1.0000 1.0019 .9200 5 1.0000 1.0026 .9981 1.0040 .9200 6 .9978 1.0040 1.0000 1.0030 7 .9200 1.0000 1.0031 .9975 1.0040 .9200 8 1.0000 1.0035 .9963 1.0040 .9200 12 1.0000 1.0049 .9953 1.0040 .9200 16 1.0000 1.0048 .9946 1.0040 .9200 20 1.0000 .9944 1.0040 1.0061 24 .9200 1.0000 1.0050 .9946 1.0040 .9200 28 1.0040 1.0000 1.0050 .9948 32 .9200 1.0000 1.0044 .9953 1.0040 .9200 36 .9957 1.0000 1.0046 1.0040 .9200 40 U -U .109 .172 .130 .132 .110 .071 .049 .040 .030 -.025 -.039 -.093 -.057 —.044 -.018 -.029 RL'-RL -.003 -.009 -.016 -.023 -.030 -.035 -.040 -.044 -.058 -.064 -.059 -.043 -.022 -.004 .011 .019 NOTE: To calculate percent change for PE, XF, Y, X and P, subtract 1 and multiply by 100. U'-U and RL'-R are differences of percents. Table 5 Estimated Elasticities for the Current Model With respect to With respect to fiscal shock (E) monetary shock (M) Quarters X P Y elapsed Y X P 1 .06 .05 .00 .29 .21 .00 .06 .00 .36 2 .08 .26 .01 .07 .08 .01 .43 3 .45 .02 .07 .04 .02 4 .58 .52 .04 .04 .08 .05 .91 .62 8 .30 .07 12 .08 .00 .91 .29 .60 .09 .08 -.0 0 .91 .05 16 .86 -.0 4 .10 .91 -.1 7 20 .08 1.06 .11 -.2 7 24 .08 -.0 3 .91 1.20 .12 -.0 5 .91 -.36 1.27 28 .08 -.0 4 .12 .91 -.3 2 32 .08 1.23 -.0 2 .09 .91 1.14 .08 -.2 3 36 .00 .08 .08 .91 -.1 2 1.04 40 T ables 4 and 5 sh o w th at o u tp u t an d p rices resp o n d qu ite slow ly to th is sh o ck . M oreover, th e m axim u m effect, w h ich o cc u rs after abou t six y ears, is relatively 24 With respect to supply shock (PE) Y X .00 -.0 0 .00 .00 .00 -.02 .00 -.01 -.0 4 .00 .00 -.0 4 .00 -.0 6 .00 -.0 6 .00 -.0 8 .00 -.0 6 .00 -.0 6 .00 -.0 6 .00 -.0 6 P .00 .01 .01 .02 .03 .05 .06 .07 .07 .07 .07 .06 .05 sm all. In fact, th e e la sticities (calcu lated w ith re sp ect to th e relative p rice o f energy) are sim ilar in m agn itu d e to th o se for fed eral exp en d itu res. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 C h a rt 1 Nominal GNP 1970 71 72 73 74 75 76 ASSESSING TH E CURRENT M O D EL’S PERFORM ANCE To provide som e in d icatio n o f m od el perfo rm an ce, th e m od el w as sim u lated ejc post d uring several p eri od s after 1969. D enoting su ch sim u lation s as e\ post m ean s that all sim u lation s w ere w ithin th e sam ple p eriod and th e exo g en o u s variables took on th e ir actu al values. All sim u lation s w ere d ynam ic; th at is, o n ce th e sim u lation w as started, th e m od el g en erated its ow n lagged values. T h e se sim ulations are su m m arized in ch arts 1-3 an d table 6. U nfortunately, th e se resu lts m ean little bv th em selves b ec a u se th ere is n o b asis for co m p ariso n . R esu lts for sim ilar sim u lation ex e rcises w ith o th e r m o d els have b ee n p u blish ed for th e 1960s an d early 1970s, bu t are n o t readily available for m o re re cen t 77 78 79 80 81 82 83 1984 period s. C on sequen tly, an y c o n c lu sio n s abo u t th e m o d el’s p erfo rm an ce are im p ressio n istic. Total Spending Growth C harts 1-3 sh ow th e resu lts o f sim u lating Y, X and P for th e full sim u lation period from 1970 th rou gh 1984. Sin ce the total sp en d in g eq u atio n co n ta in s no en d o g en o u s variables, th e m od el sim u lation sh o w n in ch art 1 sim ply show s th e fit o f th a t eq u atio n . T h a t fit obvi ously d oes qu ite poorly on a q u a rter-to -q u a rter b asis but seem s to follow th e co n to u rs over several qu arters, alm ost as if a moving average had b een applied to the actual observ ations. A d esirab le featu re o f th is eq u a tion is that the q u a rter-to -q u a rter erro rs do n ot ten d to cu m u late over tim e. T h e erro rs in th e estim ated eq u a tion are n o t correlated . Table 6 show s th at th e RMSE o f Y in cre a se s over 25 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS C h a rt 2 Real GNP 1970 71 72 73 74 75 76 77 tim e and, even w h en stan d ard ized by th e level o f GNP (SRMSE), it co n tin u es to grow as th e sim u lation period m oves tow ard th e p resen t. T h is su ggests th at th e relatio n sh ip b etw een Y an d M h as b eco m e lo o ser recen tly. Output Growth T h e relative degree o f su c c e ss in sim u lating total sp en d in g is carried over to th e sim u lation o f output. T h e m od el sim u lated X w ell over th e p eriod s, alth ough it u n d erestim ated e co n o m ic stren g th d uring th e ex p a n sio n from th e 19 7 3 -7 5 re cessio n . T h e o th e r period o f su b stan tial d ifferen ce h as o ccu rred sin ce th e third q u a rter o f 1983. T h e m o d el in d icated a recessio n , w h ich did n o t o ccu r. W h en th e m od el is sim u lated over different period s, n o c o n siste n t p attern em erg es for th e SRM SE for X. In th e 1 9 7 0 -8 4 period, th e SRM SE for X ex ceed e d th at for Y. In th e 19 7 5 -8 4 an d 1 9 7 9 -8 4 periods, how ever, the 26 78 79 80 81 82 83 1984 SRMSE for X w as less th a n for Y, a p p aren tly reflectin g th e em erging im p o rta n ce o f aggregate d em a n d sh o ck s relative to su pply-sid e sh o ck s d uring th e se p eriod s. Inflation T h e resu lts o f sim ulating th e in flation rate over th e 1 9 7 0 -8 4 period are sh o w n in ch a rt 3. G enerally sp eak ing, th e m ovem ents w ere ap p roxim ated d urin g th e 1 9 7 0 -7 7 period , bu t th e a cc e le ra tio n startin g in th e seco n d q u arter o f 1978 w as n o t p ick ed up u ntil a y e a r or so later. T h e e ss e n c e o f th e gen eral d ecelera tio n from m id-1980 w as cap tu red , b u t sin c e m id-1982, th e m odel has overestim ated in flation by ab o u t 2 p e rc e n t age poin ts. T h e se visual im p ressio n s are b o rn e ou t in th e c a lc u lation o f RMSE for th e GNP deflator. T h e sh o rtest and latest p eriod w as b est w ith a stan d ard ized RMSE o f 2.62 p e rce n t. T h e 1 9 7 5 -8 4 perio d w as th e w orst w ith an SRM SE o f 4.15 p e rce n t. T h e sim u lation for the FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 C h a rt 3 GNP Deflator 1970 71 72 73 74 75 76 77 78 79 80 81 82 83 1984 Table 6 Ex Post Simulation Results GNP (current dollars) Real GNP (1982 dollars) GNP deflator (1982 = 100) Unemployment rate (percent) Corporate Aaa bond rate (percent) 1/70-1V/84 RMSE SRMSE 46.42 2.15% 75.31 2.56 1.99 2.80 1.31 18.91 2.35 23.95 1/75— IV/84 RMSE SRMSE 78.24 2.98% 85.55 2.75 3.45 4.15 1.48 19.25 2.89 26.69 I/79-IV/84 RMSE SRMSE 107.31 3.46% 72.03 2.21 2.62 2.49 1.31 16.60 3.23 26.13 NOTE: RMSE is root mean squared error; SRMSE is standardized RMSE, that is, RMSE divided by the level of the variable and multiplied by 100. 27 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS overall period w as in betw een , w ith an SRM SE o f 2.79 p ercen t. Unemployment and Interest Rates T ab le 6 show s th at th e RMSE for sim u lation s o f the civilian u n em p loy m en t rate an d th e Aaa b o n d rate do n o t vary b y m u ch over different sim u lation period s. T h e RMSE is m o re m eaningfu l for th e se co m p ariso n s th an SRM SE b e c a u se the RMSE is alread y ex p ressed in p ercen tag e poin ts. Sim u lation s o f th e m ovem ents o f th e Aaa bo n d rate w ere g enerally u nim pressive. A lthough th e RMSE w as little different for th e alternative sim u lation period s, it in crea sed as th e sim u lation w as brou ght c lo se r to the p resen t. SUMMARY p rice level, w ith very little effect on o u tp u t an d u n e m ploym ent. F iscal a ctio n s have sm all sh o rt-ru n effects th at d isap p ear (in term s o f output) qu ite quickly. W hile th e su pply-sid e effects are n ot stron g acco rd in g to co n v en tion al ela sticities, th e se effects can b e im p o r tan t if energy p rices m ove d ram atically. T h e p erfo rm an ce o f th e m o d el is difficult to gauge, but, for th e m o st part, th e sim u lation re su lts w ere d eem ed su cce ssfu l. post sim u latio n s are th e c o n ven tion al m eth o d o f a sse ssin g a m o d e l’s p erfo rm an ce, b u t th ey are m o re m eaningfu l w h en co m p a red w ith th o se from o th e r m o d els. T h e re have b e e n n o p u b lish ed stu d ies o f h o w o th e r m o d els are p erform ing in th e 1980s. A m o re a ccu ra te evaluation aw aits co m p a ri son w ith sim ilar resu lts from o th e r cu rren t m od els. R EFER EN C E S Sin ce 1970, th e g en eral form o f th e m od el h as b een m ain tain ed , bu t several ch an g es in its sp ecificatio n an d estim atio n have b een m ade. O ne n o table ch an g e h as b een sim p lification — using rates o f ch an g e in stead o f first d ifferences. A n o th er is th e ad d itio n o f su p p ly -sid e variables — th e relative p rices o f energy an d p rice co n tro l an d d eco n tro l d u m m ies and , m ost recen tly, a d um m y in th e GNP eq u atio n to cap tu re the shift in th e relatio n sh ip sin ce 1981. O th er ch an g es in clu d ed alternative estim ates o f p o ten tial o u tp u t an d fed eral ex p en d itu res, an d a d ju stm en ts for a u to co rre latio n in several o f th e eq u atio n s. Andersen, Leonall C., and Keith M. Carlson. “A Monetarist Model for Economic Stabilization,” this Review (April 1970), pp. 7-25. Andersen, Leonall C., and Jerry L. Jordan. “Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabili zation,” this Review (November 1968), pp. 11-24. Carlson, Keith M. “Does the St. Louis Equation Now Believe in Fiscal Policy?” this Review (February 1978), pp. 13-19. Considine, William. “Public Policy and the Current Inflation,” pre pared as part of a summer intern program at the U.S. Treasury Department (September 5, 1969). de Leeuw, Frank, and Edward Gramlich. “The Channels of Mone tary Policy,” Federal Reserve Bulletin (June 1969), pp. 472-91. _________“The Federal Reserve-MIT Econometric Model,” Fed eral Reserve Bulletin (January 1968), pp. 11-46. de Leeuw, Frank, and Thomas M. Holloway. “Cyclical Adjustment of the Federal Budget and Federal Debt,” Survey of Current Busi ness (December 1983), pp. 25-40. Klein, Lawrence R., and Edwin Burmeister, eds. Econometric Model Performance (University of Pennsylvania Press, 1976). Okun, Arthur M. “ Potential GNP: Its Measurement and Signifi cance,” 1962 Proceedings of the Business and Economic Statistics Section of the American Statistical Association, pp. 98-104. Rasche, Robert H., and John A. Tatom. "Energy Resources and Potential GNP,” this Review (June 1977a), pp. 10-24. _________“The Effects of the New Energy Regime on Economic Capacity, Production, and Prices,” this Review (May 1977b), pp. D esp ite th e se ch an g es, th e p ro p erties o f th e m odel rem ain essen tially u n ch an g ed . M o n etaiy a ctio n s have a large sh o rt-ru n effect on to tal sp end ing, o u tp u t and u n em p lo y m en t; over th e long run, how ever, th e effect on total sp en d in g is alm ost en tirely reflected in th e Tatom, John A. “Energy Prices and Short-Run Economic Perfor mance,” this Review (January 1981), pp. 3-17. Yohe, William P., and Denis S. Karnosky. “Interest Rates and Price Level Changes, 1952-69,” this Review (December 1969), pp. 1838. T h e St. Louis m odel, as originally p u b lish ed in April 1970, w as d esigned to fo cu s on th e im p o rtan ce o f m o n etary a ctio n s in th e d eterm in atio n o f spending, ou tp u t an d p rice s. Its stru ctu re differed su bstan tially from o th e r ec o n o m e tric m od els at th at tim e. It c o n sisted o f th e A n d ersen -Jo rd an GNP eq u atio n and sev eral o th e r em p irical relatio n sh ip s; it w as recu rsive in form . It estim ated GNP d irectly using m o n eta iy and fiscal variables, in sh arp co n tra st to th e con vention al a p p ro ach o f estim atin g th e co m p o n e n ts o f GNP and th e n su m m in g th e m to obtain a GNP estim ate. 28 2 12 - . The follow ing article is reprinted fr o m the N ovem b er 1968 Federal Reserve Bank o f St. Louis Review. M onetary and Fiscal Actions: A Test of Their Relative Im portance in Econom ic Stabilization Leonall C. Andersen and Jerry L. Jordan H IGH em ploym ent, rising ou tp u t o f good s and services, an d relatively stable p rices are th ree w idely a cc e p te d nation al ec o n o m ic goals. R esponsib ility for ec o n o m ic stabilization actio n s to m eet th ese goals has b een assigned to m o n etaiy an d fiscal au th o rities. T h e F ed eral Reserve System h as th e m a jo r resp onsib ility for m o n etary m an ag em ent. F iscal a ctio n s involve fed eral governm ent sp en d in g plans and taxing provi sion s. G overnm ental u nits involved in fiscal actio n s are th e C on gress an d th e A d m inistration, in clu d in g th e Treasury, th e B ureau o f th e Budget, and th e C o u n cil o f E co n o m ic Advisers. T h is article rep o rts th e resu lts o f re cen t re sea rch w h ich te ste d th ree co m m o n ly h eld p ro p o sition s c o n cern in g th e relative im p o rtan ce o f m o n eta iy and fiscal a ctio n s in im p lem en tin g ec o n o m ic stabilization p o l icy. T h e se p ro p o sitio n s are: th e re sp o n se o f e co n o m ic activity to fiscal actio n s relative to th at o f m o n eta iy a ctio n s is (1) greater, (2) m ore p red ictable, an d (3) faster. Sp ecific m eanings, for th e p u rp o ses o f this article, o f th e broad term s u sed in th e se p ro p o sition s are p resen ted later. The authors give special thanks for helpful comments on earlier drafts to: Robert Basmann, Karl Brunner, James Buchanan, Albert Burger, Keith Carlson, David Fand, Milton Friedman, Gary Fromm, Michael Levy, Thomas Mayer, A. James Meigs, David Meiselman, Allan Meltzer, Richard Puckett, David Rowan, James Tobin, Robert Weintraub, and William Yohe. The authors are, of course, solely responsible for the analyses and results presented in this article. T h is article d o es n o t attem p t to te st rival e co n o m ic th eo ries o f th e m e ch a n ism by w h ich m o n etary and fiscal a ctio n s in flu en ce ec o n o m ic activity. N eith er is it in ten d ed to develop evid en ce bearin g d irectly on any cau sal relatio n sh ip s im plied by su ch th eo ries. M ore elaborate p ro ced u res th an th o se u sed h ere w ould be required in o rd er to test any th eo ries u nd erlying the fam iliar sta tem en ts regarding resu lts ex p ected from m on etary an d fiscal a ctio n s. However, em p irical rela tio n sh ip s are developed b etw een frequ en tly u sed m easu res o f stabilization a ctio n s an d e co n o m ic activ ity. T h ese relatio n sh ip s are co n siste n t w ith th e im p li ca tio n s o f som e th e o rie s o f stab ilization policy an d are in co n sisten t w ith other's, as w ill b e p o in ted out. A b rief d iscu ssio n o f th e forces in flu en cin g e c o n o m ic activity is p re sen ted first. Next, w ith th is th eory as a background , sp ecific m e a su res o f e c o n o m ic activ ity, fiscal actio n s, and m o n eta iy a ctio n s are selected . T h e resu lts o f testin g th e th ree p ro p o sitio n s n oted above, to g eth er w ith o th e r sta tem e n ts co n c e rn in g th e resp o n se o f eco n o m ic activity to m o n etary an d fiscal forces, are th en p resen ted . Finally, so m e im p licatio n s for the co n d u ct o f stab ilization policy are draw n from th e resu lts o f th e se tests. A T H E O R E T IC A L V IE W O F E C O N O M IC A C T IV IT Y O ur eco n o m ic system co n sists o f m any m arkets. Every com m od ity, service an d fin an cial a sse t is view ed as co n stitu tin g an individual m arket in w h ich a 29 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS p articu lar item is trad ed an d a p rice is d eterm in ed . All of th e se m arkets are linked to g eth er in varying d e grees, sin ce p rices in o n e m arket in flu en ce d ecisio n s m ad e in o th e r m arkets. About a ce n tu ry ago, Leon W alras o u tlin ed a fram e w ork for analyzing a co m p lex m arket eco n o m y . Su ch an analysis in clu d es a d em an d an d a su pply relatio n sh ip for eveiy co m m o d ity an d for e a c h fa cto r o f p ro d u ctio n . T rad in g in th e m arkets resu lts in p rices being esta b lish ed w h ich cle a r all m arkets, i.e., th e am ou n t offered in a m arket eq u als th e am o u n t taken from the m arket. A ccord ing to th is analysis, o u tsid e o c c u r re n ce s reflected in shifts in d em an d an d su p p ly rela tio n sh ip s ca u se ch an g es in m arket p rices and in q u a n tities traded . T h e se o u tsid e events in clu d e ch an g es in p re fe ren ces o f m arket p articip an ts, in re so u rce e n d ow m en ts, an d in tech nolog y. F in an cial asse ts w ere n o t view ed as providing utility o r satisfactio n to th e ir h o ld ers an d w ere th e re fo re e x c lu d e d from th e analysis. L a ter d ev e lo p m e n ts in e c o n o m ic th e o ry have view ed fin an cial asse ts as providing flow s o f services w h ich also provide utility o r satisfactio n to hold ers. F or exam ple, a h o ld e r o f a co m m ercial bank tim e d ep o sit receives liquidity service (ease o f co n versio n into th e m edium o f exch ange), store o f value service (ability to m ake a future p u rch ase), risk avoid ance service (little risk o f loss), an d a fin an cial yield . A cco rd ing to th is la te r view, e c o n o m ic en tities in co rp o ra te ch o ice s am on g goods, seiv ices, and fin ancial assets into th e ir d ecisio n -m ak in g p ro cesses. Exhibit 1 Classification of Market Variables Dependent Variables Prices and quantities of goods and services. Prices and quantities of factors of production. Prices (interest rates) and quantities of financial assets. Expectations based on: a. movements in dependent variables. b. expected results of random events. c. expected changes in fiscal and monetary policy. Independent Variables Slowly changing factors: a. preferences. b. technology. c. resources. d. institutional and legal framework. Events outside the domestic economy: a. change in total world trade. b. movements in foreign prices and interest rates. Random events: a. outbreak of war. b. major strikes. c. weather. Forces subject to control by: a. fiscal actions. b. monetary actions. in flu en ce ex p o rts an d th erefo re are largely an o u tsid e in flu en ce on d o m estic m arkets. T h e fact th a t e c o n o m ic en tities m ake c h o ic e s in b o th m arkets for g oo d s and seiv ices an d m arkets for fin an cial asse ts req u ires th e ad d ition o f d em an d an d su pply relatio n sh ip s for eveiy fin an cial asset. M arket in terest rates (prices o f fin an cial assets) an d ch an g es in th e stocks o u tstan d in g o f m o st fin an cial asse ts are d eterm in ed by th e m arket p ro cess along w ith p rices an d qu an tities o f good s an d seiv ices. M arket ex p e cta tio n s have also b ee n a ssign ed a significant fa cto r in m arkets, bu t th e se are not view ed as a d istin ctly in d e p en d e n t force. E x p ecta tio n s result from m arket p a rticip a n ts b asin g th e ir d ecisio n s on m ovem en ts in m ark et-d eterm in ed variables, o r they are derived from m arket re sp o n ses to th e ex p ected resu lts o f ran d o m events, su ch as th e o u tb reak o f a w ar o r th e a n ticip a tio n o f ch a n g es in fiscal o r m o n eta iy policy. T h e se th e o retical d evelop m ents have enlarged the n u m b er o f in d e p en d e n t fo rces w h ich are regarded as in flu en cin g m ark et-d eterm in ed p rices, in terest rates, qu an tities p ro d u ced o f co m m o d ities and stocks o u t stan d in g o f fin an cial a ssets. G overnm ent an d m o n e tary au th o rities are view ed as exertin g in d ep en d en t in flu en ces in th e m arket system . T h e se in flu en ces are called fiscal and m o n etary p o licies or actio n s. R an dom events, su ch as th e ou tbreak o f w ar, strikes in key in d u stries an d prolon ged drought, exert o th e r m arket in flu en ces. Growth in w orld trade an d ch an g es in foreign p rices an d in tere st rates, relative to o u r own, T h e se d ep en d en t an d in d e p en d e n t m arket vari ables are su m m arized in exhibit 1. T h e d ep en d en t variables are d eteim in e d by the in terp lay o f m arket forces w h ich resu lts from ch a n g es in th e in d e p en d e n t variables. M arket-d eterm in ed variables in clu d e p rices and qu an tities o f goo d s an d seiv ices, p rices a n d q u an tities o f factors o f p ro d u ctio n , p rices (interest rates I and qu an tities o f financial a ssets, an d ex p ecta tio n s, in d e p en d e n t variables co n sist o f slow ly ch an g in g fac tors, forces from o u tsid e o u r eco n om y, rand om events an d fo rces su b je ct to co n tro l by fiscal an d m o n etary au th o rities. A ch a n g e in an in d ep en d en t 30 FEDERAL RESERVE BANK OF ST. LOUIS variable (for exam ple, a fiscal o r a m o n eta iy action) ca u ses ch an g es in m any o f th e m ark et-d eterm in ed (dependent) variables. MEASURES O F ECONOMIC ACTIVITY AND O F MONETARY AND FISCAL ACTIONS T h re e th eo retical ap p ro ach es have b een advanced by eco n o m ists for analyzing th e in flu en ce o f m o n eta iy an d fiscal a ctio n s on e c o n o m ic activity. T h e se a p p ro a ch es are th e textbook K eynesian analysis derived from e co n o m ic thou ght o f th e late 1930s to th e early 1950s, th e portfolio ap p ro ach d eveloped over th e last tw o d ecad es, and th e m o d ern qu an tity th e o iy of m oney. E ach o f th e se th eo ries h as led to p o p u la r and fam iliar statem en ts regarding th e d irectio n , am oun t, an d tim ing o f fiscal and m o n eta iy in flu en ces on e c o n o m ic activity. As n o ted earlier, th ese th eo ries and th e ir linkages will not be tested directly, bu t th e valid ity o f som e o f th e statem en ts w h ich p u rport to re p re sen t th e im p licatio n s o f th e se th eo ries will be ex a m in ed . F o r th is p u rpose, frequ ently u sed m easu res of e c o n o m ic activity, m o n e ta ry a c tio n s, an d fiscal a ctio n s are selected . Economic Activity T otal sp end ing for goods and send ees (gross n a tional p ro d u ct at cu rren t prices) is u sed in th is article as th e m easu re o f e co n o m ic activity. It c o n sists o f total sp en d in g on final goods and services by h o u seh old s, b u sin esses and governm ents plus n et foreign invest m ent. Real ou tp u t o f goods and services is lim ited by re so u rce end ow m en ts an d techn ology, w ith th e a c tual level o f output, w ithin th is co n strain t, d eterm in ed by th e level o f total sp en d in g and o th e r factors. Monetary Actions M onetary actio n s involve prim arily d ecisio n s o f the T re a su iy and th e Fed eral Reserve System . Treasu ry m o n eta iy actio n s co n sist o f variations in its ca sh h o ld ings, d ep osits at Federal R eseive banks and at co m m ercial banks, and issu a n ce of T re a su iy cu rren cy . Fed eral R eseive m o n eta iy a ctio n s in clu d e ch an g es in its portfolio o f G overnm ent secu rities, variations in m em b er bank reserve requ irem en ts, and ch an g es in th e Fed eral Reserve d isco u n t rate. Banks and th e p u b lic also engage in a form o f m o n etaiy actio n s. C om m ercial bank d ecisio n s to hold e x cess reseives co n s ti tu te a m o n eta iy actio n . Also, b ec a u se o f differential reserve requ irem ents, th e p u b lic’s d ecisio n s to hold varying am o u n ts o f tim e d ep o