The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
The R e v ie w i.s published 10 times per year by the Research Department o f the Federal Reserve Bank o f St. Louis. Single-copy subscriptions are available to the public free o f charge. Mail requests fo r subscriptions, back issues, or address changes to: Research Department, Federal Reserve Bank o f St. Louis, P.O. Box 442, St. Louis, Missouri 63166. Articles herein may be reprinted provided the source is credited. Please provide the Bank’s Research Department with a copy o f reprinted material. Potential Output and the Recent Productivity Decline JOHN A. TATOM X ^ ) T E N T I A L o u tp u t re fe r s to th e rea l g ro ss n ation al p r o d u c t (G N P ) that is p r o d u c e d i f the e c o n o m y operates u n der h ig h -e m p lo y m e n t c o n d i tio n s. M ea su res o f p o te n tia l ou tp u t d e p e n d on m easures o f available resou rces, such as capital and labor, and re lia b le estim ates o f the rela tion sh ip b e tw e e n national output and the e m p lo y m e n t o f resou rces. S in ce 1973, the grow th o f p rod u ctivity (m easured as ou tput p er unit o f labor) has s lo w e d substantially (see chart 1), raising d o u b t a bou t the relation sh ip o f inpu t to ou tput and, th erefore, the m easu rem en t of the n ation ’ s poten tial G N P . T h is issu e is o f co n s id e r a ble im portan ce as it bears on the traditional con cern o v e r the d e g re e of resou rce u n deru tilization and the a ssociated ou tput losses in the e co n o m y . An accurate assessm ent o f poten tial output is essential to d ete r m in e the e x p e cte d gain in ou tput from a p o lic y in te n d e d to a ch ie v e full em p loy m en t. T h e relation sh ip b e tw e e n resou rce su p p lies and poten tial output also is im portant in analyzin g the ou tp ut gain from su p p ly -sid e p o licie s to increase the su pp ly o f resou rces through in crea sed w ork, saving and investm en t. dram atically sin ce 1978, it is im portant to verify that the ea rlie r e m p irica l results are co n s iste n t w ith recen t p rod u ctivity e x p e rie n ce , as w e ll as to assess the im pa ct o f this sh ock on poten tial G N P . A lso, recen t revision s o f the G N P a ccou n ts incorporate n e w inform ation on ou tp ut and in v o lv e som e c o n ceptu al chan ges that req u ire revision s in potential G N P m easures. In ad d ition , sin ce 1977 som e m o d i fications have o cc u rr e d in the m eth od s u sed b y this Bank to m easure poten tial output. T h e revision s and m od ification s are d e s c rib e d b e lo w . ENERGY PRICE SHOCKS AND PRODUCTIVITY T h is B ank’ s m easure o f p oten tial ou tp ut differs from others in that it p rovid es d irect estim ates o f the effects o f lab or force grow th, capital accum ulation and ch an ges in the relative cost o f en erg y resou rces on p rod u ctivity and e c o n o m ic ca p a city .1 T h e stabil ity of the in p u t-ou tpu t relation sh ip on w h ich this m easure is b ased, and its ability to fully a ccou n t for the unusual p rod u ctiv ity d ev elop m en ts d urin g the last d e ca d e , p r o v id e d su p p ort for the cr e d ib ility o f past estim ates. S in ce en erg y costs have in crea sed A sharp in crease in the relative p rice o f en erg y causes a red u ction in the ou tput (p rod u ctiv ity ) of existin g labor and capital resou rces, or e c o n o m ic capacity. T h e particular ch a n n els through w h ich this ch a n ge occu rs vary from firm to firm, but in clu d e ch a n gin g p ro d u ctio n m eth od s to re d u ce the use o f h ig h er-cost en erg y, the clo s in g o f plants re n d e re d u n p ro fita b le , r e d u c e d op tim a l and actual u se of existin g facilities, and the d iv e rsio n o f labor and capital resou rces to uses that e c o n o m iz e on h igh erco st en ergy. T h e s e ch an ges result in less output b e in g p ro d u ce d d e sp ite an in itially u n ch a n g ed avail ability of d o m e stic capital and lab or resou rces. As a result, m easures o f p rod u ctivity such as ou tput p er w orker, p er hour, or p er unit o f capital, d e clin e . T h e rise in e n erg y p rice s also in d u ce s a p ercen ta ge in crease in the nom inal p rices o f ou tput eq u a l to the p ercen ta ge d e clin e in p ro d u ctiv ity or poten tial out- 'T he original measures used by this Bank and the methods oi their construction are explained in Robert H. Rasche and John A. Tatom, “ Energy Resources and Potential GNP,” this Review (June 1977), pp. 10-24. The theoretical basis for the energy price effect is developed in Robert H. Rasche and John A. Tatom, “ The Effects of the New Energy Regime on Econom ic Capacity Pro duction and Prices,” this Review (May 1977), pp. 2-12. These hypotheses are further elaborated, and international evidence supporting them are presented in Rasche and Tatom, “ Energy Price Shocks, Aggregate Supply and Monetary Policy: The Theory and International Evidence,” in Karl Brunner and Allan H. M eltzer, eds., Supply Shocks, Incentives and National W ealth, C arnegie-R ochester C on feren ce Series on Public Policy, Vol. 14 (1981), pp. 9-93. 3 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 C h a rt 1 Output per Hour (Private Business Sector) 1972 d o l l a r s L a te s t d a t a p lo tte d : 3 rd q u a rte r put, s in c e less ou tp u t is p r o d u c ib le tor a g iv e n s u p p ly of m on ey. In a dd ition , the d e c lin e in p rod u ctivity shifts the d em an d for labor and capital resou rces d ow n . In the short run, th ese shifts are reflected in a fall in the real w ages o f w orkers and a d e c lin e in th e va lu e o f exist in g plan t and e q u ip m e n t relative to its rep la cem en t cost. O v er a lon g er p e rio d , the capital stock available p er w orker w ill d e clin e from the le v e l that w o u ld oth erw ise have o ccu rred , so that the long-run d e c lin e in p oten tia l output, labor p rod u ctiv ity and real w ages is larger than the initial d e clin e . T h e effe ct o f a rise in the relative p rice o f en erg y on p rod u ction is m a n ifested in a p rod u ction fu n ction approach through red u ction s in inputs (esp ecia lly 1972 d o l l a r s S o u rc e