sits at co m m ercial banks OCTOBER 1986 or cu rre n cy relative to d em an d d ep o sits are a form of m o n etary actio n , bu t are n o t view ed as stabilization a ctio n s. However, th ey are taken in to co n sid era tio n by stabilization a u th o rities in form ing th e ir ow n a ctio n s. Exhibit 2 su m m arizes th e various so u rces o f m o n eta iy actio n s related to e co n o m ic stabilization . T h e m o n eta iy b a s e 1 is co n sid ere d by b o th th e p o rt folio an d th e m o d ern qu an tity th e o iy sch o o ls to lie a strategic m o n eta iy variable. T h e m o n eta iy b a se is u n d er d irect co n tro l o f the; m o n eta iy au th o rities, w ith m a jo r co n tro l ex erted by th e F ed eral R eseive System . Both o f th e se sch o o ls c o n sid e r an in crea se in the m o n eta iy base, o th e r fo rces co n sta n t, to b e an exp an sio n aiy in flu en ce on e c o n o m ic activity an d a d ecrea se to b e a restrictive in flu en ce. T h e portfolio sch o o l h o ld s th at a ch a n g e in th e m o n eta iy b a se affects in v estm en t sp en d in g, an d th ereb y aggregate spen din g, th rou gh ch a n g es in m ar ket in terest rates relative to th e su p p ly p rice o f cap ital (real rate o f retu rn on capital). T h e m o d ern qu antity th e o iy h old s th at th e in flu en ce o f th e m o n eta iy b ase w orks through ch a n g es in th e m o n ey stock w h ich in turn affect p rices, in terest rates an d sp en d in g on good s an d seiv ices. In c re a se s in th e b a se are reflected in in crea ses in th e m o n ey sto ck w h ich in tu rn result directly an d in d irectly in in crea sed ex p en d itu res on a w h ole sp ectru m o f cap ital an d c o n su m e r g ood s. Both p rices o f good s an d in terest ra tes form th e tra n sm is sion m e ch a n ism in th e m o d ern qu an tity th eo iy . T h e m o n ey sto ck is also u sed as a strategic m o n e taiy variable in e a c h o f th e a p p ro a ch es to stabilization policies, as th e above d iscu ssio n h as im plied . T h e sim p le K eynesian a p p ro a ch p o stu lates th at a ch an g e in th e stock o f m on ey relative to its d em an d resu lts in a ch a n g e in in terest rates. It also p o stu lates th at in vestm en t sp en d in g d ecisio n s d ep en d on in terest rates, an d th at grow th in aggregate sp en d in g d ep en d s in tu rn on th e se in v estm en t d ecisio n s. Sim ilarly, in th e portfolio sch o o l o f th ou gh t, ch a n g es in th e m on ey stock lead to ch a n g es in in terest rates, w h ich are follow ed by su b stitu tio n s in asset portfolios; th en 'The monetary base is derived from a consolidated monetary bal ance sheet of the Federal Reserve and the Treasury. See Leonall C. Andersen and Jerry L. Jordan, “The Monetary Base: Explanation and Analytical Use,” in the August 1968 issue of this Review. Since the uses of the base are bank reserves plus currency held by the public, it is often called “demand debt of the Government.” See James Tobin, “An Essay on Principles of Debt Management," in Fiscal and Debt Management Policies, The Commission on Money and Credit, Prentice-Hall, Inc., Englewood Cliffs, N.J., 1963. In some analyses, Tobin includes short-term government debt out standing in the monetary base. 31 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Exhibit 2 Stabilization Actions and Their Measurement Stabilization actions 1. Monetary Actions Federal Reserve System a. open market transactions. b. discount rate changes. c. reserve requirement changes. Treasury a. changes in cash holdings. b. changes in deposits at Reserve banks. c. changes in deposits at commercial banks. d. changes in Treasury currency outstanding. 2. Fiscal Actions Government spending programs. Government taxing provisions. Frequently used measurements of actions 1. Monetary Actions Monetary base.' Money stock, narrowly defined.1 Money plus time deposits. Commercial bank credit. Private demand deposits. 2. Fiscal Actions High-employment expenditures.’ High-employment receipts.1 High-employment surplus.1 Weighted high-employment expenditures. Weighted high-employment receipts. Weighted high-employment surplus. National income account expenditures. National income account receipts. Autonomous changes in government tax rates. Net government debt outside of agencies and trust funds. ’Tests based on these measures are reported in this article. The remaining measures were used in additional tests. These results are available on request. finally, total sp en d in g is affected. In terest rates, a c co rd in g to th is la tte r sch o o l, are the key part o f th e tra n sm issio n m ech an ism , in flu en cin g d ecisio n s to h old m o n ey versus alternative fin an cial asse ts as well as d ecisio n s to invest in real a sse ts. T h e in flu en ce of ch a n g es in th e m o n ey sto ck on e c o n o m ic activity, w ithin th e m o d ern q u an tity th e o iy fram ew ork, h as alread y b ee n d iscu sse d in th e previous p aragraph.1 T h e m o n eta iy base, as noted, plays an im p ortan t role in b o th the portfolio and th e m o d ern quantity th e o iy ap p ro ach es to m o n eta iy th eo iy . However, th ere rem ain s co n sid erab le co n tro versy regarding th e role o f m o n ey in d eterm in in g e c o n o m ic activity, ran g ing from “m on ey d o es n ot m a tter” to “m o n ey is the d o m in an t facto r.” In re ce n t y ears th ere h as b een a g en eral a c c e p ta n c e that m oney, am ong m any o th er in flu en ces, is im p ortan t. T h o m as M ayer, in a re cen t book, su m m arizes th is controversy. He co n clu d e s: 2Also see Leonall C. Andersen and Jerry L. Jordan, “Money in a Modern Quantity Theory Framework” in the December 1967 issue of this Review. For an excellent analysis of these three monetary views see David I. Fand, “Keynesian Monetary Theories, Stabiliza tion Policy and the Recent Inflation,” a paper presented to the Conference of University Professors, Ditchiey Park, Oxfordshire, England, Sept. 13, 1968. 32 All in all, m u c h r e c e n t e v id e n c e s u p p o rts th e view th at th e sto ck o f m o n ey an d , th e re fo re , m o n e ta iy p o licy, h a s a su b sta n tia l e ffe ct. N ote, how ever, th a t th is re a d ing o f th e e v id e n ce is by n o m e a n s a c c e p ta b le to all e c o n o m ists . S o m e , p ro fe s so r F rie d m a n a n d Dr. W arb u rto n for exam p le, arg u e th at c h a n g e s in th e s to c k o f m o n ey d o have a d o m in a n t e ffe ct o n in c o m e , at le a st in th e lo n g ru n , w h ile o th e rs s u c h as P ro fe ss o r H ansen believe th a t c h a n g e s in th e s to c k o f m o n e y a re largely offset by o p p o site c h a n g e s in v e lo c ity .1 T h e th eo ries aside, ch a n g es in th e m o n eta iy base an d ch a n g es in th e m on ey sto ck are frequ en tly u sed as m easu res o f m o n eta iy a ctio n s. T h is article, in part, tests th e u se o f th e se variables for th is p u rp o se. M oney is narrow ly defined as th e n o n b an k p u b lic’s h old ings of d em an d d ep o sits p lu s cu rre n cy . C h an g es in th e m on ey stock m ainly reflect m ov em en ts in th e m o n e taiy b ase; how ever, th ey also reflect d ecisio n s o f c o m m ercial banks to hold e x c e ss reserves, o f th e n o n b an k p u blic to h o ld cu rre n cy an d tim e d ep o sits, an d o f the T re a su iy to hold d em an d d ep o sits at co m m ercia l banks. T h e m o n eta iy b a se reflects m o n e ta iy a ctio n s o f 3Thomas Mayer, Monetary Policy in the United States, Random House, NY, 1968, pp. 148-49. FEDERAL RESERVE BANK OF ST. LOUIS th e F ed eral Reserve, an d to a le sse r extent, th o se o f th e Treasu ry an d gold flows. But ch an g es in th e b ase have b een found to be d o m in ated by a ctio n s o f the Federal Reserve.4 O th er aggregate m easu res, su ch as m o n ey plus tim e d ep o sits, bank cred it and private d em an d d eposits, are frequ ently u sed as m o n eta iy in d icato rs (exhibit 2). T ests using th e se in d icato rs w ere also m ad e. T h e resu lts o f th e se tests did n o t ch an g e th e co n clu sio n s rea ch ed in this article; th e se resu lts are available on requ est. M arket in terest rates are n ot u sed in th is article as strategic m o n eta iy variables sin c e they reflect, to a great extent, fiscal actio n s, ex p ectation s, an d o th e r factors w h ich ca n n o t p roperly be called m o n eta iy actio n s. Fiscal Actions T h e in flu en ce o f fiscal actio n s on ec o n o m ic activity is frequently m easu red by federal governm ent sp e n d ing, ch an g es in federal tax rates, o r federal budget d eficits and su rp lu ses. T h e textbook K eynesian view has b een reflected in m any p o p u lar d iscu ssio n s of fiscal in flu en ce. T h e portfolio ap p ro ach an d th e m o d e m qu antity theory suggest alternative analy ses o f fiscal influ en ce. T h e elem en tary textbook K eyn esian view c o n c e n trates alm ost exclusively on th e d irect in flu en ce o f fiscal actio n s on total spendin g. G overnm ent sp en d in g is a d irect dem and for good s and services. Tax rates affect d isp osable in co m e, a m a jo r d eterm in an t o f c o n su m er spending, an d profits o f b u sin esses, a m ajo r d eterm in an t o f investm ent spendin g. Budget su r pluses and deficits are u sed as a m easu re o f th e net d irect in flu en ce o f sp end ing an d taxing on e co n o m ic activity. M ore advanced textbooks also in clu d e an in d irect in flu en ce o f fiscal actio n s on e c o n o m ic activ ity through ch an g es in m arket in terest rates. In eith e r case, little co n sid eratio n is gen erally given to th e m eth o d o f finan cing exp en d itu res. T h e portfolio ap p ro ach as d eveloped by T o b in a t trib u tes to fiscal a ctio n s b o th a d irect in flu en ce on e co n o m ic activity an d an in d irect in flu en ce. B oth in flu en ces take into co n sid eratio n th e fin an cin g o f gov ern m en t expenditures/' F in an cin g o f ex p en d itu res by issu a n ce o f d em and d ebt o f m o n eta iy au th o rities (the m on etary base) resu lts in th e full K eynesian m u ltip lier “For a discussion of these points, see: Karl Brunner, “The Role of Money and Monetary Policy,” in the July 1968 issue of this Review. 5Tobin, pp. 143-213. OCTOBER 1986 effect. F in an cin g by e ith e r taxes o r b orrow ing from the p u blic h as a sm aller m u ltip lier effect o n spen din g. T o b in views th is d irect in flu en ce as tem porary. T h e in d irect in flu en ce o f fiscal actio n s, a cco rd in g to Tobin, resu lts from th e m a n n er o f fin an cin g th e gov ern m en t debt, th at is, variations in th e relative a m o u n ts o f d em an d debt, sh o rt-term debt, an d lon g term debt. F or exam ple, an ex p a n sio n a iy m ove w ould be a shift from lon g-term to sh o rt-term d eb t o r a shift from sh o rt-term to d em an d d ebt. A restrictiv e a ctio n w ould resu lt from a shift in th e o p p o site d irectio n . As in th e ca se o f m o n etary actio n s, m arket in tere st rates o n fin an cial a ssets an d th e ir in flu en ce o n in vestm en t spen din g m ake up th e tra n sm issio n m ech a n ism . T h e m o d ern qu an tity th eory also su ggests th at th e in flu en ce o f fiscal a ctio n s d ep en d s o n th e m eth o d o f financing governm ent ex p en d itu res. T h is ap p roach m aintain s th at fin an cin g ex p en d itu res by e ith e r taxing o r borrow ing from th e p u b lic involves a tran sfer o f co m m an d over re so u rces from th e p u b lic to th e gov ern m en t. However, th e n et in flu en ce on total s p e n d ing resu ltin g from in terest rate an d w ealth ch a n g es is am biguous. O nly a d eficit fin an ced by th e m o n eta iy system is n ecessa rily ex p an sio n aiy .'1 H igh-em ploym ent budget co n c e p ts have b een d e veloped as m easu res o f th e in flu en ce o f fiscal a ctio n s on ec o n o m ic activity.7 In th e se bu d g et co n c e p ts, e x p en d itu res in clu d e b o th th o se for good s an d seiv ices an d th o se for tran sfer paym ents, a d ju sted lo r th e in flu en ce o f e c o n o m ic activity. R eceip ts, sim ilarly a d ju sted , prim arily reflect legislated ch a n g es in federal governm ent tax rates, in clu d in g Social Secu rity taxes. 6The importance of not overlooking the financial aspects of fiscal policy is emphasized by Carl F. Christ in “A Simple Macroeconomic Model with a Government Budget Restraint,” Journal of Political Economy, Vol. 76, No. 1, January/February 1968, pp. 53-67. Christ summarizes (pages 53 and 54) that “the multiplier effect of a change in government purchases cannot be defined until it is decided how to finance the purchases, and the value of the multiplier given by the generally accepted analysis [which ignores the government budget restraint] is in general incorrect... (the) multiplier effect of govern ment purchases may be greater or less than the value obtained by ignoring the budget restraint, depending on whether the method of financing is mainly by printing money or mainly by taxation.” 7See Keith M. Carlson, “Estimates of the High-Employment Budget: 1947-1967,” in the June 1967 issue of this Review. The highemployment budget concept was used in the Annual Report of the Council of Economic Advisers from 1962 to 1966. For a recent analysis using the high-employment budget, see "Federal Fiscal Policy in the 1960s,” Federal Reserve Bulletin, September 1968, pp. 701-18. According to this article, “the concept does provide a more meaningful measure of the Federal budgetary impact than the published measures of actual Federal surplus or deficit taken by themselves.” 33 FEDERAL RESERVE BANK OF ST. LOUIS T h e net o f re ceip ts and ex p en d itu res is u sed as a n et m easu re o f ch an g es in ex p en d itu re provisions and in tax rates. T h e se h ig h -em p loy m ent co n c e p ts are u sed in th is article as m e a su res o f fiscal a ctio n s (exhibit 2). T e sts w ere also m ad e alternatively u sing natio n al in co m e a cc o u n t governm ent ex p en d itu res an d re ceip ts, a series m easu rin g au to n o m o u s ch an g es in governm ent tax rates, a w eigh ted h ig h -em p loy m ent ex p en d itu re an d receip t series, an d a series o f U. S. g overnm ent d ebt h eld by the p u b lic plus Fed eral Reserve h old ings o f U. S. governm ent secu rities. T h ese tests did not ch a n g e th e c o n c lu sio n s re a ch ed in this article. R esu lts o f th e se tests are available o n req u est. Other Influences M easu res o f o th e r in d e p en d e n t fo rces w h ich in flu en ce e c o n o m ic activity are n o t u sed in th is article. Yet th is sh o u ld n o t b e co n stru e d to im ply th at th e se fo rces are not im p ortant. It is a cc e p te d by all e c o n o m ists th at th e n o n -m o n e ta iy an d n o n -fiscal forces listed in exh ibit 1 have an im p o rtan t in flu en ce on e c o n o m ic activity. However, reco g n itio n o f th e e x ist e n c e o f th ese “o th e r fo rce s” d o es not p reclu d e the testin g o f p ro p o sitio n s relating to th e relative im p or ta n c e o f m o n eta iy an d fiscal forces. T h e analysis p re sen ted in th is stud y provides in d irect evid en ce b e a r ing o n th e se “o th e r fo rces.” T h e in tere ste d re ad e r is en co u rag ed to read th e te ch n ica l n o te p resen ted in th e ap p en d ix to this article before p roceed in g. TESTING TH E PRO PO SITIO N S T h is sectio n rep o rts th e resu lts o f testin g th e th ree p ro p o sitio n s u n d e r co n sid eratio n . First, th e co n c e p t o f testin g a h y p o th esis is briefly d iscu ssed . Next, th e resu lts o f reg ressio n analy ses w h ich relate th e m e a su res o f fiscal an d m o n etaiy a ctio n s to total sp en d in g are rep o rted . Finally, statistics d eveloped from the reg ressio n analy ses are u sed to test th e sp ecific p ro p o sition s. The Concept o f Testing a Hypothesis In scien tific m ethod ology, testin g a h y p o th esis c o n sists o f th e sta tem e n t o f th e h yp oth esis, deriving by m ean s o f logic te stab le co n se q u e n ce s e x p e cte d from it, an d th e n taking o bserv ations from past ex p e rie n ce w h ich sh o w th e p re se n c e or a b se n c e o f th e ex p ected co n se q u e n ce s. If th e ex p e cted co n se q u e n ce s do n ot o ccu r, th e n th e h y p o th esis is said to be “n o t c o n firm ed ” by th e evid en ce. If, on th e o th e r han d , the e x p e cted co n se q u e n ce s o ccu r, th e h y p o th esis is said to b e “co n firm ed .” 34 OCTOBER 1986 It is im p o rtan t to keep the follow ing poin t in m ind. In scien tific testing, a h yp o th esis (or co n je c tu re ) m ay b e fou n d “n o t co n firm ed ” an d th erefo re refu ted as the ex p lan ation o f th e relatio n sh ip u n d e r ex am in ation . However, if it is fou n d to be “co n firm ed ,” th e h y p o th e sis ca n n o t be said to have b een proven true. In the latter case, how ever, th e h yp o th esis rem ain s an a c cep tab le pro p o sition o f a real w orld relatio n sh ip as long as it is fou nd to b e “co n firm ed ” in future tests." T h e results p resen ted in th is stud y all b e a r on w hat is co m m o n ly called a “red u ced form ” in ec o n o m ics. A red u ced -form eq u atio n is a derivable co n s e q u e n ce o f a system o f eq u atio n s w h ich m ay b e h y p o th esiz ed to rep resen t th e stru ctu re o f th e eco n o m y (i.e., a socalled stru ctu ral m odel). In o th e r w ords, all o f th e factors an d cau sal relatio n s w h ich d eterm in e total sp en d in g (GNP) are “su m m a riz e d ” in o n e eq u ation . T h is red u ced -fo rm eq u atio n p o stu la tes a ce rta in rela tion sh ip over tim e b etw een th e in d e p en d e n t variables an d th e d ep en d en t variable — to tal sp en d in g. Using ap p rop riate statistical p ro ced u res an d se le c te d m ea su res o f variables, it is p o ssib le to test w h e th e r o r not th e im p licatio n s o f th e red u ced -fo rm eq u atio n have o ccu rred in th e past. If th e im plied rela tio n sh ip s are not confirm ed , th e n th e rela tio n sh ip asserted by th e red u ced -form eq u atio n is said to have b een refu ted. However, not con firm in g th e red u ced form d o es not n ecessa rily m ean th at th e w h o le “m o d el,” an d all of th e factors an d ca u sa l relatio n s co n ta in e d in it, are denied . It m ay b e on ly th at o n e o r m o re o f th e stru c tu ral linkages o f th e m od el is in co rrect, o r th at the em p irical su rrogates c h o s e n as m e a su res o f m o n eta iy o r fiscal in flu en ce are n ot a p p rop riate.1 1 Frequ en tly on e e n c o u n te rs sta tem e n ts o r c o n je c tu res regarding factors w h ich are asse rte d to in flu en ce eco n o m ic activity in a sp e cific w ay. T h e se sta tem en ts take th e form o f red u ced -fo rm eq u atio n s, an d are so m etim es a ttrib u ted to various th e o rie s o f th e d eter m ination o f e c o n o m ic activity. As sta ted previously, th is stud y d oes n o t a ttem p t to test th e ca u sa l linkages by w h ich fiscal an d m o n etary a ctio n s in flu en ce total spending, b u t is co n c e rn e d only w ith th e co n firm a 8For a detailed discussion of testing hypotheses in reference to monetary actions, see Albert E. Burger and Leonall C. Andersen, “The Development of Testable Hypotheses for Monetary Manage ment,” a paper presented at the annual meeting of the Southern Finance Association, November 8,1968. It will appear in a forthcom ing issue of the Southern Journal of Business, University of Georgia, Athens, Georgia. 9A more specific statement relating to these considerations is pre sented in the appendix. FEDERAL RESERVE BANK OF ST. LOUIS tion o r refu tation o f rival co n je c tu re s regarding the stren gth and reliability o f fiscal an d m o n eta iy actio n s b ased on frequ ently u sed in d icato rs o f su ch actio n s. Measuring the Empirical Relationships As a step tow ard analyzing th e th ree p ro p o sition s put forth earlier, em p irical relatio n sh ip s betw een th e m easu res o f fiscal an d m o n eta iy a ctio n s an d total sp en d in g are estab lish ed . T h e se relatio n sh ip s are d e veloped by regressin g qu arter-to -q u arter ch an g es in GNP on qu arter-to -q u arter ch an g es in th e m on ey stock (M) and in th e various m e asu res o f fiscal actio n s: h ig h -e m p lo y m e n t b u d g e t s u r p lu s (R -E ), h ig h em p loy m ent ex p en d itu res (E), and h ig h -em p loy m ent receip ts (R). Sim ilar eq u atio n s w ere estim ated w here ch a n g es in th e m o n eta iy b ase (B) w ere u sed in p la ce o f th e m oney stock. C hanges in all variables w ere co m p u ted by two m eth o d s. C onventional fii-st d ifferen ces w ere c a lc u lated by su btractin g th e value for th e p reced in g qu ar te r from th e value for th e p re sen t qu arter.'" T h e o th e r m eth o d u sed is an averaging p ro ced u re used by Kareken an d Solow called cen tral d ifferen ces." T h e s tru c ture o f lags p resen t in th e regressio n s w as estim ated w ith u se o f th e Alm on lag tech n iq u e.'- T h e data are season ally ad ju sted qu arterly averages for th e period from th e first q u arter o f 1952 to the; seco n d q u arter o f 1 968.1 3 ,0Changes in GNP, R, and E are quarterly changes in billions of dollars measured at annual rates, while changes in M and B are quarterly changes in billions of dollars. Changes in GNP, R, and E are changes in flows, whereas changes in M and B are changes in a stock. Since all of the time series have strong trends, first differ ences tend to increase in size over time. Statistical considerations indicate that percent first differences would be more appropriate. On the other hand, regular first differences provide estimates of multipli ers which are more useful for the purposes of this study. Test regressions of relative changes were run and they did not alter the conclusions of this article. "John Kareken and Robert M. Solow, “Lags in Monetary Policy” in Stabilization Policies of the research studies prepared for the Com mission on Money and Credit, Prentice-Hall, Inc., 1962, pp. 18-21. 12Shirley Almon, “The Distributed Lag Between Capital Appropria tions and Expenditures,” Econometrica, Vol. 33, No. 1, January 1965, pp. 178-96. 13As a test for structural shifts, the test period was divided into two equal parts and the regressions reported here were run for each sub-period and for the whole period. The Chow test for structural changes accepted the hypothesis that the sets of parameters esti mated for each of the sub-periods were not different from each other or from those estimated for the whole period, at the five percent level of significance. As a result, there is no evidence of a structural shift; consequently, the whole period was used. OCTOBER 1986 As d iscu sse d previously, sta tem e n ts are frequ ently m ad e from w h ich ce rta in rela tio n sh ip s are ex p ected to exist b etw een m easu res o f e co n o m ic activity on the on e h and an d m easu res o f m o n eta iy an d fiscal actio n s on th e o th e r h an d . Su ch rela tio n sh ip s co n sist o f a d irect in flu en ce o f an a ctio n on CiNP an d o f an in d irect in flu en ce w h ich reflects in tera ctio n s am on g th e m any m arkets for real an d fin an cial a ssets. T h e se in te ra c tio n s w ork th rou gh th e m arket m e ch a n ism d eterm in ing th e d ep en d en t variables listed in exh ib it 1. T h e p o stu lated relatio n sh ip s are th e total o f th e se d irect an d in d irect in flu en ces. T h u s, th e em p irical rela tio n ship em b od ied in e a c h regressio n co efficien t is th e total re sp o n se (in clud in g b o th d ire ct an d in d irect re spon ses) o f GNP to ch a n g es in e a c h m easu re o f a stabilization actio n , assu m in g all o th e r fo rces rem ain co n sta n t. T h e resu lts p resen ted h ere do n o t provide a basis for sep aratin g th e d irect an d in d irect in flu en ces o f m o n eta iy an d fiscal fo rces on total spen din g, b u t this division is irrelevant for th e p u rp o ses o f th is article. T h e in tere ste d re a d e r is referred to th e ap p en d ix for fu rth er elab o ratio n o f th e se poin ts. Using th e total re sp o n se co n cep t, ch a n g es in GNP are ex p e cted to b e positively related to ch a n g es in th e m o n ey stock (M) o r ch a n g es in th e m o n eta iy b a se (B). W ith regard to th e h ig h -em p loy m en t su rp lu s (receip ts m inu s exp en d itu res), a larger su rp lu s o r a sm aller d eficit is ex p e cted to have a negative in flu en ce on GNP, an d conversely. C hanges in h ig h -em p loy m en t e x p en d itu res (E) are ex p e cted to have a positive in flu en ce an d ch a n g es in re ceip ts (R) are ex p e cted to have a negative in flu en ce w h en th e se variables are in clu d ed sep arately. C on sid erin g th at th e prirnaiy p u rp o se o f th is study is to m easu re th e in flu en ce o f a few m a jo r fo rces on ch a n g es in GNP, ra th er th an to identify an d m easu re th e in flu en ces o f all in d e p en d e n t forces, th e results o b tain ed are qu ite good (table II. T h e K- statistic, a m easu re o f th e p e rce n t o f th e v arian ce in ch a n g es in GNP exp lain ed by th e regressio n eq u ation , ranges from .53 to .73; th e se values are u sually co n sid ere d to be qu ite good w h en first d ifferen ces are u sed ra th er th an levels o f th e data. All o f th e estim ated regression co efficien ts fo r ch a n g es in th e m o n ey stock o r th e m o n eta iy b ase have th e signs im p lied in th e above d iscu ssio n (equ ation s 1.1 to 2.4 in tab le 1) an d have a high statistical sig n ifican ce in m ost ca ses. T h e esti m ated co efficien ts for th e h ig h -em p loy m en t m easu res of fiscal in flu en ce do not have th e ex p e cted signs in all ca ses an d gen erally are o f low statistical sign ifican ce. 35 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 1 Regression of Changes in GNP on Changes in Monetary and Fiscal Actions First Differences t t-1 t-2 t-3 Sum Constant R2 SE DW Central Differences t t-1 t-2 t-3 Sum Constant R2 SE DW (Equation 1.1) AM A(R-E) 1.57* -.1 5 (2.17) (.65) 1.94* -.20 (3.60) (1.08) 1.80* .10 (3.37) (.55) .47* 1.28 (1.88) (1.95) .22 6.59* (7.73) (.45) 1.99* (2.16) .56 4.24 1.54 (Equation 2.1) AM A(R-E) 1.50 -.2 4 (1.84) (.91) 2.11* -.2 3 (3.61) (1.16) .15 1.89* (3.18) (81) .52 1.06 (1.36) (1.90) .21 6.56* (8.16) (.47) 2.02* (2.48) .66 3.35 .88 AM 1.51* (2.03) 1.59* (2.85) 1.47* (2.69) 1.27 (1.82) 5.84* (6.57) 2.10 (1.88) .58 4.11 1.80 AM 1.58* (2.01) 1.57* (2.78) 1.41* (2.45) 1.26 (1.72) 5.80* (7.57) 2.00* (2.14) .72 3.03 1.14 (Equation 1.2) AE AR .36 .16 (1.15) (.53) .53* -.01 (2.15) (03) -.0 5 -.03 (.19) (.10) .11 -.78* (2.82) (32) .07 .23 (32) (.13) (Equation 1.3) AM AE 1.54* .40 (2.47) (1.48) 1.56* .54* (3.43) (2.68) 1.44* -.03 (3.18) (.13) -.74* 1.29* (2.00) (2.85) 5.83* .17 (.54) (7.25) 2.28* (2.76) .60 4.01 1.78 (Equation 2.2) AR AE .53 .32 (1.52) (1.05) -.0 4 .60* (2.44) (.17) -.11 -.1 5 (.60) (.47) -.96* .18 (3.15) (.48) .02 .34 (04) (.54) (Equation 2.3) AM AE 1.54* .63* (2.45) (2.21) 1.63* .59* (3.57) (2.61) 1.43* -.1 6 (.71) (3.16) 1.13 -.86* (3.07) (1.71) 5.74* .19 (.77) (8.45) 2.30* (3.55) .73 2.97 1.13 AB 1.02 (.49) 5.46* (3.37) 6.48* (4.10) 3.05 (1.54) 16.01* (5.67) 1.55 (1.22) .53 4.35 1.71 AB .61 (.28) 5.42* (3.16) 6.87* (3.92) 3.51 (1.71) 16.41* (6.95) 1.24 (1.14) .67 3.26 1.05 (Equation 1.4) AE AR .52 .23 (67) (1.68) .02 .37 (1.36) (07) -.1 7 -.21 (84) (.64) .14 -.93* (3.10) (39) -.5 4 .51 (.67) (89) (Equation 2.4) AE .28 (.73) .50 (187) -.2 7 (104) -1.26* (3.65) -.7 5 (1.37) AR .87* (2.55) -.0 7 (.27) -.3 3 (1.31) .35 (87) .82 (116) Note: Regression coefficients are the top figures, and their “t” values appear below each coefficient enclosed by parentheses. The regression coefficients marked by an asterisk (*) are statistically significant at the 5 percent level. R2 are adjusted for degrees of freedom. SE is the standard error of the estimate, and DW is the Durbin-Watson statistic. T h e se reg ressio n resu lts are d iscu sse d in g reater d e tail below . M o n ey a n d th e m o n eta ry b a s e — T h e total re sp o n se o f GNP to ch an g es in m o n ey o r th e m o n eta iy b ase d istribu ted over fou r q u arters is co n siste n t w ith th e p o stu lated relatio n sh ip (i.e., a positive relatio n ship), an d th e co efficien ts are all statistically signifi ca n t. T h e co efficien ts o f e a c h m e asu re o f m o n eta iy a ctio n m ay be su m m ed to provide an in d icatio n o f the overall re sp o n se o f GNP to ch an g es in m o n eta iy actio n s. T h e se su m m ed co efficien ts are also statisti 36 cally sign ifican t an d c o n siste n t w ith th e p o stu lated relatio n sh ip s. T h e resu lts o b ta in ed for m e a su res o f m o n eta iy a ctio n s w ere n ot affected sign ifican tly w h en m easu res o f fiscal a ctio n s o th e r th a n th o se rep o rted h ere w ere u sed in th e regressio n s. H ig h -em p lo y m en t b u d g e t s u r p lu s — As p o in ted out previously, th e h ig h -em p loy m en t su rp lu s o r d e ficit is often u sed a s a m e a su re o f th e d ire ctio n an d stren g th o f fiscal actio n s. E q u atio n 1.1 su m m arizes the total resp o n se o f GNP to ch a n g es in m o n ey an d ch a n g es in th e h ig h -em p loy m en t su rp lu s. T h e co ef FEDERAL RESERVE BANK OF ST. LOUIS ficien ts o f th e hig h -em p loy m en t su rp lu s estim ated for th e co n tem p o ran eo u s an d first lagged q u a rte r have th e ex p ected sign, bu t th e co efficien ts are o f veiy low statistical sign ifican ce an d do not differ significantly from zero. T h e signs o f th e co efficien ts estim ated for th e seco n d and third lagged q u arters are o p p o site to th e ex p ected signs. T h e su m o f th e co efficien ts (total resp o n se d istribu ted over fou r quarters) is estim ated to have a positive sign (op p osite th e p o stu lated sign) but is not statistically significant. T h e se resu lts p ro vide no em p irical su p p o rt for th e view that fiscal actio n s m easu red by th e hig h -em p loy m ent su rp lu s have a significant in flu en ce o n GNP. fn p rin cip le, th ese resu lts m ay have o ccu rred eith e r b e c a u se th e highem p loy m ent su rp lu s w as n o t a good m easu re o f fiscal in flu ence, o r b eca u se fiscal in flu en ce w as not im p o r tan t during th e sam p le p e rio d .1 4 E x p e n d itu re s a n d re c e ip ts — Sim ple textbook K eynesian m o d els o f in co m e d eterm in atio n usually d em onstrate, th eoretically, th at ch an g es in tax rates exert a negative in flu en ce o n e c o n o m ic activity, w hile ch an g es in governm ent ex p en d itu res exert a positive in flu en ce. E qu atio n s 1.2 and 1.3 provide tests o f th ese p rop ositions. T h e signs o f th e co efficien ts estim ated for tax receip ts are th e sam e as th e h y p oth esized signs for only th e first and se co n d lagged qu arters. However, sin ce th ese co efficien ts (individually and the sum s) are o f low statistical sign ifican ce, n o im p o rtan ce ca n b e attach ed to this variable. In clu sio n o f ch an g es in receip ts (AR) in eq u atio n 1.2 d o es n o t im prove the overall results, in term s o f R2 and th e stan d ard erro r o f estim ate, co m p ared w ith eq u atio n 1.3 from w h ich receip ts are exclu ded . T h ese resu lts provide n o su p p o rt for th e o rie s w h ich in d icate that ch an g es in tax receip ts d ue to ch an g es in tax rates exert an overall negative (or any) in flu en ce on e co n o m ic activity. T h e resu lts are co n siste n t w ith th e ories w h ich in d icate th at if th e alternative to tax reve n u e is borrow ing from th e p u blic in o rd er to fin an ce governm ent spending, th en th e in flu en ce o f sp en d in g 14lt was suggested to the authors that a weighted high-employment budget surplus might be a better measure of fiscal influence than the usual unweighted series. For an elaboration of such a weighted series, see Edward M. Gramlich, “Measures of the Aggregate Demand Impact of the Federal Budget,” in Staff Papers of the President’s Commission on Budget Concepts, U.S. Government Printing Office, Washington, D.C., October 1967. Gramlich provided weights from the FRB-MIT model of the economy for constructing a weighted series. It was further suggested that the level of the highemployment budget surplus was a more appropriate measure of fiscal actions. Coefficients of fiscal influence were estimated using both changes in the weighted series, and levels of the highemployment surplus. The results did not change any of the conclu sions of this article. OCTOBER 1986 will n ot n ecessa rily be g reater if th e fund s are b o r row ed ra th er th an o b tain ed th rou gh taxation . They are also co n siste n t w ith th e th e o iy that co n su m ers will m ain tain co n su m p tio n levels at th e ex p en se o f saving w h en th ere is a tem p orary red u ctio n in d isp o s able in co m e. T h e signs o f th e co efficien ts estim ated for highem ploym en t ex p en d itu res in eq u atio n s 1.2 an d 1.3 in d ica te th at an in cre a s e in govern m en t ex p en d itu res is m ildly stim ulative in th e q u a rter in w h ich sp en d in g is in crea sed an d in th e follow ing qu arter. However, in th e su b seq u en t two qu arters this in crea se in e x p e n d i tu res ca u ses offsetting negative in flu en ces. T h e overall effect o f a ch a n g e in ex p en d itu res d istrib u ted over four quarter’s, in d ica ted by th e sum , is relatively sm all an d n ot statistically significant. T h e se resu lts are c o n sisten t w ith m o d ern qu an tity th eo ries w h ich h o ld that governm ent spen din g, taxing, an d borrow ing p o licies w ould have, th rou gh in tere st rate an d w ealth effects, different im p acts on ec o n o m ic activity u n d e r varying c irc u m sta n c e s.1 5 Three Propositions Tested T h e em p irical re la tio n sh ip s d eveloped relating ch a n g es in GNP to ch a n g es in th e m o n ey stock an d ch a n g es in h ig h -em p loy m ent ex p en d itu res an d re ceip ts are u sed to test th e th ree p ro p o sition s u n d e r co n sid eratio n . T h e resu lts o f testin g th e prop osition s u sing ch a n g es in th e m oney stock are d iscu sse d in detail in th is sectio n . Sim ilar resu lts are rep o rted in th e acco m p an y in g tab les u sin g ch a n g es in th e m o n e tary b ase in stea d o f th e m o n ey stock. C o n clu sio n s draw n using e ith e r m ea su re o f m on etary a ctio n s are sim ilar. P ro p o sitio n I sta tes th at fiscal a ctio n s exert a larger in flu en ce on e c o n o m ic activity th an do m on etary a ctio n s. A test o f th is p ro p o sition involves an ex am in a tion o f th e size o f th e reg ressio n co efficien ts for highem p lo y m en t ex p en d itu res r elative to th o se for- m on ey an d th e m on etary b ase."1P rop osition I im plies th at the 15John Culbertson points out that in a financially constrained economy (i.e., no monetary expansion to finance government expenditures), expenditures by the government financed in debt markets in compe tition with private expenditures can very possibly “crowd out of the market an equal (or conceivably even greater) volume that would have financed private expenditures.” He asserts that it is possible to have a short-lived effect of government spending on total spending if the financial offsets lag behind its positive effects. The results obtained for AE in this article are consistent with his analysis. See John M. Culbertson, Macroeconomic Theory and Stabilization Policy, McGraw-Hill, Inc., New York, 1968, pp. 462-63. '6Since little response of GNP to AR was found, further discussions consider only AE. 37 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 2 Measurements of the Relative Importance of Monetary and Fiscal Actions Beta Coefficients AB AR Quarter AM AE First Differences (equations 1 .2 and 1 .4 ) .14 .05 t .24 t-1 .26 .20 t-2 .24 -.0 2 -.01 .03 t-3 .20 - .30 .02 .07 Sum .94 Central Differences (equations 2 .2 and 2 .4 ) t .26 .20 .09 t-1 .26 .23 - .01 t-2 .23 - .06 - .03 .05 t-3 .20 - .36 Sum .95 .01 .10 AE AR Partial Coefficients of Determination AM AE AR AB AE AR .06 .31 .37 .17 .91 .09 .14 -.0 8 -.3 6 -.21 .16 .01 -.0 5 .04 .16 .07 .14 .12 .06 .45 .02 .08 * .13 * .01 * * * * * .18 .24 .04 .38 .01 .03 .01 .16 .02 .05 * .01 * .01 .04 .31 .40 .20 .95 .11 .19 -.1 0 -.4 7 -.2 7 .25 -.0 2 -.09 .10 .24 .07 .13 .11 .05 .53 .04 .10 .01 .16 * .02 * * * .01 * .16 .23 .05 .49 .01 .06 .02 .21 .04 .11 * .03 .01 .03 ’ Less than .005. co efficien ts for AE w ould be larger, w ithou t regard to sign, th a n th o se for AM an d AB. T h e co efficien ts p resen ted in table 1 are n ot ap p ro priate for th is test b ec a u se th e variables have different tim e d im e n sio n s and are a m ixture o f sto ck s and flow s. An ap p rop riate m easu re is developed by ch a n g ing th e se reg ressio n co efficien ts to “b eta co e fficie n ts” w h ich elim in ate th ese difficu lties (table 21. T h e se co e f ficien ts take in to co n sid eratio n the p ast variation o f ch a n g es in e a c h in d ep en d en t variable relative to the p ast variation o f ch an g es in GNP.1 T h e size o f b eta 7 co efficien ts m ay be, therefore, d irectly co m p ared as a m easu re o f th e relative co n trib u tio n o f e a c h variable to variations in GNP in th e test period. A ccord ing to table 2, th e beta co efficien ts for ch a n g es in m o n ey are g reater th an th o se for ch an g es in h ig h -em p loy m en t ex p en d itu res for th e q u arter in w h ich a ch an g e o cc u rs and during th e two follow ing qu arters. T h e co efficien ts for ch an g es in th e m o n eta iy b ase are g reater for th e tw o q u arters im m ediately follow ing a ch an g e in th e b ase. In th e lagged qu arters in w h ich th e b eta co efficien ts for AE are largest, a negative sign is asso cia te d w ith th e reg ressio n co e f ficien t, in d icatin g a lagged c o n tra ctio n a iy effect o f "Arthur S. Goldberger, Econometric Theory. John Wiley & Sons, Inc., December 1966, New York, pp. 197-200. 38 in crea sed ex p en d itu res. As a m easu re o f th e total co n trib u tio n over th e fou r qu arters, th e su m o f th e b eta co efficien ts for ch a n g es in m o n ey an d th e m o n e ta iy b ase are m u ch g reater th a n th o se for ch a n g es in exp en d itu res. Proposition I m ay also be tested bv th e u se o f partial co efficien ts o f d eterm in atio n . T h e se sta tistics are m e a su res o f th e p e rce n t o f variation o f th e d ep e n d en t variable rem ain ing after th e v ariation a c c o u n te d fo r by all o th e r variables in th e reg ressio n h as b e e n su b tra cted from th e total variation. P ro p o sitio n I im plies th at larger co efficien ts sh o u ld b e o b seiv ed fo r fiscal actio n s th an for m o n eta iy a ctio n s. T able 2 p re sen ts th e partial co efficien ts o f d eterm in a tio n for th e vari ab les u n d e r co n sid era tio n . F o r th e q u a rte r o f a ch an g e an d th e su b seq u en t tw o qu arters, th e se co efficien ts for AM are m u ch g reater th a n th o se for AE. W ith regard to AB, th e co efficien ts are abou t eq u al to th o se for AE in th e first q u a rter an d are m u ch g rea ter in the two su b seq u en t qu arters. T h e partial co e fficie n ts o f d eterm in atio n for th e total co n trib u tio n o f e a c h policy variable to ch a n g es in GNP over fo u r q u arters m ay b e d eveloped. T ab le 2 show s th a t th e partial co efficien ts o f d eterm in atio n for th e overall re sp o n se o f AGNP to AM an d AB ran ge from .38 to .53, w h ile th o se fo r AE are virtually zero. O th er im p licatio n s o f th e resu lts p resen ted in table 1 m ay b e u sed to test fu rth er th e relative stren g th o f th e re sp o n se o f GNP to alternative govern m en t a ctio n s FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 3 Simulated Response of an Increase in Government Expenditures Financed by Monetary Expansion (millions of dollars) Quarter 1 2 3 4 5 6 7 8 Increase in Government Expenditures Impact Cumulative Effect Effect Change in Expenditures on GNP on GNP $1000 0 0 0 - 1000 0 0 0 - $400 540 30 740 400 540 30 740 Required Increase in Money Impact Cumulative Change in Effect Effect Money Stock on GNP on GNP $400 940 910 170 - 230 - 770 - 740 0 u n d e r co n d itio n s w h ere “o th e r th in g s” are held c o n stant. T h ree alternative actio n s are assu m ed taken by stabilization au th o rities: (1) th e rate o f govern m en t sp en d ing is in crea sed by $1 billion an d is fin an ced by e ith e r borrow ing from the p u blic o r in creasin g taxes; (2) th e m o n ey sto ck is in crea sed by $1 billion w ith no ch an g e in th e budget positio n ; an d (3) th e rate of governm ent sp en d in g is in crea sed by $1 billion for a y ea r an d is fin an ced by in creasin g th e m on ey sto ck by an equal am ount. T h e im p act on total sp en d in g o f th e first two actio n s m ay be m easu red by u sing th e su m s o f th e regression coefficien ts p resen ted for eq u atio n 1.3. A $1 billion in crease in th e rate o f governm ent sp en d in g w ould, after four qu arters, resu lt in a p erm an en t in crea se o f $170 m illion in GNP. By co m p ariso n , an in cre a se o f the sam e m agnitu d e in m o n ey w ould resu lt in GNP being $5.8 billion p erm an en tly h ig h er after fou r qu arters. T h e resu lts o f the last actio n are p re sen ted in table 3.'“ T h e ann u al rate o f governm ent sp en d in g is a s su m ed to be in creased by $1 billion in th e first q u arter an d h eld at that rate for th e follow ing th ree qu arters. T his w ould requ ire an in crea se in m o n ey o f $250 m illion d uring e a c h o f th e fou r q u arters to fin an ce the h ig h er level o f exp en d itu res. Sin ce w e are in terested only in th e result o f finan cing th e original in crea se in 1The authors wish to give special thanks to Milton Friedman for B suggesting this illustration and table 3. However, the formulation presented here is the sole responsibility of the authors. $250 250 250 250 0 0 0 0 $ 385 775 1135 1458 1072 682 323 0 $ 385 1160 2295 3753 4825 5507 5830 5830 Total Response in GNP Impact Cumulative Effect Effect on GNP on GNP $ 785 1315 1105 718 672 142 353 740 $ 785 2100 3205 3923 4595 4737 5090 5830 ex p en d itu res by m o n eta iy exp an sio n , ex p en d itu res m u st b e red u ced by $1 b illion in th e fifth qu arter. If ex p en d itu res w ere h eld at th e h ig h er rate, m on ey w ould have to co n tin u e to grow at $250 m illion p er quarter. A ccord ing to table 3, GNP w ould rise to a p erm an en t level $5.8 b illion h ig h er th an at th e b eg in ning. T h is in crea se in GNP resu lts en tirely from m o n e tary expan sion . A ccord in g to th e se th ree tests, th e reg ressio n resu lts im plied by P rop osition I did n o t o ccu r. T h erefo re, th e p ro p o sition th at th e re sp o n se o f total d em an d to fiscal a ctio n s is g reater th a n th at o f m o n eta iy a ctio n s is n ot confirm ed by th e evidence. P ro p o sitio n l i hold s that th e re sp o n se o f eco n o m ic activity to fiscal a ctio n s is m o re p red ictab le th a n th e resp o n se to m o n eta iy in flu en ce. T h is im p lies th at the regression co efficien ts relative to th eir stan d ard errors (this ratio is called th e “t-value”), relatin g ch a n g es in E to ch a n g es in GNP, sh o u ld b e g rea ter th an th e co rre s p on d in g m easu res for ch a n g es in M an d in B. T h e g reater the t-value, th e m o re co n fid e n ce th ere is in th e estim a ted reg ressio n co efficien t, an d h en ce, th e g reater is th e reliability o f th e estim a ted ch a n g e in GNP resu lting from a ch a n g e in th e variable. T h e se tvalues are p resen ted in table 4. An exam in ation o f th is table in d ica tes g rea ter tvalues for th e regressio n co efficien ts o f th e tw o m o n e tary variables th an for th e fiscal variable, ex cep t for the third q u arter after a ch an g e. Also, th e t-values for th e 39 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Table 4 Measurement of Reliability of the Response of GNP to Monetary and Fiscal Actions (“t-values” of regression coefficients1 ) Quarter First Differences t t-1 t-2 t-3 Sum AM AE AR AB AE AR 2.03 2.85 2.69 1.82 6.57 1.15 2.15 0.19 2.82 0.13 0.53 0.03 0.10 0.32 0.32 0.49 3.37 4.10 1.54 5.67 0.67 1.36 0.84 3.10 0.89 1.68 0.07 0.64 0.39 0.67 Central Differences t t-1 t-2 t-3 Sum 2.01 2.78 2.45 1.72 7.57 1.52 2.44 0.60 3.15 0.04 1.05 0.17 0.46 0.48 0.54 0.28 3.16 3.92 1.71 6.95 0.73 1.87 1.04 3.65 1.37 2.55 0.27 1.31 0.87 1.16 ’t-values associated with equations 1.2,1.4, 2.2 and 2.4 in table 1. sum o f th e regression co efficien ts for AM an d AB are large, w hile th o se for AE are n o t statistically significant from zero. Sin ce th e reg ressio n resu lts im p lied by Proposition II did not appeal', th e p ro p o sition is not con firm ed. P ro p o sitio n III states th at th e in flu en ce o f fiscal a ctio n s o n e co n o m ic activity o cc u rs faster th an th at o f m o n etaiy actio n s. It is tested by exam in ing the c h a ra c te ristics o f th e lag stru ctu re in th e regressio n s. P rop o sition III im plies that beta co efficien ts for AE should b e g reater th an th o se for AM in th e q u arte r o f a ch an g e an d in th o se im m ed iately follow ing. It also im plies th at th e m ain resp o n se o f GNP to fiscal actio n s o cc u rs w ith in few er qu arters th an its re sp o n se to m o n eta iy a ctio n s. T h e b eta co efficien ts are plotted in th e ch a r ts .1 A ” ch an g e in th e m o n ey stock in d u ce s a large an d alm ost equal re sp o n se in each o f th e fou r q u arters. T h e larg est re sp o n se o f GNP to ch an g es in the m o n eta iy base 19The Almon lag structure was developed by using a fourth degree polynomial and constraining the coefficients for t-4 to zero. The regressions indicate that four quarters constitute an appropriate response period for both fiscal and monetary actions. Equations using up to seven lagged quarters were also estimated, but there was little response in GNP to fiscal and monetary actions beyond the three quarter lags reported. 40 o cc u rs in th e first an d se c o n d q u arters after a ch an g e. T h e b eta co efficien ts for ch a n g es in M are g rea ter th an th o se for ch a n g es in E for th e q u a rte r o f a ch a n g e an d th e follo w in g q u a rte r, in d ic a tin g c o m p a ra tiv e ly sm aller re sp o n se o f GNP to fiscal a ctio n s in th e se first tw o qu arters. M oreover, th e largest co efficien t for AE o cc u rs for th e third q u a rter after a ch an g e. T h e ex p ected regressio n re su lts im p lied by P ro p o si tion III w ere n o t found. T h erefo re, th e p ro p o sition th at th e m a jo r im p a ct o f fiscal in flu en ce on e c o n o m ic activity o cc u rs w ith in a sh o rte r tim e in teiv al th an m o n eta iy in flu en ce is n ot con firm ed . S u m m a ry — T h is sectio n tested th e p ro p o sitio n s that th e re sp o n se o f e c o n o m ic activity to fiscal a ctio n s relative to m o n eta iy a ctio n s is (I) larger, (II) m ore p red ictab le an d (III) faster. T h e resu lts o f th e tests w ere not co n siste n t w ith any o f th e se p ro p o sition s. C on sequen tly, e ith e r th e co m m o n ly u sed m e a su res o f fiscal in flu en ce do n ot co rrectly in d ica te th e d egree an d d ire ctio n o f su ch in flu en ce, o r th e re w as n o m ea su rable n et fiscal in flu en ce on to tal sp en d in g in th e test period. T h e test resu lts are co n s is te n t w ith an alternative set o f p ro p o sition s. T h e re sp o n se o f e c o n o m ic activity to m on etary a ctio n s co m p a red w ith th at o f fiscal a ctio n s is (I') larger, (II') m o re p red ictab le, an d (III') faster. It sh ould b e rem em b ered that th ese alternative FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 mm M easu res of Lag Response E q u a tio n 1.2 First Differences E q u a tio n 1.4 E quation 2.2 Central Differences Equation 2 .4 Beta coefficients are for changes in the money stock (AM), the monetary base (AB), high-employment expenditures (AE), and high-employment receipts (AR). These beta coefficients are calculated as the products of the regression coefficient for the respective variables times the ratio of the standard deviation of the variable to the standard deviation of GNP. 41 FEDERAL RESERVE BANK OF ST. LOUIS p ro p o sitio n s have n o t b een proven true, bu t th is is alw ays th e ca se in scien tific testin g o f h y p oth esized rela tio n sh ip s. N evertheless, it is asse rte d h ere that th e se alternative p ro p o sition s are ap p rop riate fo r the co n d u ct o f stabilizatio n policy until evid en ce is p re sen ted proving o n e o r m ore o f th em false. T h e re is a m a jo r qu alification to th e se statem en ts. Sin ce th e p ro p o sitio n s w ere tested u sin g th e period first q u arter 1952 to seco n d q u arter 1968, it is im p lic itly a ssu m ed in m aking th e se sta tem e n ts that th e gen eral en viron m en t prevailing in th e test period h old s for th e im m ed iate future. Implications for Economic Stabilization Policy R ejectio n o f th e th ree p ro p o sition s u n d e r ex am in a tion and a c c e p ta n c e o f th e alternatives offered c a n y im p ortant im p licatio n s for th e co n d u ct o f e c o n o m ic stab ilization policy. All o f th e se im p licatio n s p o in t to th e advisability o f g reater re lia n ce b ein g p laced on m o n eta iy a ctio n s th a n on fiscal actio n s. Su ch a reli a n ce w ould re p rese n t a m arked d ep artu re from m ost p resen t p ro ced u res. T h e finding th at sta tem e n ts w h ich assert that ch a n g es in tax rates have a significant in flu en ce on total sp en d in g are not su p p o rted bv th is em pirical investigation su ggests th at p ast efforts in th is regard have b ee n overly o p tim istic. F u rth erm o re, th e finding th at th e re sp o n se o f total sp en d in g to ch an g es in governm ent ex p en d itu res is sm all co m p ared w ith the re sp o n se o f sp en d in g to m o n e ta iy actio n s strongly su ggests that it w ould be m o re ap p rop riate to place g reater relian ce on th e latter form ol stabilization a ctio n . Find ing o f a stron g em p irical re latio n sh ip b etw een e c o n o m ic activity and e ith e r o f th e m easu res o f m o n e ta iy a ctio n s p o in ts to th e co n c lu sio n th at m o n eta iy a ctio n s can an d sh o u ld play a m ore p ro m in en t role in ec o n o m ic stabilization th an they have up to now . Fu rth erm o re, failure to reco gn ize th e se relatio n sh ip s ca n lead to u n d esired ch an g es in e c o n o m ic activity b eca u se o f th e relatively sh o rt lags and stron g effects attribu table to m o n etaiy actio n s. Evid ence w as found w h ich is c o n siste n t w ith the pro p o sition th at th e in flu en ce o f m o n eta iy a ctio n s on e co n o m ic activity is m o re ce rta in th a n th at o f fiscal actio n s. Sin ce m o n etaiy in flu en ce w as also fou nd to be stro n g er and to o p erate m ore quickly th a n fiscal in flu en ce, it w ould ap p e ar to b e inap p rop riate, for stabilization pu rp oses, for m o n eta iy au th o rities to 42 OCTOBER 1986 w ait veiy long for a d esired fiscal a ctio n to b e ad o p ted an d im p lem en ted . E vid ence fou n d in th is stud y su ggests th at the m oney sto ck is an im p o rtan t in d ica to r o f th e total th ru st o f stab ilization actio n s, b o th m o n etary an d fiscal. T h is poin t is argued on tw o g ro u n d s. First, ch a n g es in th e m on ey sto ck reflect m ainly w h at m ay be ca lled d iscretio n ary a ctio n s o f th e F ed eral Reserve System as it u ses its m a jo r in stru m en ts o f m o n etary m an ag em en t — o p en m arket tra n sa ctio n s, d isco u n t rate ch an g es, an d reserve requ irem en t ch a n g es. S e c ond, th e m on ey stock reflects th e jo in t a ctio n s o f the T re a su iy an d th e F ed eral Reserve System in fin an cin g new ly crea ted govern m en t d ebt. Su ch a ctio n s are based on d ecisio n s regarding th e m o n etiz a tio n o f new d ebt by Fed eral Reserve a ctio n s, an d T reasu ry d e c i sion s regarding ch an g es in its b a la n ces at Reserve banks an d co m m ercia l banks. A ccord in g to th is s e c ond point, ch a n g es in govern m en t sp en d in g fin an ced by m o n eta iy ex p an sio n are reflected in ch a n g es in th e m o n etary b a se an d in th e m o n ey stock. A n u m b er o f eco n o m ists m ain tain th at th e m a jo r in flu en ce o f fiscal a ctio n s re su lts o n ly if e x p en d itu res are fin an ced by m o n eta iy ex p a n sio n . In p ractice, the Fed eral Reserve d oes n o t buy secu rities from th e Gov ern m en t. In stead , its o p en m arket o p era tio n s an d o th e r a ctio n s provide fund s in th e m arkets in w h ich both th e governm ent an d private sector's borrow . T h e relatio n sh ip s exp ressed in table 1 m ay b e u sed to p ro ject th e ex p ected co u rse o f GNP, given a lte rn a tive assu m p tio n s about m o n eta iy an d fiscal a ctio n s. Su ch p ro jectio n s n ecessa rily a ssu m e th at th e en v iron m en t in th e period u sed for estim a tio n an d th e aver age relatio n sh ip s o f th e re ce n t p ast h o ld in th e future. T h e p ro jectio n s are n ot able to take in to co n sid era tio n the in flu en ces o f o th e r in d e p en d e n t forces; th erefore, they are n o t su itable for ex a ct fo recastin g p u rp o ses. However, th ey do provide a u seful m easu re o f m o n e tary an d fiscal in flu en ces on ec o n o m ic activity. An exam p le o f su ch p ro jectio n s u sin g eq u atio n 1.3 is p re sen ted in table 5. E qu atio n 1.3 related q u arter-to q u arter ch an g es in GNP to ch a n g es in th e m o n ey stock an d ch a n g es in h ig h -em p loy m en t ex p en d itu res, b o th d istrib u ted over four qu arters. A ssu m ptions u sed in co m p u tin g th e p ro je c tio n s of qu arterly ch a n g es in GNP rep o rted in table 5 in clu d e: (a) h igh -em p loym ent ex p en d itu res w ere p ro jecte d th rou gh th e se c o n d q u a rte r o f 1969 u n d e r th e a s su m p tio n th at federal sp en d in g in fiscal 1969 w ill be about 5 p e rce n t (or $10 billion) g rea ter th a n fiscal 1968; FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 (b) federal sp en d in g w as a ssu m ed to co n tin u e in creasin g at a 5 to 6 p ercen t rate in the first two qu arters o f fiscal 1970; an d (c) q u a rter-to -q u a rter ch a n g es in the m on ey stock w ere p ro jecte d from 111/68 to IV/69 for fou r alternative co n sta n t an n u al grow th rates for m oney: 2 p ercen t, 4 p ercen t, 6 p e rce n t, an d 8 p ercen t. Table 5 Projected Change in GNP with Alternative Rates of Change in Money Stock1 Assumed Rates of Change in Money Stock2 Quarter 1968/III3 IV 1969/1 II III IV 2% 4% 6% 8% 17.9 14.6 12.0 11.0 6.8 8.0 17.9 16.0 15.0 15.2 12.3 13.7 17.9 17.5 18.0 19.4 18.0 19.4 17.9 19.0 20.7 23.7 23.4 25.2 'First differences of quarterly data. All variables are in billions of dollars. Projections are based on coefficients of equation 1.3 in table 1. 2Assumed alternative rates of change in the money stock from III/68 to IV/69. Preliminary estimate by the Department of Commerce. T h e h ig h est grow th rate o f th e m on ey stock (8 p er cent) in d icates co n tin u ed rapid rates o f ex p an sio n in GNP d uring th e next five qu arters. T h e slow est grow th rate o f m o n ey (2 p ercen t) in d ica tes so m e slow ing o f GNP grow th in the fourth q u a rte r o f th is y e a r and fu rth er grad ual slow ing th rou gh o u t m o st o f n ext year. T h e p ro jectio n s in d ica te that if th e re cen t d e c e le r ated grow th in th e m o n ey stock (less th a n 4 p ercen t from Ju ly to O ctober) is co n tin u ed , an d grow th o f governm ent sp en d in g is at abou t th e rate in d icated above, the eco n o m y w ould probably re a ch a n o n in flatio n aiy grow th rate o f GNP in about th e third q u a rter o f 1969 an d w ould th e n a cc e le ra te slightly. T h e se p ro jectio n s, o f co u rse, m ake no assu m p tio n s regarding th e V ietnam War, strikes, ag ricu ltu ral situ a tions, civil d iso rd ers o r any o f the m any o th e r n o n c o n trol lable exo g en o u s forces. Appendix 1 T h e s p e cific h y p o th e sis u n d erly in g th e an aly sis in th is stu d y is e x p re sse d by th e fo llo w ing re latio n : If th is re la tio n (2) w ere e m p irica lly e stim a te d , th e follow ing w ou ld b e obtained:-' (1) Y = f (E, R, M, Z), (3) AY = a.AE + ouAR + a:,AM + a,AZ, w h e re: Y = total sp en d in g , E = a variable su m m a riz in g g o v ern m en t e x p e n d itu re a c tio n s ; R = a variable su m m a riz in g g o v ern m en t taxin g a c tio n s ; M = a variable su m m a riz in g m o n e ta iy a c tio n s ; Z = a variable su m m a riz in g all o th e r fo rees th a t in flu e n c e total sp e n d in g .2 E x p re ssin g th is relatio n in te rm s o f th e c h a n g e s o f e a ch variable y ie ld s: (2 ) Ay = f (Ae, a r , Am, a z ) . 1The authors would like to give special thanks to Karl Brunner for useful discussion regarding the points made in this note. 2See exhibit 1 for a listing of "other forces” which influence total spending. w h ere th e valu es fo r a „ a ,, a „ an d ot4 are re g re ssio n o f th e o b serv ed v alu es o f AY o n valu es o f AE, AR, AM, a n d AZ. In (3) th e c o e fficie n ts ( a ’s) are th e to tal re s p o n s e o f AY e a c h o f th e fo u r in d e p e n d e n t variab les. e stim a te d by th e o b se iv e d v alu e o f the to c h a n g e s in As d is c u ss e d in th e text, tim e se rie s fo r E, R, a n d M have b e e n s e le c te d o n th e b asis o f fre q u en tly u sed in d ic a to rs o r m e a su res o f fiscal a n d m o n e ta iy a c tio n s . T h e p u rp o se o f th is stu d y w as to te st so m e fre q u en tly e n c o u n te re d rival c o n je c tu r e s reg ard in g th e in flu e n c e o f fiscal an d m o n e ta iy fo rce s o n e c o n o m ic activity, n o t to q u an tify all fo rc e s in flu e n cin g o u r e co n o m y . T h e re fo re , a tte n tio n h e re h a s b e e n d ire c te d tow ard e stim a tin g th e m a g n itu d e an d sta tistic a l 3For purposes of this note the lags of the independent variables are ignored. 43 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS reliability o f th e re s p o n s e o f AY to AE, AR, a n d AM. H owever, AZ c a n n o t b e sim p ly ign o red . T h e re a d e r w ill n o te th at th e re is n o c o n sta n t te rm in eq u a tio n (3) s in c e th e e ffe ct o f "all o th e r fo rc e s " in flu e n cin g sp e n d in g are su m m arize d by a,A Z. H owever, in th e re su lts re p o rte d in tab le 1 o f th is stu d y, a c o n s ta n t te rm is re p o rte d fo r e a c h e q u a tio n . T h e s e c o n sta n t te rm s a re an e stim a te o f a , tim e s th e average a u to n o m o u s n o n -m o n e ta iy a n d n o n fiscal fo rce s s u m m a riz e d in Z. In a co m p le x m arket e co n o m y , it is p o ssib le fo r m o n e ta iy an d fiscal a c tio n s to exert an in d ire c t a s w ell as a d ire c t in flu e n ce on AY. T h is in d ire c t in flu e n c e w o u ld o p e ra te th ro u g h AZ. O ne fo im o f th e re la tio n b e tw e e n AZ an d m o n e ta iy an d fiscal fo rc e s is sh o w n by: (4) AZ = b„ + b,AK + b,AR + b:,AM. T h e e m p irica l v alu es o f a „ a , a n d a „ w h ic h w e re e sti m ated by re g re ssio n an aly sis an d re p o rte d in th is stud y, e m b o d y b o th th e d irec t an d th e in d ire c t re s p o n s e s o f total sp e n d in g to m o n e ta iy an d fiscal a c tio n s . U sing AE as an exam p le, th e e x p re ssio n (a, + b ,a j is an e stim a te of a „ the to tal re s p o n s e o f AY to AE. T h e d ire c t re s p o n s e is a,, a n d th e in d ire ct re s p o n s e is b ,a 4. C o n se q u e n tly , th e e q u a tio n e s ti m ated an d re p o rte d in th is stu d y (for e xam p le, e q u a tio n 1.2 in tab le II is: (51 AY = b„a, + (a, + b ,a 4)AE + la, + b,a,)A R + (a:l + b ,a4)AM; w h ere b„a4 is th e “c o n s ta n t” re p o rte d in tab le 1. It it w ere kn ow n th a t b „ b , a n d b , ar e z e ro , it co u ld b e c o n c lu d e d th at th e re are n o in d ire c t e ffe cts o f m o n e ta iy an d fiscal fo rces o p e ra tin g th ro u g h Z o n Y, o n ly d ire c t e ffe cts w h ic h are 44 m e a su re d by a,, a, an d a,. S in ce th is c a n n o t b e e sta b lish e d co n clu siv e ly , it c a n n o t b e ru le d o u t th a t AZ m ay in c lu d e so m e in d ire c t m o n e ta iy a n d fiscal fo rc e s in flu e n c in g e c o n o m ic activity. T h e c o n s ta n t term is e stim a te d to b e q u ite large an d sta tistic a lly sig n ifican t. T h is p ro v id es in d ire c t e v id e n c e th a t AZ is e x p la in e d to so m e e x te n t b y fa c to rs o th e r th a n AE, AR, an d AM. T h e valu e o f b„a, is a m e a su re o f th e average effect o f “o th e r fo r c e s ” o n AY, w h ic h o p e ra te th ro u g h AZ. As a n o th e r te st o f th e in d e p e n d e n c e o f AZ from m o n e ta iy a n d fiscal fo rc e s, th e total tim e p erio d w as d ivided in to tw o s u b -sa m p le s an d th e e q u a tio n s w e re e stim a te d fo r th e s e s u b -sa m p le s. T h e C h ow te st (see text) w as a p p lied to th e s e ts o f re g re ssio n c o e fficie n ts e stim a te d from th e s u b -s a m p les c o m p a re d to th e w h o le sa m p le ; th e h y p o th e sis th at th e re w e re n o s tru c tu ra l sh ifts in th e tim e p e rio d c o u ld not b e re je c te d , im p lyin g no c h a n g e in th e siz e o f b„a4. If th e re w ere a sig n ifican t in d ire c t in flu e n c e o f AE, AR, a n d AM o p e ra tin g th ro u g h AZ, b„a4 w o u ld c h a n g e a lo n g w ith c h a n g e s in th e s e in d e p e n d e n t v ariab les. S in ce th is in te r c e p t w as fo u n d to b e s ta b le ov er th e te st p erio d , th is p ro v id es fu rth e r e v id e n ce th at AZ is in flu e n c e d b y fa c to rs o th e r th a n m o n e ta iy a n d fiscal fo rc e s. T h e re s u lts from th e s u b -s a m p le s in d ic a te th a t th e re w ere d iffe re n c e s in th e relativ e v ariability o f th e in d e p e n d en t v ariab les b e tw e e n th e tw o s u b -s a m p le s. T h is te n d s to s tre n g th e n th e c o n c lu s io n s o f th is a rticle s in c e th e re s p o n s e o f AGNP to AM o r AB w as g re a te r ev en in th e first s u b -sa m p le (1/53 to 1/60) in w h ic h th e variability o f AM an d AB w as s m a lle r th a n th e variability o f AE an d AR. T h e follow ing article is re p rin te d fro m the A pril 1970 F e d e ra l R eserv e Bank o f St. Louis Review. A M onetarist Model for Econom ic Stabilization Leonall C. Andersen and Keith M. Carlson T HE m o n etarist view th at ch an g es in th e m o n ey stock are a prim ary d eterm in an t o f ch an g es in total spending, and should th ereb y b e given m a jo r e m p h a sis in e co n o m ic stabilization program s, h as b ee n o f grow ing in terest in re ce n t y ears. From th e m id -1980s to the m id-1960s, m o n etaiy p o licy received little e m p h asis in e co n o m ic stabilization policy. P resu m ed failure o f m on etary p olicy d uring th e early y ears o f the G reat D epression , along w ith th e d evelopm ent and gen eral a cc e p ta n ce o f K eyn esian eco n o m ics, resu lted in a m ain em p h asis on fiscal a ctio n s — fed eral gov ern m en t sp en d in g an d taxing program s — in e c o n o m ic stabilization plans. M onetary policy, in so fa r as it received any atten tio n , w as gen erally ex p ressed in term s o f m arket rates o f in terest. Growing recogn ition o f th e im p o rtan ce o f m o n ey an d o th e r m o n etary aggregates in th e d eterm in atio n o f spending, output, and p rice s h as b ee n fostered by th e ap p arent failure o f stabilizatio n p o licy to cu rb th e inflation o f th e last h a lf o f th e 1960s. Sharply rising m arket in terest rates w ere in terp reted to in d icate significant m o n etaiy restraint, w hile th e R evenue an d E xp en d itu re C ontrol Act o f 1968 w as co n sid ere d a m a jo r m ove tow ard fiscal restraint. Offering helpful suggestions throughout the study were Denis Karnosky of this Bank, William P. Yohe of Duke University and Visiting Scholar at this Bank, 1969-70, and David Fand of Wayne State University. Susan Smith provided programming assistance and Chris topher Babb and H. Albert Margolis advised on statistical problems. The authors thank the following for their comments on earlier drafts, without implying their endorsement of either the methods of analysis or the conclusions: F. Gerard Adams, Philip Cagan, E. Gerald Corrigan, Richard Davis, Ray Fair, Edgar Fiedler, Milton Friedman and members of the Money and Banking Workshop at the University of Chicago, Edward Gramlich, Harry G. Johnson, John Kalchbrenner, Edward Kane, Michael Keran, Allan Meltzer, Franco Modigliani, George Morri son, David Ott, Joel Popkin, Thomas Saving, Roger Spencer, Henry Wallich, Clark Warburton, Manfred Willms, and Arnold Zellner. D espite th ese policy d evelopm en ts, total sp en d in g co n tin u ed lo rise rapidly u ntil late 1969, an d th e rate of inflation a ccelera ted . T h o se h old ing to th e m o n etarist view w ere n ot su rp rised by th is lack o f s u c c e s s in cu rb in g excessive grow th in total spen din g, largely b eca u se th e m on ey stock grew at a h istorically rapid rate d uring th e fou r y ea rs en d in g in late 1968. E c o n o m ic d evelop m en ts from 1965 th rou gh 1969 w ere in g en eral ag reem en t w ith th e ex p e cta tio n s o f th e m o n e tarist view. T h is article develops a m od el d esign ed to analyze eco n o m ic stab ilization issu e s w ithin a fram ew ork w h ich focu ses on th e in flu en ce o f m o n eta iy ex p a n sion on total spen din g. M ost o f th e m a jo r eco n o m etric m od els have n ot a ssign ed an im p o rtan t role to th e m on ey stock or to an y o th e r m o n eta iy aggregate.' Fu rth erm ore, m o st eco n o m etric m o d els co n ta in a large n u m b er o f behavioral h y p o th eses to b e em piri- 'Frank de Leeuw and Edward M. Gramlich, “The Federal ReserveMIT Econometric Model,” Federal Reserve Bulletin (January 1968), pp. 11-40, and “The Channels of Monetary Policy: A Further Report on the Federal Reserve MIT Econometric Model,” Federal Reserve Bulletin (June 1969), pp. 472-91; James S. Duesenberry, Gary Fromm, Lawrence R. Klein, and Edwin Kuh (ed.), The Brookings Quarterly Econometric Model of the United States (Chicago: Rand McNally, 1965), and The Brookings Model: Some Further Results (Chicago: Rand McNally, 1969); Michael K. Evans and Lawrence R. Klein, The Wharton Econometric Forecasting Model, 2nd Enlarged Edition (Philadelphia: University of Pennsylvania, 1968); Maurice Liebenberg, Albert A. Hirsch, and Joel Popkin, “A Quarterly Econo metric Model of the United States: A Progress Report," Survey of Current Business (May 1966), pp. 423-56; Daniel M. Suits, “The Economic Outlook for 1969,” in The Economic Outlook for 1969, papers presented to the Sixteenth Annual Conference on the Eco nomic Outlook at The University of Michigan (Ann Arbor: University of Michigan, 1969), pp. 1-26. For a discussion of the role of money in these models, see David I. Fand, “The Monetary Theory of Nine Recent Quarterly Econometric Models of the United States,” forth coming in the Journal of Money, Credit, and Banking. 45 FEDERAL RESERVE BANK OF ST. LOUIS cally estim ated and in tegrated w ith ea ch oth er, b e ca u se th ey are d esigned to aid in u n d erstan d in g th e d eterm in atio n o f m any ec o n o m ic m ag n itu d es. By co m p ariso n , th e m odel p resen ted in th is article is qu ite sm all. It is d esigned to provide in form ation on th e m o st likely co u rse o f m ovem ent o f ce rta in strategic eco n o m ic variables in re sp o n se to m o n eta iy an d fiscal actio n s. T h e m o d el p re sen ted h ere is th e a u th o rs’ ow n ver sion o f h o w m o n etary an d fiscal a ctio n s in flu en ce th e eco n om y. O ther eco n o m ists (includ ing th o se o f a m o n etarist persu asion ) m ay p refer to develop certa in a sp e cts o f th e m o d el in a different w ay. Tw o su ch m o d ification s are p re sen ted in ap p en d ix C. T h e m od el is co n sid ere d o p en to revision, bu t is p resen ted at th is tim e w ith a view to stim u latin g o th e rs to jo in in quantifying relatio n sh ip s th at are g en erally a sso cia te d w ith the m o n etarist view. T h is article is divided in to five m a jo r sectio n s. A gen eral m o n etarist view o f th e re sp o n se o f spendin g, o utput, and p rices to m o n eta iy an d fiscal a ctio n s is su m m arized first. Next, th e sp e cific featu res o f th e m od el are d iscu sse d w ithin a form al fram ew ork o f analysis. S tatistical estim ates o f th e m o d e l’s p ara m e ters are p re sen ted in th e third se ctio n . T h e fou rth sectio n te sts th e p erfo rm an ce o f th e m o d el w ith sev eral d ynam ic sim u lation ex p erim en ts. Finally, by sim u latin g th e re sp o n se o f th e eco n o m y to alternative rates o f m o n e ta iy exp an sio n , an illu stration is p ro vided o f h o w th e m od el ca n b e u sed for cu rren t stab ilization analysis. GENERAL MONETARIST VIEW T h e gen eral m o n etarist view is th at th e rate o f m o n etary ex p an sio n is th e m ain d eterm in an t o f total spen ding, co m m o n ly m easu red by gross n atio n al p ro d u ct (GNP).’ C hanges in to tal sp end ing, in turn, in flu en ce m ovem en ts in output, em p lo y m en t an d th e gen eral p rice level. A b asic p rem ise o f this analysis is th at th e eco n o m y is b asically stab le an d n o t n e c e s sa r ily su b je ct to recu rrin g p erio d s o f severe re ce s sio n an d inflation. M ajor b u sin ess cy cle m ovem ents that have o ccu rred in th e p ast are attrib u ted prim arily to large sw ings in th e rate o f grow th in th e m o n ey stock. T h is view regarding aggregate e c o n o m ic relatio n sh ip s differs from prevailing view s w h ich co n sid e r 2General references on the monetarist view are Karl Brunner, "The Role of Money and Monetary Policy,” this Review (July 1968), pp. 9-24; David I. Fand, “Some Issues in Monetary Economics,” this Review (January 1970), pp. 10-27, and “A Monetarist Model of the Monetary Process," forthcoming in the Journal of Finance. 46 OCTOBER 1986 aggressive p o licy a ctio n s n ece ssa ry to p ro m o te sta b il ity. M o n etarists gen erally h o ld th at fiscal a ctio n s, in th e a b se n c e o f a cco m m o d ativ e m o n etary a ctio n s, ex ert little n et in flu en ce on total sp en d in g an d th erefo re have little in flu en ce on o u tp u t an d th e p rice level. G overnm ent sp en d in g u n a c co m p a n ie d by a c c o m m o dative m o n etary ex p an sio n , th a t is, fin a n ce d by taxes o r b orrow ing from th e p u blic, re su lts in a crow d ing ou t o f private ex p en d itu res w ith little, if any, n e t in crea se in total sp en d in g. A ch a n g e in th e m o n ey stock, on th e o th e r h an d , ex erts a stron g in d e p en d e n t in flu en ce on to tal sp en d in g. M o n etarists co n c lu d e th at a ctio n s o f m o n e ta iy a u th o ritie s w h ic h re su lt in ch a n g es in th e m o n ey sto ck sh o u ld b e th e m ain tool o f e co n o m ic stabilization . S in ce th e eco n o m y is c o n sid ered to b e basically stable, an d sin c e m o st m a jo r b u sin ess cy cle m ov em en ts in th e p a st have resu lted from in ap p ro p riate m ov em en ts in th e m o n ey stock, co n tro l o f th e rate o f m o n etary e x p a n sio n is th e m e a n s by w h ich e co n o m ic in stability ca n b e m in im ized . T h e th eo retica l h eritage o f th e m o n eta rist p o sitio n is th e qu an tity th eory o f m on ey.3 T h is th eo ry d ates b ack to th e cla ssica l eco n o m ists (particularly David Ricardo) in th e early 1800s. T h e q u an tity th eo ry in its sim p lest form is ch a ra cte riz ed as a rela tio n sh ip b e tw een th e sto ck o f m o n ey an d th e p rice level. C lassical eco n o m ists co n c e n tra te d on th e lo n g -ru n a sp e cts of th e qu an tity th eo ry in w h ich ch a n g es in th e m o n ey sto ck resu lt in ch an g es o n ly in n o m in al m ag n itu d es, like th e p rice level, b u t have n o in flu en ce o n real m agn itu d es like ou tp u t an d em p loy m en t. T h e qu an tity th eory o f m on ey in its m o d ern form reco g n izes th e im p o rtan t in flu en ce th at ch a n g es in th e m o n ey stock ca n have o n real m ag n itu d es in th e sh o rt run, w hile in flu en cin g o n ly th e p rice level in th e long run. T h e m o d ern qu an tity th eo ry p o stu la tes th at in th e sh o rt run a ch a n g e in th e rate o f grow th in m on ey is follow ed w ith a m o d erate lag by ch a n g es in total sp en d in g an d output, w h ile ch a n g es in th e p rice level follow w ith a so m ew h at lo n g er lag.4 T h e se 3The classic work on the quantity theory is Irving Fisher, The Pur chasing Power of Money (New York: Macmillan, 1911). For an extensive review of the quantity theory literature, see Arthur W. Marget, The Theory of Prices: A Re-examination of the Central Prob lems of Monetary Theory (New York: Prentice-Hall, 1938), volume II, pp. 3-133. “Many of the ideas prevalent in current monetarist doctrine can be found in the writings of O ^ k Warburton in the 1940s and early 1950s. Many of his im p la n t articles have been reprinted in his Depression, Inflation, and Monetary Policy, Selected Papers, 19451953 (Baltimore: The Johns Hopkins Press, 1966). See also Milton Friedman (ed.), Studies in the Quantity Theory of Money (Chicago: University of Chicago Press, 1958), and Lloyd W. Mints, Monetary Policy in a Competitive Society (New York: McGraw-Hill, 1951). OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Exhibit 1 Model in Algebraic Form (1) Total Spending Equation AY, = f, (AM,... AM,_„, AE, - AE,^„) (2) Price Equation AP, = f2 (D,... D,_n, APA,) (3) Demand Pressure Identity D, = AY, - (XF, - X,_,) (4) Total Spending Identity AY, = AP, + AX, (5) Interest Rate Equation R, = f3 (AM,, AX,... AX,_„, AP,, AP*) (6) Anticipated Price Equation AP‘, = f4 ( A P ... AP (7) Unemployment Rate Equation U, = f5 (G„ G,_,) (8) GNP Gap Identity „ xf, -x, G ,= x f, AY, = AP, = D, = AX, = R, = APA, = U, = G, = Endogenous Variables change in total spending (nominal GNP) change in price level (GNP price deflator) demand pressure change in output (real GNP) market interest rate anticipated change in price level unemployment rate GNP gap Exogenous Variables' AM, = change in money stock ,AE, = change in high-employment Federal expenditures XF, = potential (full-employment) output 'Other than lagged variables. ch an g es in total spending, o u tp u t and p rice s are in th e sam e d irectio n as th e ch an g e in th e rate o f m o n e taiy expansion . T h e m o d ern qu an tity th eo ry still a cc e p ts th e longrun p o stu lates o f its o ld er version. A ch an g e in th e rate o f m o n etary ex p an sio n in flu en ces only n o m in al m ag n itu d es in th e long run, nam ely, total sp en d in g (GNP) an d th e p rice level. Real m agnitu d es, n otably output a n d em ploym en t, are u n affected .3 Follow ing th e sh o rtrun re sp o n ses to a ch an g e in th e rate o f m o n eta iy growth, total sp end ing an d th e p rice level grow at rates d eterm in ed by th e rate o f in crea se in m oney, 5See Milton Friedman, “The Role of Monetary Policy,” American Economic Review (March 1968), pp. 1-17. w hile ou tp u t m oves tow ard an d resu m es a lon g-run grow th path. Su ch grow th in o u tp u t is little in flu en ced by th e rate o f m o n eta iy ex p a n sio n . In stead , it is d eter m in ed by grow th in th e e c o n o m y ’s prod uctive p o te n tial, w h ich d ep en d s on grow th o f natu ral reso u rces, cap ital stock, lab or force an d productivity. GENERAL FORM O F TH E M ODEL A sum m ary o f th e m o d el is p re sen ted in algebraic form in exhibit 1, along w ith a listing o f variables classified as to w h eth er th ey are en d o g en o u s or exo g e n o u s to th e m odel (for a g rap h ical illu stration o f the m odel, see ap p en d ix B). T h is g en eral form o f the m od el su m m arizes its essen tia l featu res, ignoring pro b lem s o f d im en sio n ality an d lag length. 47 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS Equations o f the Model E qu atio n (1) is th e total sp en d in g eq u atio n . T h e ch a n g e in to tal sp en d in g (AY) is sp ecified as a fu n ctio n o f cu rren t and p ast ch an g es in th e m o n ey sto ck (AM) an d cu rren t and past ch an g es in hig h -em p loy m ent fed eral ex p en d itu res (AE). T h is general sp ecifica tio n rep rese n ts th e re d u ce d form for th at cla ss o f s tru c tu res w h ich h as AM and AE as exo g en o u s variables. In th is form th e total sp en d in g eq u atio n rem ain s u n c o m m itted as to stru ctu re; it is p o ten tially co n siste n t w ith b o th K eynesian an d qu antity th eo ry m odels." (The m agn itu d e an d sign ifican ce o f th e estim ated p ara m e ters d eterm in e w h eth er th e d ata co n fo rm m ore clo sely to a K eynesian o r a qu antity theory.) E qu ation (2) sp ecifies th e ch an g e in th e p rice level (AP) as a fu n ctio n o f cu rre n t an d p a st d em an d p re s su res (D) an d an ticip ated p rice ch an g es (AP'). D em and p ressu re is d efined in eq u atio n (3) as th e ch an g e in total sp en d in g m inu s th e potential in crea se in ou tp u t (X1 — X). T h e p rice eq u atio n is an alternative to th e stand ard sh o rt-ru n Phillips cu iv e relatio n gen eralized to in clu d e ch an g es in to tal sp en d in g an d a n ticip a te d p rice s.7 (Set: append ix A for fu rth er d evelop m en t of this relationship.) E q u atio n (4) d efin es a ch an g e in to tal sp en d in g in term s o f its co m p o n e n ts, th e p art asso cia te d w ith ch a n g es in th e p rice level (AP) an d th e part asso cia ted w ith ch an g es in o u tp u t (AX).” W ith AY d eterm in ed by eq u atio n (1), and AP by eq u atio n (2), AX ca n b e derived from eq u atio n (4). E qu atio n (5) sp ecifies th e m arket rate o f in tere st (R) as a fu n ctio n o f cu rren t ch an g es in th e m o n ey stock (AM), cu rre n t an d p ast ch an g es in o u tp u t (AX), cu rren t p rice ch a n g e (AP), an d a n ticip ate d p rice ch an g e (AP'). T h e p rice a n ticip atio n s term is in clu d ed to cap tu re th e F ish er effect.9 T h e a n ticip ate d p rice fu n ctio n is 6For further discussion of the structural versus the reduced form of a model, see Michael Keran, “Monetary and Fiscal Influences on Economic Activity— The Historical Evidence,” this Review (Novem ber 1969), pp. 5-24; Edward M. Gramlich, “The Usefulness of Monetary and Fiscal Policy as Discretionary Stabilization Tools,” forthcoming in the Journal of Money, Credit, and Banking; and Richard G. Davis, “How Much Does Money Matter? A Look at Some Recent Evidence,” Federal Reserve Bank of New York Monthly Review (June 1969), pp. 119-31. 7See Roger W. Spencer, “The Relation Between Prices and Employ ment: Two Views," this Review (March 1969), pp. 15-21. 8The change in the price level, AP, and the change in output, AX, are defined in dollar units so that their sum is equal to the change in total spending, AY. 9For a detailed study of interest rates and the Fisher effect, see William P. Yohe and Denis S. Karnosky, “ Interest Rates and Price 48 defined in eq u atio n (6). A n ticip ated p rice ch a n g e (AP') in th e cu rre n t p eriod is a ssu m ed to d ep e n d on p ast p rice ch a n g es (AP). E q u atio n (7) is th e u n em p lo y m en t rate eq u a tio n a n d is a tran sfo rm atio n o f th e GNP gap (G), as d efin ed in eq u atio n (8), in to a m e a su re o f u n e m p lo y m en t re la tive to th e lab or force. T h is tran sfo rm atio n is b a se d on “O ku n’s L aw .''1 0 Workings o f the Model T h e w orkings o f th e m o d el are su m m arized by a flow diagram (exhibit 2). O nly variables in th e cu rre n t p erio d are sh o w n in th e diagram ; lagged variables, w ith th e ex cep tio n o f p a st ch a n g es in p rice s, are o m itted . T h e re la tio n sh ip th a t d e te rm in e s to tal sp en d in g is th e fu n d am en tal o n e am o n g th o se th at d eterm in e th e en d o g en o u s variables o f th e m od el. T o tal sp en d in g is d eterm in ed by m o n eta ry a ctio n s an d fiscal a ctio n s (federal sp en d in g fin a n ce d by taxes o r b orrow ing from th e p u blic), th o u gh n o d ire ct in fo r m atio n is provided as to how su ch a ctio n s affect spen din g. T h e ch a n g e in to tal sp en d in g is co m b in ed w ith p o ten tial (full em ploym ent) o u tp u t to provide a m e a su re o f d em an d p ressu re. A n ticip ated p rice ch an ge, w h ich d ep en d s on p ast p rice ch an g es, is co m b in ed w ith d em an d p ressu re to d eterm in e th e ch a n g e in th e p rice level. T h e total sp en d in g id en tity en a b les th e ch a n g e in output to be d eterm in ed , given th e ch a n g e in total sp en d in g an d th e ch a n g e in p rices. T h is m eth o d o f d eterm in in g th e ch a n g e in to tal sp en d in g a n d its division b etw een o u tp u t ch a n g e an d p rice ch a n g e differs from m o st e c o n o m e tric m o d els. A stan d ard p ra c tice in ec o n o m e tric m o d el bu ild in g is to d e te r m ine o u tp u t an d p rice s sep arately, th e n co m b in e th e m to d eterm in e total sp en d in g. T h e ch a n g e in output, th e ch a n g e in p rices an d in a n ticip a te d p rices, along w ith th e ch a n g e in th e m o n ey stock, d eterm in e m arket in tere st ra tes. T h e flow diagram show s th a t th e m arket in tere st rate d oes Level Changes, 1952-69,” this Review (December 1969), pp. 1838. '“Arthur M. Okun, “ Potential GNP: Its Measurement and Signifi cance,” 1962 Proceedings of the Business and Economic Statistics Section of the American Statistical Association, pp. 98-104. Okun’s Law relates the GNP gap to the unemployment rate as follows: XF, - X, = .03(U, - 4)X,. The number .03 is a productivity factor and 4 is defined as the unemployment rate consistent with full resource utilization. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Exhibit 2 Flow Diagram of Model n o t ex e rcise a d irect role in th e m o d el in th e d eterm i n atio n o f spendin g, output, and p rices. To d eterm in e th e u n em p loy m en t rate, th e ch an g e in o utput is first co m b in ed w ith p o ten tial o u tp u t to d eterm in e th e GNP gap relative to p o ten tial ou tp u t. T h e GNP gap is th e n tran sfo rm ed in to th e u n em p lo y m ent rate. Summary T h e m odel h as b ee n p resen ted in gen eral form to sh ow th e b asic linkages p o stu lated am o n g m oney, federal exp en d itu res, p rice s an d o u tp u t. T h e p u r pose o f th e follow ing statistical sectio n is to estim a te th e re sp o n se o f o u tp u t an d p rice s to m o n eta iy an d fiscal a ctio n s, n ot to test a h y p o th esized stru ctu re. T h e focu s is on th e re sp o n se in th e sh ort run — period s of tw o o r th ree y ears — b u t th e lon g -ru n p ro p erties o f th e m odel also are exam in ed . ESTIMATION O F TH E M ODEL T h e gen eral form o f th e m od el in d ica tes th o se vari ables th at are in clu d ed in ea ch eq u atio n . E stim atio n requires selectio n o f th e alg eb raic form o f the eq u a tio n s an d th e te ch n iq u e s to b e u sed in estim ation . E ach o f th e eq u atio n s o f th e m o d el is estim ated by ord inary least squ ares. Lag stru ctu res, w ith o n e ex c e p tion, are estim ated by th e Alm on lag te ch n iq u e . T h e rep o rted relatio n sh ip s reflect co n sid era b le ex p eri 49 FEDERAL RESERVE BANK OF ST. LOUIS m en tatio n w ith th e n u m b er o f lags an d th e degree o f th e p oly n om ial." T h e sam p le period starts w ith 1953 for th e sp end in g eq u atio n an d w ith 1955 for all the o th ers. T h e d ata are qu arterly and, w ith th e ex cep tio n o f in tere st rates, are season ally ad ju sted . C riteria u sed in th e selectio n o f th e eq u atio n s w ere m inim izin g th e stand ard erro r o f estim ate an d elim i natin g serial co rrelatio n in th e estim ated resid u als. In ad d ition, th e signs and statistical sig n ifican ce o f the estim ated co efficien ts received co n sid eratio n , along w ith th e p attern o f th e lag d istribu tion . S in ce th e se criteria frequ en tly co u ld n ot be satisfied sim u lta n e ously, an elem en t o f su bjectivity w as p re sen t in s e le c t ing th e “b e s t” eq uation. Total Spending T h e ch an g e in total sp en d in g is sp ecified as a fu n c tion o f cu rren t an d past ch an g es in th e m oney stock (dem and d ep o sits and cu rre n cy h eld by th e n o n b an k public) an d in h ig h -em p loy m ent F ed eral e x p e n d i tu res (exp end itu res on good s an d services p lu s tra n s fer pay m en ts a d ju sted to rem ove th e in flu en ce o f variations in e co n o m ic activity on u n em p loy m en t ben efit paym ents). T h e c h o ice o f th e p a rticu la r eq u a tion (table 1) is b ased on previous w ork by A nd ersen an d Jo rd a n .'2 Im p licit in th is c h o ice is th e a ssu m p tio n th at th e ch an g e in th e m oney sto ck is an exo g en o u s variable. A m o re co m p lete m odel w ould sp ecify a m e ch an ism w h ereby th e m o n ey stock is d eterm in ed by a ctio n s o f th e m o n etary au thorities, th e pu blic, an d th e banking system . T h e p attern o f th e co efficien ts in d icates a large an d rapid in flu en ce o f m o n eta iy a ctio n s on to tal sp en d in g relative to that o f fiscal a ctio n s.1 C hanges in high3 em p lo y m en t exp en d itu res, w ith th e m o n ey stock h eld co n stan t, first have a posiuv'e in flu en ce on total sp en d - "For discussion of the use and interpretation of the Almon lag technique, see Keran, p. 10. ,2Leonall C. Andersen and Jerry L. Jordan, “Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabiliza tion," this Review (November 1968), pp. 11-24. See also Keran, pp. 5-24. l3Andersen and Jordan tried several measures of fiscal actions in their basic equation. The best results were obtained by using only high-employment expenditures, rather than the high-employment surplus or both high-employment expenditures and receipts. They justify their choice by appealing to the notion that financing expendi tures by borrowing from the public and taxes have essentially the same impact on total GNP. For some results that contradict those of Andersen and Jordan, see E. G. Corrigan, “The Measurement and Relative Importance of Fiscal Policy,” forthcoming in Federal Re serve Bank of New York Monthly Review. It should be repeated that, a priori, specification of the total spending equation was sufficiently general as to be consistent with a number of theories of GNP determination. 50 OCTOBER 1986 Table 1 Total Spending Equation Sample Period: I/1953-IV/1969 Constraints: 4th Degree Polynomial (m_, = e_, = 0; m5 = e6 = 0) =2.67 + 2 m.AM,., + 2 e^Et-i (3.46) i = 0 i= 0 m0 = 1.22 m, = 1.80 m2 — 1.62 m3 = .87 m4 = .06 2m; = 5.57 (2.73) (7.34) (4.25) (3.65) ( 12) (8.06) R2 = .66 SE = 3.84 DW = 1.75 e0 = .56 ( 2.57) e, = .45 ( 3.43) 62 ~ .01 ( .08) ®3 = -.4 3 ( -3.18) e* = -.5 4 ( - 2.47) Se, = .05 ( .17) Symbols are defined as: AY, = dollar change in total spending (GNP in current prices) in quarter t AM,_i = dollar change in money stock in quarter t - i AEt_, = dollar change in high-employment federal expendi tures in quarter t - i NOTE: “t” statistics appear with each regression coefficient, enclosed by parentheses. R2 is the percent of variation in the dependent variable which is explained by variations in the independent variables. SE is the standard error of the estimate. DW is the Durbin-Watson statistic. ing, bu t th e in flu en ce b e c o m e s sign ifican tly negative after th ree qu arters. F isca l actio n s, u n a c co m p a n ie d by ch a n g es in m oney, have little n et effect o n GNP over five q u a rters.1 F o r sh o rt p eriod s, an d for ex ten d ed 4 period s in w h ich th e rate o f ch a n g e o f fed eral ex p e n d itu res is eith e r a ccelera tin g o r d eceleratin g , fiscal ef fects are significant. T h e estim a ted co e fficie n ts for ch a n g es in m o n ey an d ch a n g es in fed eral e x p e n d i tu res are in g en eral ag reem en t w ith th e m o n eta rist view o f th e re sp o n se o f to tal sp en d in g to th e se two variables. T h e sp ecifica tio n o f the total sp en d in g eq u atio n , as sh ow n in table 1, has b een criticiz ed as b ein g in c o m plete in th at it allegedly ig n ores th e effects o f in tere st rates o n v elo city .1 However, sin c e th e sp e n d in g eq u a 5 l4Andersen and Jordan, p. 18, indicate that these results are consist ent with a “crowding-out” theory of effects of government spending. 15See Paul S. Anderson, “Monetary Velocity in Empirical Analysis,” in Controlling Monetary Aggregates, Proceedings of the Monetary Conference held on Nantucket Island (June 1969), pp. 37-51, and the discussion of that paper by Leonall C. Andersen, pp. 52-55. See also Henry A. Latane, “A Note on Monetary Policy, Interest Rates and Income Velocity,” Southern Economic Journal (January 1970), pp. 328-30. FEDERAL RESERVE BANK OF ST. LOUIS tion is a red u ced form , su ch effects ait; em b od ied in th e coefficien ts o f m o n ey ."1 Total Spending Equation T h e total sp en d in g eq u atio n is th e co rn e rsto n e o f the m odel, providing its m o n etarist ch a ra cte r. T h e focu s o f this paper, how ever, is on d eterm in in g th e division o f th e ch an g e in total sp en d in g b etw ee n p rice an d ou tp u t ch an g es. P rice ch an g es are estim ated as a fu n ctio n o f (1) cu rre n t and p ast d em an d p ressu re, an d (2) an ticip ated p rice ch an g e. D e m a n d p r e s s u r e — As a m e asu re o f d em an d p re s su re on p rices, th e ch an g e in total sp en d in g is related to th e potential ch an g e in o u tp u t (GNP in co n sta n t p rice s).1 T h e se tw o variables, w h en co m b in ed , p ro 7 vide a m easu re o f th e eco n o m y 's d em an d for goods an d services relative to its cap acity to su pply goods an d services. T h e ch an g e in p rices is sp ecified as a positively related lin ear fu n ctio n o f th is m easu re o f d em an d p ressu re (see ap p en d ix A). D em and p ressu re, D„ is defined as: D, = AY, - (X1, - X,_,), w h ere AY, is th e ch an g e in total sp en d in g in q u arter t; XK is p o ten tial (full em ploym ent) GNP in 1958 p rice s in , q u a rter t; an d X,_, is real GNP in th e previous quarter."* Given th e GNP gap, d efined as X1, — X,_,, th e larger is th e ch an g e in total sp en d in g (AY,), th e g reater is th e spillover into h ig h er p rices. Given AY„ th e larger is X', — X, _ „ the greater is th e ex p an sio n o f o u tp u t and th e less th e spillover into h ig h er p rices. In ad d itio n to cu rre n t values, p ast values o f th e d em an d p ressu re variable are in clu d ed in th e p rice eq u ation . T h e p u rp o se o f in clu d in g p ast values is to allow for lags in th e d eterm in atio n o f p rices in re 16See A. A. Walters, “Monetary Multipliers in the U. K.: 1880-1962,” Oxford Economic Papers (November 1966). "This measure was apparently first used by Ray Fair of Princeton University. See his “The Determination of Aggregate Price Changes,” forthcoming in the Journal of Political Economy. For a similar specification of a price equation, see Milton Friedman, “A Theoretical Framework for Monetary Analysis,” also forthcoming in the Journal of Political Economy. See also a paper by William Considine of Stanford University, “Public Policy and the Current Inflation,” prepared as a part of a summer intern program at the U.S. Treasury Department (September 5,1969). 18The series on potential output is based on that used by the Council of Economic Advisers. Currently, potential output is estimated to be rising at a 4.3 percent annual rate. For alternative estimates of potential output, see Fair, “The Determination of Aggregate Price Changes.” OCTOBER 1986 sp o n se to ch an g in g d em an d . F u rth erm o re, th e im p act of ch an g in g d em an d th rou gh ch an g in g in p u t p rices an d co sts o f p ro d u ctio n is given a c h a n c e to o p erate by in clu d in g lagged values for th e d em a n d p ressu re variable. A nticip ated p r ic e c h a n g e — T h e o th e r in d e p en d e n t variable in clu d ed in th e p rice eq u atio n is a m easu re o f a n ticip a te d p rice ch an g e (AP\). T h e p u rp o se o f in clu d in g th is variable as a fa cto r in flu en cin g cu rren t ch a n g es in th e p rice level is to allow a n ticip a tio n s o f future p rice m ovem en ts to in flu en ce th e d ecisio n s o f m arket p articip an ts. Sin ce su ch a variable is n o t o b servable, it h a s to b e co n stru c te d . T h is is a c c o m p lish ed by assu m in g th at a n ticip a tio n s abou t future p rice ch a n g es are form ed on th e b asis o f p ast p rice ex p erien ce. T h e m easu re o f p rice a n ticip a tio n s u sed in th is stud y is a b y -p rod u ct o f estim atin g lon g-term m arket in terest ra te s.11Yohe an d K arnosky sh o w ed th a t lon g 5 term m arket in terest rates resp o n d to p rice a n ticip a tio n s o f borrow ers an d len d ers, sin ce co m m itm e n ts to borrow an d len d fu n d s requ ire an a sse ssm en t o f a n tic ip ated ch a n g es in th e p rice level for th e p eriod o f the loan . T h e p rob lem co n sists o f isolatin g this p rice effect on m arket in terest ra tes from facto rs in flu en cin g th e real rate. In the p ro cess o f co n stru ctin g a m e a su re o f a n tici pated p rice ch an g e, past ch a n g es in p rices are a d ju sted by a su m m ary m easu re o f cu rre n t e co n o m ic co n d itio n s. Sin ce p rice ch a n g es ten d to lag ch a n g es in total spending, th e degree o f re so u rce u tilization as m easu red by th e u n em p lo y m en t rate is u sed as a lead in g in d ica to r o f future p rice m ov em en ts.20 F or exam ple, if u n em p lo y m en t is rising relative to the lab or force, d ecisio n -m ak in g e co n o m ic u n its w ould ten d to d isco u n t cu rren t inflation in form ing a n ticip a tion s about future p rice m ovem en ts. R eflecting this co n sid eratio n , th e p rice ch a n g e in e a c h q u a rter is divided by an ind ex o f th e u n em p lo y m en t rate ap p li cab le to that quarter. T h u s th e m ea su re o f p rice a n tici p atio n s w ould b e less for a given inflation rate a c c o m p an ied by high o r risin g u n em p lo y m en t th an w h en u n em p loy m en t is low o r falling. T h e sp ecific defin ition o f p rice a n ticip a tio n s is sh o w n in table 2. T h e w eights and th e len gth o f th e lag '9For other ways of handling expectations, see appendix C on alterna tive price equations. “ For purposes of exposition the unemployment rate was not included in the definition of anticipated price change in exhibits 1 and 2. 51 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 2 Table 3 Anticipated Price Definition Price Equation Sample Period: 1/1955— IV/1969 Constraints: 2nd Degree Polynomial (d_, * 0;d6 = 0) 5 AP, = 2.70 + 2 d,D,_, + .86APA, (7.07) i = 0 (8.55) (Pi from Long-Term Interest Rate Equation) APA, = Y,-2{ [ ( 17 ..P ) .01 + 1]'/4 - 1} 2 i=1 U' i/4 P , = p2 = p3 = P4 = P5 = Pe = Pi .02 .03 .04 .06 .06 .07 p7 = .08 p8 = 08 p9 = .08 PlO ~ -OS = 08 P12 = -07 P it = Pl4 = P,5 = P .6 = P,7 = 2p, = P l3 .06 .06 .04 .03 .02 .96 Symbols are defined as: APA = anticipated price change (scaled in dollar units) in t quarter t Pt_, = annual rate of change in GNP deflator (1958 = 100) in quarter t - i Ut_,/4 = index of unemployment as a percent of labor force (base = 4.0) in quarter t - i Y,_2 = total spending (GNP in current prices) in quarter t - 2 perio d w ere o btain ed from th e estim ated long-term in terest rate eq u atio n .-1 E stim a ted p r ic e eq u a tio n — T h e estim ated p rice eq u ation is show n in table 3, w here AP, is defin ed as th e d o llar ch an g e in total sp en d in g d ue to p rice ch a n g es in q u arter t. T h e in flu en ce on p rice s o f th e d em an d p ressu re variable, D , i s significant and p o si tive for five qu arters but very sm all th e re a fte r.- T h e p a ttern o f in flu en ce is o n e o f stead y d ecay, w ith 70 p e rce n t o f th e total effect o f d em an d p ressu re taking p la ce in th e first th ree qu arters an d 95 p e rce n t in th e first five qu arters. A nticipated p rice change, rep rese n te d by APA is a „ sign ificant d eterm in an t o f cu rre n t p rice ch an g e. T h ou g h significant, th e m easu re o f th e im p act o f this variable sh o u ld n o t be tak en too literally, b e c a u se its co n stru ctio n in d icates th at it ca n n o t really b e view ed 21The price expectations variable as shown in table 2 is scaled in dollar units. This transformation is made because prices are esti mated as the dollar change in total spending due to price changes. “ When the price equation is estimated with the components of D,_i separated, the coefficients for the AY, portion are not statistically significant at the five percent level, implying that the gap portion (XF, - X,_,), explains most of the changes in AP,. However, there may be collinearity problems which influence the estimated coeffi cients. Furthermore, the D,_i form is used because, theoretically, it is a measure of excess demand (see appendix A). 52 d0 = d, = d2 = d3 = .02 (2.63) .02 (6.33) .02 (6.63) .01 (2.93) R2 = .87 SE = 1.07 DW = 1.41 d„ = .01 (1.86) d5 = * (1.38) 2d, = .09 (9.18) Symbols are defined as: AP, = dollar change in total spending (GNP in current prices) due to price change in quarter t D, = AY, - (XF, - X,.,) AY, = dollar change in total spending (GNP in current prices) in quarter t XF, = potential output in quarter t X,_, = output (GNP in 1958 prices) in quarter t - 1 APA, = anticipated price change (scaled in dollar units) in quarter t NOTE: "t” statistics appear with each regression coefficient, enclosed by parentheses. R2 is the percent of variation in the dependent variable which is explained by variations in the independent variables. SE is the standard error of the estimate. DW is the Durbin-Watson statistic. *Less than .01 in d ep en d en tly o f th e d em a n d p re ssu re variable.-1T h e in flu en ce o f th e se tw o variables sh o u ld p e rh a p s be view ed in co m b in atio n , ra th er th a n as in d e p en d e n t an d sep a ra te in flu en ces.2 4 D eterm ina tion o f o u tp u t — Given AY, as d eter m in ed by th e to tal sp en d in g eq u atio n , an d AP, from th e p rice eq u atio n , th e d ollar ch a n g e in to tal sp en d in g d ue to o u tp u t ch an g es, d efin ed as AX„ ca n b e derived ziFrom the standpoint of the model as a unit, price anticipations are important only in determining the division of total spending between prices and output, not the level (or change) of spending itself. To allow for the possible direct influence of price expectations on total spending, the spending equation was estimated with the price anticipations variable. The coefficient of the price anticipations variable was not significant for this specification. 24There is, however, some evidence that the price anticipations vari able may be interpreted as an independent and separate influence. When the price equation is estimated without APA„ the sum of the coefficients on D,_, is only slightly more than shown in table 3 and the standard error is increased considerably. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 from th e follow ing identity: AY, = AP, + AX, + (P, - P,_,) ( X, - X,_,). T h e cro ss-p ro d u ct term is a ssu m ed equal to zero.-' Thus, AX, = AY, - AP,. T h e u nem p loy m en t rate plays a role in th e m odel, rep resen tin g th e m ean s by w h ich p ast p rices are a d ju sted to take in to a cc o u n t varying e c o n o m ic c o n d i tion s in th e form ation o f an ticip ated p rice ch an g es. To estim ate th e u nem p loy m en t rate, th e u nem p lo y m en t rate is regressed on cu rre n t an d lagged values o f th e GNP gap, exp ressed as a p e rce n t o f p o ten tial GNP. T h is eq u atio n is estim ated by u n co n stra in ed ord inary least squ ares and is show n in table 4. Interest Rates In terest rates do n o t fu n ctio n exp licitly in th e m odel as a p art o f a tran sm issio n m e ch a n ism ru n n in g from ch a n g es in th e m o n ey sto ck to o u tp u t an d p rices. T h e estim ated long-term in tere st rate eq u atio n plays a vital role in th e m odel, how ever, providing th e in form ation to ca lcu late th e m easu re o f p rice an ticip atio n s. M arket in terest rates are sp ecified to d ep en d on cu rren t and p ast rates o f ch a n ge o f o u tp u t (X), th e cu rren t rate o f ch an g e in th e m o n ey sto ck (M), and cu rren t and past rates o f ch an g e in p rices (P) a d ju sted by an in d ex o f th e u nem p lo y m en t rate. T h is sp e cifica tion draw s on Sarg ent's work, w h ich w as exp lored fu rth er by Yohe and Karnosky.2 " L o n g -term in terest ra te — T h e lon g -term m arket rate (R1,) is m easu red by th e rate on sea so n e d c o rp o rate Aaa bo n d s. C han ges in o u tp u t an d p rice s (ad ju sted for u nem ploym ent) lagged for 17 q u arters p ro vide th e m o st satisfactory resu lts. T h e estim ated eq u ation for th e lo n g -term rate is show n in table 5. 25The value of this cross-product term was calculated from 1953 to the present and provides ample justification for the assumption that it be equated to zero for purposes of the model here. Also note that AP, is defined in dollar units, that is, as (P, - P,_,)X,_,, not (P, - P,_,). AX, is defined analogously. ^Thomas Sargent, “Commodity Price Expectations and the Interest Rate,” Quarterly Journal of Economics (February 1969), pp. 127-40, and Yohe and Karnosky, pp. 31-34, 38. The estimated interest rate equations also contain a dummy variable (0 for 1955-60 and 1 for 1961-69). The significance of this dummy variable indicates a shift of structure within the sample period. Questions can be raised about this procedure, but it is felt that a price expectations variable should not be constructed on the basis of a sample period containing only an expansion like 1961-69. Including the dummy variable leaves unexplained that factor (or factors) which changed the relationship, but it does provide a way of estimating a set of coefficients on prices that is based on a sample period reflecting varying economic cir cumstances. Table 4 Unemployment Rate Equation Sample Period: I/1955-IV/1969_________ U, = 3.90 + .04G, + .28G,_, (72.50) (1.10) (6.80) R2 = .92 SE = .30 DW = .60 Symbols are defined as: U, = unemployment as a percent of labor force in quarter t XF, = potential output in quarter t X, = output (GNP in 1958 prices) in quarter t NOTE: “t” statistics appear with each regression coefficient, enclosed by parentheses. R2 is the percent of variation in the dependent variable which is explained by variations in the independent variables. SE is the standard error of the estimate. DW is the Durbin-Watson statistic. T h e resu lts reflect, in a gen eral way, th e view stresse d by m o n eta rists th at a ch a n g e in th e rate of m o n eta iy ex p a n sio n in flu en ces m arket in tere st rates in th ree stages.27 First, th e liquid ity effect o f a n in crea se in th e rate o f ch a n g e o f th e m o n ey stock on m arket in terest rates is negative. Seco n d , an in cre a s e in th e rate o f m o n eta iy ex p a n sio n in flu en ces th e rate o f ch a n g e in output, w h ich in tu rn h a s a positive in flu en ce on m arket in terest rates. Finally, an in cre a s e in m o n ey grow th in flu en ces th e rate o f ch a n g e in p rices, w h ich h as a positive effect on m arket in tere st rates. S h o rt-term in terest ra te — The sh o rt-term in terest rate (Rs,) w h ich is estim a ted is th e fou r- to six-m o n th co m m ercia l p ap er rate. T h e eq u atio n is sh o w n in table 6. Price ch a n g es are fou nd to e n te r significantly for a sh o rte r lag perio d th a n in th e lo n g -term rate eq u atio n . Also, th e sh o rt-term rate, as m easu red by th e fou r- to six-m o n th co m m e rcia l p a p e r rate, is m u ch m o re sensitive to ch a n g es in o u tp u t an d th e m o n ey stock th a n is th e long-term rate as m easu red by th e rate on sea so n ed co rp o ra te Aaa b o n d s. Time Response to Monetary Actions: A Summary T h e p attern o f th e co efficien ts in th e eq u atio n s provides in form ation abou t th e tim e re sp o n se o f total spen din g, output, an d p rices to m o n eta iy an d fiscal actio n s. T h e eq u atio n s in d ica te th at m o n e ta iy a ctio n s 27See Friedman, “The Role of Monetary Policy,” p. 6. 53 FEDERAL RESERVE BANK OF ST. LOUIS g en erally affect total sp en d in g w ith a tw o- to th reeq u a rte r lag. A ch an g e in th e rate o f grow th o f total sp en d in g is acco m p a n ie d by a sim u ltan eo u s ch an g e in th e rate o f grow th o f output, and it is not until th ree qu arters la te r th at th e re sp o n se o f p rice s to a ch an g e in d em an d p re ssu re bu ild s to 70 p e rce n t o f th e total. T h e re sp o n se o f p rice s to a ch an g e in to tal sp en d in g is y et slow er w h en th e re are a n ticip a tio n s o f a high rate o f inflation. T h e sp en d in g eq u atio n (table 1) in d icates that ab o u t h a lf o f th e to tal re sp o n se to a ch a n g e in m o n e tary grow th o cc u rs in th e first tw o qu arters, an d about 80 p e rce n t in th e first th ree qu arters. T h e p atte rn o f co efficien ts in th e p rice eq u atio n (table 3) in d icates th at th e effect o f a ch an g e in total sp en d in g is reflected first in o u tp u t an d la ter in p rices. O perating throu gh th e d em an d p ressu re variable, a b o u t a fou rth o f th e re sp o n se o f p rices to a ch an g e in to tal sp en d in g is in th e first qu arter, w h ich is about tw o qu arters after th e ch an g e in m o n etary actio n s. Over 70 p e rce n t o f th e p rice re sp o n se is in th e first th ree qu arters, and 95 p e rce n t in the first five qu arters. T h e re sp o n se o f th e p rice level to ch an g es in total sp en d in g is also in flu en ced by a n ticip a te d p rices. T h e g reater th e a n ticip ate d rise in p rices, th e longer d e layed is th e re sp o n se o f th e p rice level to a d eclin e in th e rate o f ch an g e in total spending. TESTS OF THE M O D E L’S PERFORMANCE T h e eq u atio n s o f th e m o d el are to b e view ed as a unit, an d th e sp ecificatio n o f th e m odel is su ch that given th e ch an g e in m o n ey (AM), an d th e ch an g e in h ig h -em p loy m en t ex p en d itu res (AE), th e m o d el can be solved in th e follow ing seq u e n ce: for th e ch an g e in total sp en d in g (AY), th e ch an g e in th e p rice level (AP), th e ch an g e in real ou tp u t (AX), th e u nem p loy m en t rate (U) an d th e long- an d sh o rt-term in tere st rates (RLand Rs). T h e exp lanatory p o w er o f ea ch o f th e eq u atio n s sh o w n in tables 1 - 6 m ay be a cc e p ta b le by co n v en tion al stand ard s, but th is provides no g u aran tee th at th e m o d el w ill perform satisfactorily as a u nit. T h ere are in terd e p en d e n cie s in th e m od el th at have to be tak en in to a cc o u n t w h en evaluating th e co m p lete m od el. O f in terest in evaluating th e m od el as a u n it is the im plied pattern for th e en d o g en o u s variables w h en only an initial set o f lagged en d o g en o u s vari ables and th e tim e path s o f th e exo g en o u s policy variables (m oney sto ck an d h ig h -em p loy m ent federal 54 OCTOBER 1986 Table 5 Long-Term Interest Rate Equation Sample Period: 1/1955-IV/1969 Constraints: 2nd Degree Polynomial (x_, ^ 0, p_, =£ 0; x17 = p17 = 0) RL, = 1.28 - .06M, + 1.42Z, (5.20) (-3.53) (11.01) 16 16 + 2 x, X,_, + X i= 0 i= 0 p( (U,_/4) rP^ r,-i R2 = .92 SE = .28 DW = .69 Xo = .02 ( 3.85) x, = .02 ( 4.35) x2 = .02 ( 4.44) x3 = .02 ( 4.08) x„ = .02 ( 3.54) x5 = .02 ( 3.03) x6 x7 xs x9 x,o x„ = = = = = = .01 ( 2.61) .01 ( 2.27) .01 ( 2.01) .01 ( 1.80) .01 ( 1.64) .01 ( 1.50) x,2 = x,3 = x,4 = Xi5 = x,6 = 2x, = p0 = .02 ( 1.23) p, = .03 ( 3.05) p2 = .04 ( 5.96) p3 = .06 (10.82) p4 = .06 (17.34) p5 = .07 (19.66) p6 p7 p8 p9 p,„ p„ = = = = = = .08 (17.13) .08 (14.49) .08 (12.64) .08 (11.37) .08 (10.47) .07 ( 9.81) p12 = .06 ( 9.29) Pi3 = .06 ( 8.89) p,4 = .04 ( 8.57) p,5 = .03 ( 8.30) Pi6 = 02 ( 8.07) 2p( = .96 (19.04) .01 ( 1.38) .01 ( 1.28) * ( 1.20) * ( 1.13) * ( 1.07) .20 ( 2.88) Symbols are defined as: RL, = Moody’s seasoned corporate Aaa bond rate in quarter t M, = annual rate of change in money slock in quarter t Z, = dummy variable in quarter t (0 for 1/1955-IV/1960 and 1 for 1/1961 — IV/1969) X,_i = annual rate of change in output (GNP in 1958 prices) in quarter t - i Pt_( = annual rate of change in GNP deflator (1958 = 100) in quarter t - i U,_j/4 = index of unemployment as a percent of labor force (base = 4.0) in quarter t - i NOTE: “t” statistics appear with each regression coefficient, enclosed by parentheses. R2is the percent of variation in the dependent variable which is explained by variations in the independent variables. SE is the standard error of the estimate. DW is the Durbin-Watson statistic. "Less than .01 expen d itu res) are a ssu m ed know n. T o co n d u ct su c h a test, several d yn am ic sim u lation ex p e rim e n ts w ere p erform ed . T h e se sim u lation s take th e form o f e * post d yn am ic sim u lation s an d an epc ante d yn am ic sim u la tion .3 * 28For a discussion of the different ways of assessing the tracking ability of econometric models, see Carl F. Christ, “Econometric Models of the Financial Sector,” forthcoming in the Journal of Money, Credit, and Banking. For a discussion of simulation proce dures and results with an income-expenditure model, see Evans and Klein, pp. 50-69. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 6 Short-Term Interest Rate Equation Sample Period: 1/1955— IV/1969_________ Constraints: 2nd Degree Polynomial (x_, 4 0, p., * 0; x„ = p„ = 0) 10 10 p Rst = —1.13 - 17M, + .92Z, + 2 x, X,_, + 2 p,(— — ) (-3.14) (-6.36) (5.03) i = 0 i = 0 U'-’/4 R2 = .90 SE = .47 DW = .69 Xo = x, = x2 = x3 = x„ = x5 = x6 = x7 = x8 = x9 = x,0 = Zx, = .08 .08 .08 .08 .08 .07 .07 .06 .05 .03 .02 .71 ( 8.10) ( 8.99) ( 8.97) ( 8.45) ( 7.83) ( 7.25) ( 6.76) ( 6.36) ( 6.02) ( 5.75) ( 5.52) ( 8.24) Po =’ Pi = P2 = P3 = P4 = Ps = Pe = Pz = Ps = P9 = Pio ” SPi = -.0 4 .04 .09 .14 .17 .18 .19 .18 .15 .11 .06 1.27 (-.9 1) ( 1.53) ( 8.26) (18.52) (13.17) (10.29) ( 8.92) ( 8.14) ( 7.64) ( 7.30) ( 7.04) (16.89) Symbols are defined as: Rs, = four- to six-month commercial paper rate in quarter t M, = annual rate of change in money stock in quarter t Z, = dummy variable in quarter t(0 for 1/1955-IV/1960 and 1 for 1/1961-I V/1969) X,_i = annual rate of change in output (GNP in 1958 prices) in quarter t - i P,_i = annual rate of change in GNP deflator (1958 = 100) in quarter t - i U,_i/4 = index of unemployment as a percent of labor force (base = 4.0) in quarter t - i NOTE: “t” statistics appear with each regression coefficient, enclosed by parentheses. R2 is the percent of variation in the dependent variable which is explained by variations in the independent variables. SE is the standard error of the estimate. DW is the Durbin-Watson statistic. Ex Post Dynamic Simulations An e\ post d ynam ic sim u lation is co n fin ed to the sam ple period from w h ich th e estim ated relatio n sh ip s are derived. A ctual values for all cu rren t an d lagged exog en ou s variables are u sed , bu t only initial actu al values for th e lagged en d o g en o u s variables are used. T h e m od el g en erates so lu tio n values for th e en d o g e n o u s variables in th e first sim u lation period, w h ich are th en u sed to g en erate so lu tio n values for th e se co n d period, and so on for e a c h su cce ed in g p erio d .-’ A co m p ariso n o f th e se ca lcu la ted tim e p ath s for the en d o g en o u s variables w ith th e ir a ctu al tim e paths en ab les o n e to form ulate so m e ju d g m en t as to h ow w ell th e m od el perform s as an in terd e p en d e n t unit in tracking th e m ovem ents o f ce rta in strategic e co n o m ic variables. /> post d ynam ic sim u lation s w ere co n d u cte d for ■■ several su b p erio d s w ithin th e sam p le period (1 9 5 5 69). T h e resu lts for th e en tire sam p le perio d are su m m arized in ch art 1 on th e next page. W hen sim u la tion s are c o n d u cte d for su b p erio d s w ithin th e 1 9 5 5 -6 9 period, th e p attern o f m ovem en t as sh o w n for th e w h o le p eriod sim u lation te n d s to hold, b u t th e levels are c lo se r to th e actual values at th e b egin n in g o f ea ch su bperiod. C hart 1 in d icates th at th e m od el ten d s to track the m ovem ent o f th e en d o g en o u s variables qu ite w ell d u r ing th e 1 9 5 5 -6 9 period. Sin ce criteria for jud ging th e p erfo rm an ce o f the m odel in su ch a sim u lation have n o t b ee n developed, an y co n c lu sio n s are n ecessa rily su bjective.3 T h e te n d en cy for th e m od el to avoid d i 0 verging sharply from th e a ctu al p ath for ex ten d ed period s is an esp ecially im p o rtan t featu re. Su ch a featu re provides so m e b asis for tru stin g th e tracking ability o f th e m o d el over several qu arters, even if o n a qu arter-b y-qu arter b asis it m ay ap p e a r to b e off th e mark. To gain ad d ition al in form ation ab o u t th e pred ictive p erfo rm an ce o f th e m odel, a co m p a riso n is m ad e w ith an e?t post sim u lation from a n o th e r m od el. R esu lts of an e * post sim u lation for 1963 an d 1964 have b een p u b lish ed for th e W harton m od el. T h e resu lts for th e m od el are co m p a red w ith th o se o f th e W h arton m od el in table 7. T h e period 1 9 6 3 -6 4 in clu d es th e 1964 tax cut, w h ich , a cco rd in g to the W h arton m odel, is co n sid ered an im p o rtan t fa cto r in flu en cin g e c o n o m ic d evelop m en ts in 1964. However, th e St. Louis m od el, w h ich d o es n ot em p h asize su ch fiscal actio n s, did ab o u t as well, on average, for th e y ea rs 1963 an d 1964 (see table 7). T h e m ain d ifference to b e rem em b ered in evaluat ing th e se sim u lation s is th at th e St. Louis m o d el c o n tain s th ree prim ary exo g en o u s variables, w hile th e W h arton m od el co n ta in s 43. ^See de Leeuw and Gramlich, “The Channels of Monetary Policy . . . p. 485. “ See Robert H. Rasche and Harold T. Shapiro, “The F.R.B.-M.I.T. Econometric Model: Its Special Features,” American Economic Re view, Papers and Proceedings (May 1968), p. 142. 55 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS C h a rt I Results of Ex Post Dynamic Simulation A n n u a l R a te s o f C h a n g e 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 N o t e : P r e d ic t e d v a lu e s b a s e d o n e s t im a t e o f m o d e l f o r s a m p le p e r i o d e n d in g I V / 1 9 6 9 . 56 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 7 Alternative Ex Post Simulations: Actual Minus Predicted1 Comparison of Wharton and St. Louis Models for 1963-64______ Nominal GNP2 Real GNP Price Level3 Unemployment Rate4 Wharton St. Louis Wharton St. Louis Wharton St. Louis Wharton St. Louis 0.4 -0 .4 -3 .9 -0.1 0.2 1963: I -4 .6 0.9 0.3 0.4 -0 .7 -0.1 II -0 .2 0.3 0.2 0.7 0.1 2.5 0.6 -0 .3 0.1 III 1.3 1.5 0.9 -0.1 2.2 -0 .4 -0.2 2.1 0.5 1.2 IV 0.9 0.2 -1 .4 1.7 2.7 1.4 1964: I 0.9 -0 .3 0.6 0.2 3.2 2.3 -0 .3 1.4 II 1.1 0.1 0.6 0.2 4.0 -2 .7 1.7 -0 .5 0.8 III 1.5 1.6 0.2 2.2 -6 .8 -0.4 -1 .7 0.9 1.2 0.2 IV 0 1.55 -1.88 -0.28 0.49 0.11 0.76 1.16 0.16 Average Error Root Mean 2.92 3.09 0.33 0.60 1.49 1.28 0.21 Squared Error 2.00 'Sample period: Exogenous variables: Wharton: 1948-1964 Wharton: 43 St. Louis: 1955-1969 St. Louis: 3 2Billions of dollars. 3Computed from the level of implicit price deflator. “Percent. Sources: M. K. Evans and L. R. Klein, The Wharton Econometric Forecasting Model, 2nd, Enlarged Edition (Philadelphia: University of Pennsylvania, 1968); and Federal Reserve Bank of St. Louis. T h e co m p ariso n is n ot m ean t to im ply th at th e St. Louis m od el is su p erior. R ather, th e su ggestion is offered th at a sm all m o d el c o n stru c te d w ith in a m o n e tarist fram ew ork m ay y ield as m u ch in form ation abou t th e key aggregates as a large stru ctu ral m od el. In sum m ary, sm all m o n etarist m o d els m ay b e u sefu l as a guide in th e form u lation o f stabilizatio n policy. Ep Ante Dynamic Simulations An e?c ante d yn am ic sim u lation is like an e\ post d ynam ic sim ulation , excep t th at it ex ten d s b ey o n d th e sam p le period . To co n d u ct su ch a sim u lation for this m odel, it w as n ecessa ry to re-estim ate th e m o d el for a su b p erio d w ithin th e full sam p le p eriod . All eq u atio n s o f th e m od el w ere re-estim ated w ith d ata th rou gh 1967. T h e period o f th e e^c ante d yn am ic sim u lation is 1968 and 1969. T h e resu lts are su m m arized in ch a rt 2 and in tables 8 and 9. T h e s u c c e ss o f th e e\ ante d ynam ic sim u lation can be a ssessed by co m p arin g it w ith th e track ing reco rd o f th e f',v post sim u lation for th e sam e period. A co m p ariso n o f th e erro rs a sso cia te d w ith th e e,\ ante sim u latio n w ith th o se o f th e c.x post sim u lation (w here th e erro rs in b o th ca se s are co m p u te d w ith re feren ce to a ctu a l values) su ggests that any stru ctu ral shifts that o cc u rre d in th e 1 9 6 8 -6 9 p erio d w ere n o t o f su ch a m agn itu d e th at th e c,\ ante trackin g ability o f th e m odel w as significantly d ifferent from that o f c.x post sim u lation . Any c o n c lu sio n s ab o u t th e tracking ability o f th e m od el are n ecessa rily tentative, b e c a u se th ey are b a sed o n only o n e e\ ante d yn am ic sim u lation ex p eri m en t. N evertheless, th e se resu lts provide a tentative basis for co n fid e n ce in the tracking ability o f the m o d el in estim atin g th e ec o n o m ic resp o n se to m o n e ta iy an d fiscal actio n s. U nfortunately, it is difficult to co n d u ct ad d ition al tests o f th is type for o th e r su b p e riods in th e sam ple, b ec a u se d egrees o f freed o m are severely re d u ce d w h en th e sam p le perio d is sh o rt en ed furth er. 57 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 USING THE M ODEL FOR CURRENT ANALYSIS T h e m o d el is u sed in th is sectio n to sim u late th e effects o f p o ssib le fu tu re rates o f m o n etary ex p an sio n o n spendin g, output, p rices, u n em p loy m en t an d in terest rates, given th e e c o n o m ic circ u m sta n ce s o f late 1969 an d early 1970. Sim u lation o f th e se alternative co u rses o f m o n eta iy actio n su ggests how th e m od el m ay provide in fo im atio n w h ich will be helpful to policym akers. C h a r t II Results of Ex Ante Dynamic Simulation A n n u a l R ates o f C h a n g e Sim u lations o f th e m od el are co n d u cte d only for alternative rates o f m o n e ta iy ex p an sio n . T h is is d o n e b e c a u se o f th e n atu re o f th e resu lts for th e sp en d in g eq u ation . T h e n et effect o f a ch an g e in federal ex p e n d itu res o n total sp en d in g (GNP) over a five-quarter p e riod is very sm all if u n a cco m p a n ied by m o n e ta iy e x pan sion. Short-Run Projections T h e resu lts o f sim ulating th e m od el for alternative grow th rates o f m oney, an d for th e grow th o f federal ex p en d itu res as p ro je c te d in th e fiscal 1971 budget, are sh ow n in table 10. T h e se sim u lation resu lts reflect th e a cceleratin g inflation o f th e p ast several y ea rs an d th e fiscal and m o n eta iy restrain t in fo rce throu gh ou t 1969 an d early 1970. T h e se p ro je c tio n s assu m e th at em p irical relatio n sh ip s b ased on p ast ex p e rie n ce will co n tin u e to h o ld in th e n e a r future. R ates o f ch an g e in th e m o n ey stock w ere co m p u ted from th e first q u arter o f 1970. T h re e alternative rates are sh o w n in table 10. T h e “n o -ch an g e c a s e ” co rre sp o n d s to th e co u rse o f m o n eta iy a ctio n s in th e s e c o n d h a lf o f 1969. T h e “th ree p e rce n t c a s e ” co rre s p o n d s to th e tren d rate o f in crea se in m o n ey from 1961 to 1965. Finally, th e “six p e rce n t c a s e ” re p rese n ts m o n eta iy actio n s sim ilar to th o se o f 1967 an d 1968. N o -ch a n g e c a s e — A co u rse o f no ch an g e in the m on ey sto ck from th e first q u arter o f 1970 w ould lead to fu rth er re d u ctio n in th e rate o f in cre a se o f total sp en d in g in 1970 and 1971 (table 10). A slow ing o f total sp en d in g along w ith upw ard p re ssu res on p rice s from th e past inflation (anticip ated p rice effect) w ould lead to co n tin u ed d eclin es in o u tp u t th rou gh 1971. Su ch a restrictive co u rse o f m o n eta iy a ctio n s w ou ld slow th e rate o f p rice in crease to a 4 p e rce n t rate by late 1970 a n d to a 2 p e rce n t rate by late 1971. T h e d eclin e in o u tp u t w ould be acco m p a n ie d by a rise in th e u n e m p loym ent rate to over 7 p e rce n t by late 1971. T h e effects o f su ch restrictive m o n eta iy a ctio n s on 58 N o te : P r e d ic te d v a lu e s b a s e d o n e s tim a te o f m o d e l fo r s a m p le p e r io d e n d in g IV /1 9 6 7 . FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 8 Model Simulations 1968 1 GNP Level (billions of dollars) Actual Ex Ante1 Ex Post2 Annual Rate of Change in Y Actual Ex Ante Ex Post Annual Rate of Change in X Actual Ex Ante Ex Post Annual Rate of Change in P Actual Ex Ante Ex Post Unemployment Rate (percent) Actual Ex Ante Ex Post Corporate Aaa Rate (percent) Actual Ex Ante Ex Post Commercial Paper Rate (percent) Actual Ex Ante Ex Post II 1969 III IV 835.3 858.7 876.4 892.5 834.1 856.7 878.9 899.9 834.6 856.7 877.7 897.8 1 II III IV 908.7 924.8 942.8 952.2 917.6 932.3 945.9 957.2 914.9 929.4 943.4 955.1 9.7 9.1 9.4 11.7 11.3 11.0 8.5 10.8 10.2 7.6 9.9 9.5 7.5 8.1 7.8 7.3 6.6 6.5 8.0 6.0 6.1 5.9 5.7 5.4 7.4 7.6 6.7 4.0 7.0 5.7 3.2 5.9 4.8 2.6 4.0 3.0 2.0 2.4 1.6 2.2 -0 .4 1.7 0.7 1.1 0.0 3.7 3.3 3.8 4.0 3.4 4.0 4.0 3.6 4.3 4.3 3.8 4.5 4.9 4.0 4.7 5.2 4.1 4.9 5.4 4.2 5.0 4.7 4.2 5.0 3.7 3.9 3.9 3.6 3.8 3.7 3.6 3.5 3.5 3.4 3.3 3.4 3.4 3.2 3.4 3.5 3.2 3.5 3.6 3.3 3.7 3.6 3.5 3.9 6.1 5.8 5.9 6.3 5.9 6.0 6.1 6.0 6.1 6.2 6.4 6.5 6.7 6.5 6.7 6.9 6.7 6.9 7.1 7.0 7.2 7.5 7.1 7.4 5.6 5.1 5.8 6.1 5.1 5.8 6.0 5.1 5.8 6.0 5.6 6.5 6.7 5.7 6.6 7.5 6.0 7.0 8.5 6.6 7.8 8.6 6.7 8.0 4.0 4.9 5.1 Key to Abbreviations: Y = Nominal GNP X = Real GNP P = GNP price deflator 'Simulation based on equations estimated through IV/1967. Simulation based on equations estimated through IV/1969. in terest rates w ould b e to k eep th e lon g -term in terest rate at re cen t levels th rou gh 1970, m ainly b e c a u se o f the effects o f p ast inflation. By early 1971, th e slow er advance o f p rices in 1970 and th e slow ing o f o utput grow th w ould lead to d eclin es in th e lon g -term rate. T h e sh o rt-term in tere st rate, on th e o th e r han d , w ould hold at re cen t levels only tem porarily, partly b eca u se of co n tin u ed restrictive m o n e ta iy actio n s. Sh o rt-term rates w ould drop sh arp ly by th e seco n d h a lf o f 1970, reflecting prim arily th e slow ing o f o u tp u t grow th. Sin ce th e p rice lags are sh o rte r for th e sh o rt-term rate, th e effects o f p ast inflation are n o t so pervasive as for th e long-term rate. T h r e e p e r c e n t c a s e — G rowth o f th e m o n ey sto ck at a th ree p e rce n t an n u al rate is p resen ted to illu strate th e effects o f a m o d erate ex p a n sio n o f m on ey. T h is rate co rresp o n d s to th e tren d rate o f in c re a s e in m o n ey from 1961 to 1965. In th e cu rre n t e co n o m ic situation, a th ree p e rce n t rate o f ex p a n sio n in m on ey w ould rep rese n t a m o d erate easin g o f m o n etary p o l icy from its restrictive in flu en ce o f late 1969 an d early 1970. T h e effect o f su ch ex p an sio n w ould b e to m aintain 59 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS grow th in total sp en d in g at a rate abou t th e sam e as in th e fourth q u arter o f 1969. Given th e in flu en ce o f past inflation o n p rices, o u tp u t w ou ld d eclin e slightly th rou g h 1970, bu t w ould resu m e its in cre a se by 1971. T h e effect on p rices in 1970 w ould be little d ifferent from th e n o -ch an g e case, but by late 1971 th e differ e n c e w ould b e m arked. In th e th ree p e rce n t case, p rices w ould still b e rising at a th ree p e rce n t rate by late 1971 co m p ared w ith a tw o p e rce n t rate fo r th e n o ch an g e case. M od erate ex p an sio n o f th e m o n ey sto ck lead s to a rise in th e u n em p lo y m en t rate th rou g h 1970 an d 1971. In general, for th is m od el, th e u n em p loy m en t rate rises as long as o u tp u t grow s at le ss th a n a fou r p e rce n t rate. T h e lon g -term in terest rate w ould rem ain at re ce n t levels th rou gh o u t 1970, and n ot until early 1971 w ould th e effect o f slow er p rice in c re a se s an d o u tp u t grow th b e en o u g h to offset th e effects o f p ast in flation. T h e sh o rt-term in tere st rate w ould fall m o re qu ickly th a n th e long-term rate bu t w ould not fall as m u ch by late 1971 as in th e n o -ch an g e case. Su ch a p atte rn for th e sh o rt-term rate illu strates th e sh o rt- an d lon g er-ru n in flu en ce o f q u ick en ed m o n etary ex p an sio n . Six p e r c e n t c a s e — A six p e rce n t an n u al rate of in cre a se in m o n ey is sh o w n to illu strate th e effects o f a su d d en shift to a very rapid rate o f m o n etary ex p a n sion in th e seco n d q u arter o f 1970. Su ch in cre a se in m o n ey w ould be about th e sam e as d uring 1967 and 1968. A m a jo r effect o f shifting to rapid m o n etary ex p a n sion w ou ld b e to a d v a n c e th e r a t e o f to tal sp en d in g grow th. By late 1971, total sp en d in g w ould be in c re a s ing at an 8 p e rce n t rate w ith su ch m o n etary a ctio n s. T h e rate o f p rice in crea se w ould fall som ew h at, h o w ever, b ec a u se o f p ast restrictive m o n etary actio n s. B ut th e gain in p rice p erfo rm an ce w ould b e sm all, b ec a u se by late 1971 p rices w ould still be in creasin g at a 4 p e rce n t rate. T h e effects o f p ast m o n etary an d fiscal a ctio n s, along w ith p ast inflation, w ould lead to a d eclin e in o u tp u t th rou gh m id-1970. F rom th e n th rou gh 1971, ou tp u t grow th w ould in crease. D esp ite a shift to a very rapid rate o f m o n etary grow th, u n em p lo y m en t w ou ld rise u n til m id -1971. T h is in crea se in u nem p loy m en t w ould follow b ec a u se o f th e co n tin u ed in flu en ce o f past nro n etaiy an d fiscal actio n s. By late 1971, th e recovery in o u tp u t grow th w ould b e p u sh in g th e u n em p lo y m en t rate b ack dow n. A shift to rapid m o n etary ex p an sio n h as a p ro n o u n ced effect on m arket in terest rates. T h e lon g term rate w ould stay at re ce n t levels th rou g h 1971, b e c a u se th e in flu en ce o f p ast p rice s (an ticip ated p rice 60 Table 9 Ex Ante vs. Ex Post Simulation: Actual Minus Predicted — 1/1968-IV/1969 (root mean squared errors) Ex Post Ex Ante 1.45 GNP’ 1.26 0.99 1.58 GNP in 1958 Prices' GNP Deflator' 0.25 0.76 0.14 Unemployment Rate 0.20 0.17 0.26 Corporate Aaa Rate 0.44 Commercial Paper Rate 1.26 'Computed from actual minus predicted annual rates of change. effect) w ou ld n o t b e offset by a su sta in e d re d u ctio n in o u tp u t grow th. T h e sh o rt-te rm rate w ou ld fall, in re sp o n se to th e tem p o rary re d u ctio n in o u tp u t grow th, b u t th e d eclin e by late 1971 w o u ld b e le ss th a n for eith e r th e n o -ch a n g e o r th ree p e rce n t ca se. Implications o f the Model f o r the Long Run S h o rt-ru n p ro sp e cts for ec o n o m ic variables te n d to d o m in a te p o licy m a k e rs’ d e c is io n s . H ow ever, th e lo n g er-ru n co n se q u e n ce s o f alternative m o n eta ry p o l icie s sh o u ld also b e given co n sid era tio n . T h is m o d el is in co m p le te for lo n g -ru n analy sis; n ev erth eless, it y ield s re su lts th a t are o f in tere st a n d m ay n o t b e too far rem oved from re su lts th a t m igh t evolve from a m o re co m p lete sp ecifica tio n .3' W h en sim u latio n s are co n d u cte d for lon g period s in to th e fu tu re (30 years), th e m o d el d em o n stra te s p ro p erties c o n siste n t w ith th o se ex p o u n d ed by th e cla ssica l eco n o m ists. Over th e lon g run, m o n etary a ctio n s have n o effect on real m ag n itu d es; th e ra te of grow th o f output, th e u n em p lo y m en t rate an d th e real rate o f in tere st all te n d to m ove tow ard som e eq u ilibriu m rate, regard less o f w h ic h rate o f m o n ey 31The shortcomings of the model for the long-run analysis are quite evident. There are no assumptions specified as to labor force growth and productivity. Furthermore, there is no investment func tion and, therefore, the capital stock is not an endogenous variable. All long-run assumptions are embodied in assumptions about the growth rate of potential output. With these assumptions, policy actions cannot affect the economy’s long-run growth rate. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 10 Simulation of Alternative Rates of Monetary Expansion Actual Projected Projected Rate of IV/1969 Change in M' 1/1970 11/1970 111/1970 IV/1970 1/1971 0 Percent Annual Rate of change in Y X P Unemployment Rate Corporate Aaa Rate Commercial Paper Rate 3 Percent Annual Rate of change in Y X P Unemployment Rate Corporate Aaa Rate Commercial Paper Rate 6 Percent Annual Rate of change in Y X P Unemployment Rate Corporate Aaa Rate Commercial Paper Rate 11/1971 111/1971 IV/197 4.0 (5.1 )2 -0.4 (0.1) 4.7 (5.0) 3.6 (4.0) 7.5 (7.4) 8.6 (8.0) 3.5 -1 .3 4.9 4.3 7.4 7.6 3.1 1.6 4.7 4.8 7.5 7.6 1.0 -3 .3 4.4 5.2 7.4 7.1 0.3 -3 .6 4.0 5.8 7.4 6.5 0.7 -2 .7 3.5 6.4 7.3 5.7 0.9 -2 .0 3.0 6.9 7.1 5.0 2.1 -0 .4 2.5 7.4 6.9 4.4 0.9 -1 .0 1.9 7.7 6.7 3.6 4.0 (5.1)2 -0.4 (0.1) 4.7 (5.0) 3.6 (4.0) 7.5 (7.4) 8.6 (8.0) 3.5 -1 .3 4.9 4.3 7.4 7.6 3.8 0.8 4.7 4.8 7.3 7.2 2.9 -1 .5 4.5 5.2 7.3 6.8 3.1 -1 .0 4.2 5.6 7.3 6.4 4.1 0.3 3.8 6.0 7.3 5.9 4.4 0.8 3.5 6.3 7.2 5.5 5.5 2.2 3.2 6.5 7.1 5.0 4.3 1.4 2.9 6.7 6.9 4.5 4.0 (5.1)2 -0.4 (0.1) 4.7 (5.0) 3.6 (4.0) 7.5 (7.4) 8.6 (8.0) 3.5 -1 .3 4.9 4.3 7.4 7.6 4.6 0.1 4.7 4.8 7.2 6.7 4.8 0.2 4.5 5.1 7.2 6.5 6.0 1.6 4.3 5.4 7.2 6.3 7.6 3.3 4.2 5.6 7.3 6.1 7.8 3.7 4.0 5.6 7.3 5.9 8.9 4.8 3.9 5.7 7.3 5.7 7.7 3.8 3.8 5.7 7.2 5.5 Key to Abbreviations: Y = Nominal GNP P = GNP Price Deflator X = Real GNP M = Money Stock 'Rates of change in money projected from 1/1970. High-employment Federal expenditures projected on basis of fiscal 1971 budget, as released in January 1970. 2Model estimates. grow th is m ain tain ed . T h e effects o f alternative rates o f m o n etary ex p an sio n are on n o m in al m agn itu d es, nam ely, to tal spending, p rices, an d m arket in terest rates. B ased on th e assu m p tio n s o f th e m od el, a six p e r c e n t rate o f grow th in m o n ey along w ith a six p e rce n t grow th rate in fed eral ex p en d itu res, for exam ple, w ould lead u ltim ately to ab o u t a six p e rce n t rate o f grow th in total spen ding, a fou r p e rce n t rate o f grow th in output, a two p e rce n t rate o f in crea se in p rice s and m arket in terest rates ab o u t tw o p e rce n tag e p o in ts in e x cess o f th e real rate. Alternatively, a tw o p e rce n t grow th rate in m o n ey w ould resu lt ap p roxim ately in a two p e rce n t grow th in total spending, a fou r p e rce n t rate o f grow th in output, a two p e rce n t rate o f d eclin e in p rice s and m arket in terest rates abou t tw o p e rc e n t age p o in ts belo w th e real rate. Over th e lon g run, th e m o d el in d ica tes th at high em p lo y m en t an d p rice sta bility are co m p atib le. SUMMARY T h e m ain p u rp o se o f th is stu d y h as b ee n to quantify th e effects o f m o n etary an d fiscal a ctio n s w ithin a sm all-m od el fram ew ork an d th ereb y offer an a ltern a tive to existing larg e-scale e c o n o m e tric m o d els. Su ch a m od el h as b een form u lated an d th e effects o f m o n e ta iy an d fiscal a ctio n s on sp en d in g, output, p rices, em p lo y m en t an d in tere st rates have b ee n estim ated . T h e m o d el d eveloped in th is article is prim arily “m o n eta rist” in ch a ra cte r. T h e estim ated eq u atio n s in d icate th at m o n eta iy a ctio n s, as m easu red by ch a n g es in the m o n ey stock, play a strategic role. F iscal actio n s, as m easu red by h ig h -em p loy m en t fed 61 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS eral exp en d itu res, have so m e sh o rt-ru n effects, b u t for p erio d s o f a y e a r o r m o re th e n et effect o n spending, ou tp u t and p rices is n ear zero. Sim u lation s o f altern a tive rates o f m onetary ex p an sio n p ro d u ce sh o rt-ru n an d lon g -ru n re sp o n ses w h ich are c o n siste n t w ith th e gen eral m o n etarist view o f th e eco n om y. O ne o f th e c h ie f advantages o f this m od el is th at it d ep en d s prim arily on in form ation ab o u t o n ly two variables — th e m oney sto ck and hig h -em p loy m ent ex p en d itu res.3 C on sid erable insigh t can be gained abou t th e p attern o f ex p e cted m ovem ents o f certa in 32This feature has led John Deaver to conjecture that the standard error of forecast in the Andersen-Jordan model may be far lower than that of the FRB-MIT model. See his “Monetary Model Build ing,” Business Economics (September 1969), p. 30. strategic e co n o m ic variables by co n sid erin g a lte rn a tive co u rses o f m o n etary a n d fiscal a ctio n s. However, sin ce th e m od el is lim ited to on ly m o n etary an d fiscal in flu en ces, to th e ex clu sio n o f o th e r in d e p en d e n t forces, it is n ot su itab le for ex a ct forecastin g .*' Its prim ary p u rp o se is to m easu re th e gen eral p a tte rn o f in flu en ce o f m o n etary an d fiscal a ctio n s on several strategic ec o n o m ic variables. Sin ce th e eco n o m y is view ed as b ein g b asically stable, o th e r fa cto rs in flu en cin g to tal spen din g, o u tp u t an d p rices are n ot co n sid ere d to be o f great im p o rta n ce in estim a tin g th e re sp o n se to m o n etary an d fiscal a ctio n s. “ See Andersen and Jordan, pp. 15,23, 24, and Leonall C. Andersen, “Money in Economic Forecasting,” Business Economics (Septem ber 1969), p. 17. Appendix A Explanation of the P rice Equation The price equation (omitting timescripts and lags) is AP = f (D, AP*), where D, demand pressure, is defined as D = AY - (X1 - X). AY is the change in total spending, (X1 — X) is the GNP gap, that is, the difference between potential and actual output, and AP' is anticipated price change. This specification of the price equation is based on standard theoiy of macroeconomic equilibrium. M acroeconomic equilibrium can be depicted graphically as in figure 1. The solid downward-sloping line, X", is the total spending line, which represents the combinations of prices and output consistent with a particular level of total spending, Y. This total spending line can be interpreted as total demand for output. The upward-sloping line, labeled Xs, is the total supply line. This line corresponds to that combination of prices and output which maximizes profits of firms, given the prices of factors of production, the degree of competition among firms and the stock of human and physical capital (defined to embody the state of technology). The intersection of total supply and total demand deter mines the levels of output and prices. The equilibrium price level is that level which equates the amount of output supplied with the amount demanded. The focus of the model is on the change in prices and the 62 Figure I M a c r o e c o n o m i c E qu i l i b r i um (Determination o fO u tp u t and Prices) FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 ch a n g e in o u tp u t. In term s o f figure 1, c h a n g e s in p ric e s a n d o u tp u t are b ro u g h t a b o u t by sh ifts in d e m a n d and/or sup p ly. S in ce X" is d raw n fo r a level o f to tal sp e n d in g , a shift o f th at lin e up w ard a n d to th e rig h t to Xy+ v re p re s e n ts an A in c re a se in to tal sp e n d in g . If th e to tal su p p ly lin e re m ain s fixed, th e effect o f AY o n p ric e s d e p e n d s o n (1) th e m ag n i tu d e o f AY, an d o n (2) th e slo p e o f th e to tal su p p ly lin e, Xs. tw een X1 — Xs a n d th e slo p e o f Xs. A ssu m in g th at th is re la tio n sh ip is a p p ro x im a te ly lin e a r w ith in th e ran g e o f e x p e r ie n c e s in c e 1955, a n d th at th e o bserved v alu es fall on th e su p p ly lin e, th e effect o f a v ariab le slo p e fo r X s c a n b e ap p ro x im a te d by X' — X. In th is w ay th e term [AY — (X1 — X)] b rin g s to g e th e r b o th th e m ag n itu d e o f d e m a n d sh ift a n d the slo p e o f th e su p p ly lin e. T h e p u rp o se o f th e m od el is to e stim a te th e re s p o n s e o f sp en d in g , ou tp u t, a n d p ric e s to m o n e ta iy a n d fiscal a ctio n s, n o t to te st a h y p o th e siz e d stru c tu re . C o n se q u e n tly , ra th e r th a n a tte m p t to d e te rm in e th e s h a p e o f th e total su p p ly lin e e m p irically , its v ariable slo p e is p ro x ied by th e d ifferen ce b e tw e e n p o te n tia l o u tp u t a n d a ctu a l o u tp u t. As d raw n in figure 1, th e re is a o n e -to -o n e re la tio n sh ip b e T h e o th e r term in th e p ric e e q u atio n , a n tic ip a te d p rice c h a n g e, AP*, is c o n sid e re d a s a se p a ra te in flu e n c e o n p ric e s . In te rm s o f figure 1, th e a n tic ip a te d p ric e term is a shift p a ra m e te r fo r th e total su p p ly lin e (an in c re a s e in AP* sh ifts Xs up w ard an d to th e left). In clu d in g it in th is w ay a llow s fo r th e in flu e n c e o f p a st p ric e s o n c u rre n t p ric in g p o lic ie s o f firm s a n d facto rs o f p ro d u c tio n . Appendix B G raphical Illustration of the Model T h e w ork in gs o f th e m o d e l c a n b e d e m o n s tra te d w ith g ra p h ical te ch n iq u e s . Figu re 2 is a re p re s e n ta tio n o f th e c o re o f th e m od el, sh o w in g th e d ete rm in a tio n o f c h a n g e s in sp en d in g , o u tp u t, a n d p ric e s. Panel A o f figure 2 is a g ra p h ica l re p re s e n ta tio n o f the to tal s p e n d in g e q u a tio n w ith AM o n th e h o riz o n ta l axis an d AY o n th e v ertical axis. C h an g e s in AE sh ift th e to tal s p e n d ing lin e. P an el B sh o w s p ric e s (AP) as a fu n ctio n o f AY. A s h o rt-ru n p rice lin e (AP,) is d raw n c o n s is te n t w ith e m p iric a l resu lts sh ow in g th at AP is n o t v e iy sen sitiv e to AY in th e sh o rt ra n . Im p o rta n t d e te rm in a n ts o f th e p o sitio n o f th e s h o rt-ru n p ric e lin e are th e siz e o f th e GNP g ap a n d a n tic ip a te d p rice ch a n g es. T h e lo n g -ru n p ric e lin e (A P(LR)) is d raw n to sh o w th e re la tio n sh ip b e tw e e n AP a n d AY w h e n th e GNP gap is z ero a n d a n ticip a te d p ric e s are eq u al to a c tu a l p ric e s. Its s lo p e (45 d eg rees from its o rig in o n th e AY axis) is b a s e d o n th e m o n e ta rist view th a t in th e lo n g ru n , AM in flu e n c e s only AP. P anel C e x p re ss e s th e to tal sp e n d in g id e n tity in g ra p h ica l te rm s. T o tal s p e n d in g is d ivided b e tw e e n o u tp u t an d p ric e s ; to reflect th is, th e lin e in p a n e l C is d raw n as a 45 In p a n e l D, th e AX, lin e sh o w s th e re la tio n sh ip b e tw e e n m o n ey (AM) a n d o u tp u t (AX) as d erived from th e o th e r th re e p an e ls. T h e e q u a tio n fo r th is lin e is n o t sh o w n in e x h ib it 1 in th e text, b u t it c a n b e d erived from th e o th e r e q u a tio n s o f th e m od el. Figu re 2 is d raw n to re p re s e n t a n in itial eq u ilib riu m fo r a given AM, w h ic h h a s a s so c ia te d w ith it th e sh o rt-ru n p ric e a n d o u tp u t lin es, AP, a n d AX,. T h e effect o f a c h a n g e in AM, given AE, is sh o w n as a m o v e m e n t alo n g th e s p e n d in g lin e in p an el A from © to © . Given th e in itial p ric e lin e, AP,, and th e c h a n g e d AY, th e effect o n p ric e s a n d o u tp u t is sh o w n in p a n e ls B, C an d D as a m o v em en t from © to © . T h is c a se illu stra te s th e im p a c t o f a c h a n g e in AM in th e sh o rt ru n . F o r lo n g e r p erio d s, a n tic ip a te d p ric e c h a n g e s a n d th e GNP gap w ill also ch a n g e ; th ey b e c o m e e n d o g e n o u s va ria b les in a lo n g -ru n m o d e l. T o illu stra te th e e ffe cts fo r th e long ru n, th e lo n g -ru n p ric e lin e , AP(LR), in p a n e l B, is relevan t. T h e in te rp re ta tio n o f th e lo n g -ru n p ric e lin e is th a t c h a n g e s in AM are re fle cte d o n ly in AP, w ith AX d e te rm in e d by c o n sid e r a tio n s o f r e s o u rc e g ro w th a n d te ch n o lo g y . T h e h o riz o n ta l lint; in p a n e ls G a n d D is th e lo n g -ru n re latio n b e tw e e n p ric e s an d o u tp u t. In th e sh o rt-ru n , th e s o lu tio n o f th e m o d el n e e d not lie o n d eg ree lin e w ith its p o sitio n d e te rm in e d by th e m ag n itu d e o f total sp e n d in g (AY). T h e re is a fam ily o f 45 d eg re e lin es, o n e fo r e a c h p o ssib le AY. A lso in c lu d e d in p an el C is a h o riz o n ta l lin e re p re se n tin g th e lo n g -ru n g row th rate of o u tp u t. It is sh o w n a s a h o riz o n ta l lin e to in d ic a te th a t long- th e lo n g -ru n p rice lin e in p a n e l B lin e in p a n e ls G a n d D). However, a (show n as a sh ift o f th e AP a n d AX te n d to m ove e q u ilib riu m to w ard ru n o u tp u t grow th is e x o g en o u sly d e te rm in e d by re s o u rc e o u tp u t lin es, as a n tic ip a te d p ric e s a d ju s t to a ctu a l p ric e s grow th an d tech n o lo g y . a n d th e GNP gap g o e s lo zero . (or th e lo n g -ru n o u tp u t s u c c e s s io n o f sh o rt ru n s lin e s to AP, an d AXJ will th e lo n g -ru n p ric e an d 63 OCTOBER 1986 FEDERAL RESERVE BANK OF ST. LOUIS F ig u r e II Model in Graphical Form (A) S p e n d i ng Equati on Appendix C Alternative P rice Equations T h e m o d e l c o n sis ts , for th e m o st part, o f e q u a tio n s w h ic h have b e e n e stim a te d in p rev iou s stu d ie s. T h e p u rp o se o f th is p a p e r is to c o m b in e th e e q u a tio n s in a w ay w h ic h re p re s e n ts th e g en eral m o n e ta rist view. T h e p rim ary d istin g u ish in g featu re o f th is m o d el, o th e r th a n th e re d u ce d -fo rm to tal sp e n d in g e q u a tio n , is th e in c lu sio n o f a p ric e a n ticip a tio n s v ariable in th e p ric e e q u a tio n . Tw o altern ativ e m e th o d s o f in tro d u c in g p ric e a n tic i p a tio n s w ere c o n sid e re d . O ne altern ativ e b y p a sse s the p re c is e form o f th e p ric e a n ticip a tio n s fu n c tio n a n d u se s th e lo n g -term m ark et in te re st rate (yield o n c o rp o ra te Aaa b on d s) as an in d e p e n d e n t v ariable in th e p ric e e q u a tio n . T h e o th e r a ltern ativ e b y p a sse s b o th p ric e e x p e c ta tio n s an d in te re st rates, an d in tro d u c e s c h a n g e s in m o n ey as an 64 (B) Price E q u a t i o n in d e p e n d e n t v ariable in th e p ric e e q u a tio n . S u c h a s p e c i fica tio n allow s m o n e ta ry a c tio n s to serve a s a p ro x y for a n ticip a te d p ric e s . A Market Interest Rate in the Price Equation T h e first altern ativ e re p la c e s th e p ric e a n tic ip a tio n s vari ab le w ith th e lo n g -te rm m ark et in te re st ra te .' T h e ra tio n a le is th at th e p ro c e ss o f p ric e a n tic ip a tio n s fo rm a tio n is so co m p le x th a t it d efies m e a s u re m e n t. H ow ever, th e re se e m s to b e ag re em e n t th a t th e level o f m ark e t in te re s t ra te s re fle cts a n ticip a te d p ric e c h a n g e s , h o w ev er fo rm ed . T h u s th e m ark et in te re st ra te c a n b e u se d a s a p ro x y fo r p rice a n ticip a tio n s. 'The suggestion for using the interest rate in the price equation came from the Money and Banking Workshop at the University of Chicago. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Table 11 Alternative Price Equations (Dependent Variable — APt) Sample Period: 1/1955— IV/1969 ________ Independent Variable “ Dt_, AP*t (R\_,X,_,).01 SAM, , Constant R2 * * 2.70 .87 Price Anticipations Specification’ .09 .86 (7.07) (9.18) (8.55) .11 1.93 .88 .09 Interest Rate Specification2 (8.72) (8.73) (4.10) .06 Money Stock Specification3 2.61 2.11 .86 (4.61) (7.82) (4.22) SE DW 1.07 1.41 1.04 1.49 1.11 1.37 ’Where i goes from 0 to 5. 2Where i goes from 0 to 10. Where i goes from 0 to 9. ‘ Not estimated for this equation. NOTE: Regression coefficients are the top figures; their “t” statistics appear below each coefficient, enclosed by parentheses. R2 is the percent of variation in the dependent variable which is explained by variations in the independent variable. SE is the standard error of the estimate. DW is the Durbin-Watson statistic. S in ce in te re st rate s re fle ct factor's o th e r th a n p ric e a n tic i p atio n s, in clu d in g th e in te re st rate d o es n o t prov id e a c le a n m e a su re o f p ric e a n ticip a tio n s . U sing th e m ark et in te re st rate allow s th o se facto rs in flu e n c in g th e real rate o f in te re st to e n te r in d ire ctly as a n in flu e n c e o n p ric e s . In g en eral, how ever, it h a s b e e n arg u ed th a t th e real r ate o f in te re st is very stable. Fo llow in g th is reaso n in g , th e p ric e e q u a tio n w as e sti m ated by in clu d in g th e lo n g -te rm in te re st rate. T h e re su lts are sh o w n in tab le 11. T h e co e fficie n t o f th e inter e st rate variable is sig n ifican t at th e five p e rc e n t level for th is s p e c i ficatio n , a n d th e su m o f th e c o e fficie n ts fo r th e d em a n d p re ssu re v ariable is ro u g h ly th e sa m e a s fo r th e p ric e a n tic i p a tio n s v ersio n o f th e e q u a tio n . H ow ever, th e le n g th o f th e lag s tru c tu re is lo n g er, in d ic a tin g th a t th e re s p o n s e o f p ric e s to ch a n g e s in d em a n d p re s su re m ay b e slo w e r th a n in th e b a s ic e q u atio n . B ut th is n e e d n o t im ply th a t p rices are slo w e r to r e sp o n d to m o n e ta iy a c tio n s , s in c e th e m ag n i tu d e o f th e in te re st rate c o n trib u tio n to p ric e c h a n g e is sm a lle r th a n w ith th e p ric e a n tic ip a tio n s s p e cific a tio n . Money in the Price Equation Several o bserv ers have b e e n critic a l o f p ric e e q u a tio n s th at d o n o t in c lu d e m o n e ta iy v ariables d irectly . As sh o w n in th e text, e x clu d in g m o n e ta iy v ariab les fro m th e p rice e q u a tio n d o es n o t n e c e ss a rily im p ly a n o n -m o n e ta ry th e o ry o f in flatio n .2 S u ch a c o n c lu s io n c a n n o t b e d erived by e x a m in in g th e p ric e e q u a tio n alo n e, b u t re q u ire s an e x a m in a tio n 2See Fand, “Some Issues in Monetary Economics,” pp. 20-23. o f th e w h o le m o d el, a n d th e lin kages b e tw e e n m o n ey an d p ric e s in p a rticu la r. T h e s e c o n d altern ativ e th a t is c o n sid e r e d is b a s e d o n th e c e n tra l p ro p o sitio n o f th e q u a n tity th e o iy — th a t ch a n g e s in m o n ey a re u ltim ately re fle cte d in ch a n g e s in th e p rice level. A ccord in gly, c u rre n t a n d p a s t c h a n g e s in m o n e y are u se d as a p roxy to m e a su re a n tic ip a te d m o v e m e n ts in p ric e s .3 T h o u g h th is ra tio n a le fo r in c lu d in g m o n ey is s o m e w h at n a rro w e r th a n th at p ro p o se d by s o m e m o n e ta iy e c o n o m ists, th e d ire c t an d in d ire c t e ffe cts o f m o n e y a re b e in g m e a su re d o n c e it is in c lu d e d in th e p ric e e q u a tio n . T h e p ric e e q u a tio n in corp or a tin g c u rr e n t a n d lagged v alu es o f c h a n g e s in m o n e y is sh o w n in tab le 11. E x c e p t for th e c u rre n t q u arter, th e c o e fficie n ts are sig n ifican t fo r n in e lagged qu artet's. T h e effect o f in c lu d in g c h a n g e s in m o n ey lo w ers th e su m o f th e c o e fficie n ts o n th e d e m a n d p re s su re v ariable, h ow ever. T h e overall e x p la n a to ry p o w e r o f th e e q u a tio n is ab o u t th e sam e as for th e p rice a n ticip a tio n s m o d el. Comparison o f the Alternatives T o co m p a re th e p ric e e q u a tio n in th e text w ith th e tw o altern ativ e s in th is ap p e n d ix , th e m o d el w as sim u la te d w ith e a c h o f th e th re e d ifferen t s p e c ific a tio n s from 1965 th ro u g h 1969. T h e p e rio d sta rtin g in 1965 is u se d b e c a u s e th e re la tive tra ck in g ab ility o f th e m o d e ls d u rin g a p e rio d o f a c c e le r a tin g in flatio n is e sp e c ia lly relevan t in a s se s sin g th e c u rre n t 3This suggestion was made by Professors David Fand and Allan Meltzer. 65 FEDERAL RESERVE BANK OF ST. LOUIS O CTOBER 1986 Table 12 GNP Price Deflator Alternative Ex Post Simulations: Actual Minus Predicted (compounded annual rates of change) Price Anticipations Interest Rate Money Stock Specification Specification Specification 0.6 1965 I 0.3 -0 .5 II 0.2 -0 .2 -1 .0 -0 .7 III -1 .0 -1 .7 IV -0 .2 -0 .4 -1 .0 1.0 1966 I 1.0 0.4 1.4 II 1.5 1.0 III 0.8 0.8 0.6 IV 0.2 0.1 0.3 1967 I -0 .3 -0 .3 -0 .2 II -0 .8 -0 .8 -0 .8 III 1.0 1.1 0.9 IV 1.3 1.4 1.1 0.4 1968 I 0.4 0.3 II 0.5 0.7 0.6 III 0.2 0.7 0.4 IV 0.2 0.9 0.3 0.6 1969 I 1.4 0.7 II 0.7 1.6 0.8 III 0.8 1.7 1.0 IV 0.1 0.9 0.5 Average Absolute Error 0.60 1965-69 0.86 0.71 1965-67 0.71 0.74 0.79 0.44 1968-69 1.04 0.58 Root Mean Squared Error 1965-69 0.50 0.96 0.63 0.67 1965-67 0.76 0.79 1.27 1968-69 0.25 0.39 e c o n o m ic s itu a tio n . S in ce th e p ric e e q u a tio n is th e o n ly p art th a t v aries from o n e m o d e l to th e n ex t, o n ly th e re su lts fo r th e rate o f c h a n g e o f p ric e s are re p o rte d (see ta b le 12). T h e p ric e a n ticip a tio n s s p e c ific a tio n h a s th e sm a lle st average a b so lu te e rro r a n d th e sm a lle st ro o t m e a n sq u a re d e rro r fo r th e p e rio d . D u rin g th e last tw o y e a rs o f th e p erio d , 1968 a n d 1969, e a c h o f th e altern ativ e s p e c ific a tio n s te n d s to u n d e re s tim a te p ric e c h a n g e s . H ow ever, fo r 1968 a n d 1969, th e p ric e a n ticip a tio n s s p e c ific a tio n ag ain lias b o th th e sm a lle st average a b so lu te e rro r a n d ro o t m e a n sq u a re d error. Conclusions An e x a m in a tio n o f th e m o d el re fle ctin g th re e d ifferen t s p e c ific a tio n s fo r th e p ric e e q u a tio n in d ic a te s th a t n o n e o f th e s p e c ific a tio n s is cle a rly s u p e rio r as ju d g ed by c o n v e n 66 tio n a l c rite ria . A p o licy m a k e r m ig h t w ell c o n s id e r th e re su lts p ro v id ed by e a c h o f th e th re e . W h en s im u la tio n s a re p e rfo rm e d fo r 3 0 -y e ar p e rio d s b e g in n in g in 1970, th e p ric e a n tic ip a tio n s v ersio n las p re s e n te d in th e text) a p p r o a c h e s c lo s e s t a lo n g -ru n c la s sic a l so lu tio n . F o r th e o th e r tw o s p e c ific a tio n s th e u n e m p lo y m e n t ra te d o e s n o t sta b iliz e at th e s a m e level fo r a ltern ativ e g ro w th ra te s o f m o n ey . T h e s e tw o a ltern a tiv e s y ie ld th e sam e eq u ilib riu m grow th ra te s o f o u tp u t fo r altern ativ e grow th ra te s o f m o n ey , b u t s in c e th is ra te is a p p ro a c h e d asy m p to tically , u n e m p lo y m e n t s ta b iliz e s at a d iffe re n t rate fo r e a c h altern ativ e g row th ra te o f m o n e y .4 “Supplementary materials relating primarily to the long-run simula tions are available on request. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Leonall C. A ndersen: A Biographical Sketch A n d y w as b orn in M arshall, M innesota, on N o v e m b e r 4, 1924. He atten d ed Gustavus A d olp h u s C ollege, w h e re he received a B.S. in business ad m in istration in 1949; th ree years later, h e e arn ed an M.S. in E con om ics from the U niversity o f Illinois. He jo in e d the faculty at St. O laf C ollege in N orth field, M innesota, in 1955 as an assistant professor, w h ile sim u ltaneou sly p u rsu ing graduate w o rk at the U niversity o f M innesota. In 1962, having been aw a rd ed his Ph.D., he jo in e d the Research D ep artm en t at the Federal Reserve Bank o f St. Louis. E xcept fo r a on e-year leave in 1969 as a sen ior staff e c o n o m ist at the C ou n cil o f E con o m ic Advisers in W ash ington , D.C., A n d y re m ain ed in St. Louis for 16 years. At the St. Louis Federal Reserve Bank, he rose from e con o m ist (1962-65) to sen ior ec o n o m is t (196566), to vice p resid en t (1966-71), to sen ior vice p resid en t (1971-74), to e c o n o m ic adviser to the p resid en t (1974-78). In 1978, A n d y retired from the Bank to accep t an a p p oin tm en t as p ro fessor o f banking in the C ollege o f Business A d m in istration at the U niversity o f Florida. In 1981, he retu rn ed h o m e to M in n esota to his alm a m ater, Gustavus A d olp h u s, as a p ro fessor o f econ om ics, a p ositio n he h eld until his death on O ctob er 27, 1985. 67 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 Leonall C. A ndersen: A Selected Bibliography “T h e In c id en c e o f M on etary an d Fiscal M easures on the Structure o f O utput,” T h e Review o f E c o n o m ic s a n d Statistics, August 1964. “ F ed eral Reserve O p en M arket T ransaction s an d th e M o n e y Sup p ly,’’ Federal Reserve Bank o f St. Louis Review, A p ril 1965. “ Im p lem en ta tio n o f F ed eral Reserve O p en M arket P o lic y in 1964’’ (w ith Jules M. Levine), Federal Reserve Bank o f St. Lou is Review, June 1965. “ Seasonal M ovem en ts in Finan cial Variables — Im p a ct o f F ed eral Reserve an d Treasu ry,” B u siness a nd G o v ern m en t Review, U n iver sity o f M issouri, July-August 1965. “A P rin cipal C om p o n en ts T est o f Bank Debits as a L ocal E con o m ic In d ic a to r” (w ith W illard T. C arleton), T h e Jo u rn a l o f B usiness, O ctob er 1965. “ Federal Reserve O p en M arket O perations in 1965: O bjectives, Actions, and A c c o m p lish m e n ts ” (w ith Elaine R. G oldstein), F e d eral Reserve Bank o f St. Louis Review, June 1966. “A M e th o d o f U sing D iffu sion In d exes to In d ica te th e D irectio n o f N ational E con o m ic A ctivity ” in P a pers a n d P ro ceed in g s, Business and E con o m ic Statistics Section, An nual M e etin g o f the A m erican Statistical Association, August 1966. “A T est o f M o n e y M arket C on d ition s as a M eans o f Short-Run M o n e ta iy M a n a gem en t” (w ith Jules M. Levine), T h e National Banking Review, S ep tem ber 1966. “T h re e A p p ro a ch es to M o n e y Stock D eterm in a tion ,” F ed eral R e serve Bank o f St. Louis Review, O ctob er 1967. “ F ed eral Reserve D efensive O p eration s an d Short-Run C on tro l o f the M o n e y Stock,” T h e Jo u rn a l o f Political E co n o m y , M arch -A p ril 1968. “ 1967 — A Y ear o f C onstraints on M o n e ta iy M a n a g e m en t” (w ith M ich ael O. Rigg), Federal Reserve Bank o f St. Louis Review, M ay 1968. “T h e M o n e ta iy Base — E xplan ation and A n alytical U se” (w ith Jerry L. Jordan), Federal Reserve Bank o f St. Louis Review, August 1968. “ M on etary and Fiscal A ctions: A T est o f T h e ir Relative Im p o rta n ce in E con o m ic S tabilization ” (w ith Jerry L. Jordan), F ed eral Reserve Bank o f St. Louis Review, N o v e m b e r 1968. “T w o M o n e ta iy T erm s D efin ed ,” Banking, Journal o f the A m erican Bankers Association, M arch 1969. “ M o n e ta iy and Fiscal Actions: A T est o f T h e ir Relative Im p o rta n ce in E con o m ic Stabilization — R ep ly” (w ith Jerry L. Jordan), Federal Reserve Bank o f St. Louis Review, A p ril 1969. 68 FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 “Asset M an a gem en t an d C om m ercia l Bank P ortfolio Behavior: T h eo iy and P ra ctice” (w ith Albert E. Burger), T h e Jo u rn a l o f F inan ce, M ay 1969. “ C om m en t on T h e C h oice o f In term ed iate E con o m ic Targets,’ ” A m erica n E c o n o m ic Review, M ay 1969. “ M o n e ta iy V elocity in E m pirical Analysis: D iscu ssion,” Controlling M onetary A ggrega tes, Federal Reserve Bank o f Boston, June 1969. “A d d ition a l E m pirical E viden ce on the R everse-Causation A rgu m en t,” Federal Reserve Bank o f St. Louis Review, August 1969. “An Evaluation o f the Im pacts o f M o n e ta iy and Fiscal P olicy on E con o m ic A ctivity,” P apers a nd P ro ceed in gs, Business and E c o n o m ic Statistics Section, Annual M e etin g o f the A m erican Statisti cal Association, August 1969. “ M o n e y and E con o m ic F orecastin g,” B usiness E co n o m ics, S ep tem ber 1969. "T h e D evelo p m en t o f E xp lan atoiy E con o m ic H ypoth eses fo r M o n e ta iy M a n a gem en t” (w ith Albert E. Burger), S o u th ern Jo u rn a l o f B usiness, O ctob er 1969. “ M o n e y M arket C on d ition s as a G uide fo r M on etary M a n a ge m en t” in T argets a nd Indicators o f M onetary Policy, Karl Brunner, ed., C h an d ler Publishing C om pany, San Francisco, California, 1969. “A M on etarist M o d e l fo r E con o m ic S tabilization ” (w ith K eith M. Carlson), Federal Reserve Bank o f St. Louis Review, A p ril 1970. “ Selection o f a M o n e ta iy A ggregate fo r Use in the FO M C D irective,” O pen M arket Policies a nd O perating P ro c e d u re s — S ta ff Studies, Board o f G overnors o f the F ederal Reserve System, July 1971. “ Im p act o f M on eta ry P olicies on the G row th o f Banking in the ’70s,” Financial Analysts Jo u rn a l, July-August 1971. “A M on etarist V iew o f D em a n d M an agem en t: T h e U n ited States E xp erien ce,” Federal Reserve Bank o f St. Louis Review, S eptem ber 1971. “T h e St. Louis E con o m etric Forecastin g M o d e l,” B usiness E c o no m ics, Sep tem ber 1971. “A M on etarist V iew o f C urrent Stabilization P olicy ,” A rkansas B usi n ess and E c o n o m ic Review, M ay 1972. “A Look at T e n M on th s o f Price-W age C on trols,” F ed eral R eseive Bank o f St. Louis Review, June 1972. “A n E con o m etric Analysis o f the R elation o f M o n e ta iy Variables to the B ehavior o f Prices and U n e m p lo y m e n t” (w ith K eith M. Carlson), T h e E c o n o m e tric s o f P rice D eterm ination, O tto Eckstein, ed., Board o f G overnors o f the F ederal Reserve System, June 1972. 69 FEDERAL RESERVE BANK OF ST. LOUIS 70 OCTOBER 1986 “T h e A p p ro p ria te T im e Fram e fo r C on tro llin g M o n eta ry Aggregates: T h e St. Louis E v id e n ce ” (w ith D enis S. Karnosky), Controlling M onetary A g g reg a tes II: T h e Im plem en ta tion , C o n feren ce P ro ceed in gs, Federal Reserve Bank o f Boston, S ep tem b er 1972. “ P rop erties o f a M on etarist M o d e l fo r E con o m ic Stabilization ,” K redit u n d Kapital, P roceed in gs o f th e First K on sta n zer Sem inar on M o n e ta iy T h e o ry and M o n e ta iy Policy, U n iversity o f K on stanz, W est Germ any, Karl Brunner, ed., 1972. “ Statistics fo r Public Financial P olicy,” Statistics: A G uide to the Unknown, Judith M. Tanur, ed., H old en -D a y Inc., San Francisco, California, 1972, pp. 321-25. “A C om p a rison o f Stabilization P olicies 1966-67 an d 1969-70," J o u r nal o f M oney, Credit and Banking, Vol. 5, Part I, February 1973, pp. 26-38. "T h e State o f the M on etarist D ebate,” Federal Reserve Bank o f St. Louis Review, Sep tem ber 1973. "C o m m e n t on 'A N ote on the Effects o f G overnm en t F in an ce on A ggregate D em an d for G oods and Services,’ ” Public F in a n ce, In ternational Q uarterly Journal, T h e Hague, N etherlands, N o. 3-4, 1973. "A M on etarist A p p ro a c h to F orecastin g,” M eth o d s a n d T e c h n iq u e s o f B u siness F o reca stin g (revised ed ition ), Butler, Kavesh an d Platt, eds., Prentice-H all, Inc., E n g le w o o d Cliffs, N e w Jersey, A p ril 1974. “T h e U.S. E con o m ic Outlook: P o lic y A sp ects,” R eco rd , C o n feren ce Board, June 1974. "St. Louis M o d e l R evisited ” (w ith Keith M. Carlson), International E c o n o m ic Review , June 1974. “A M on eta ry M o d e l o f N om in al In co m e D eterm in a tion ,” Federal Reserve Bank o f St. Louis Review, June 1975. “ O bserved In c o m e V elocity: A M is u n d e rs to o d Issue in M o n e ta iy P olicy ,” Federal Reserve Bank o f St. Louis Review, August 1975. “ Selection o f a M on eta ry A ggregate fo r E c o n o m ic Stabilization ,” Federal Reserve Bank o f St. Louis Review, O ctob er 1975. “ Is T h e re a Capital Shortage?: T h e o ry and R ecen t E m p irical Evi d e n c e ,” Jo u rn a l o f F inan ce, M ay 1976. "D ilem m as in th e T ren d s o f Capital and P rod u ctivity,” T h e Jo u rn a l o f Portfolio M anagem ent, Sum m er 1976. "T h e O u tlook for Long-R un E con o m ic G row th ,” B u siness E c o no m ics, S ep tem ber 1976. "S om e C on sid eration s in Using M o n e ta iy A ggregates in the Im p le m en tation o f M o n e ta iy P o lic y ” (w ith D enis S. Karnosky), Federal Reserve Bank o f St. Louis Review, Sep tem ber 1977. FEDERAL RESERVE BANK OF ST. LOUIS OCTOBER 1986 “A M o n e ta iy In terpretation o f In fla tion ” (w ith Denis S. Karnosky), Analysis o f Inflation: 1965— 1974, Joel Popkin, ed., N ation al Bureau o f E con o m ic Research, 1977. ' An E xplanation o f M ovem en ts in the Labor-Force Participation Rate, 1957-76,” Federal Reserve Bank o f St. Louis Review, August 1978. 71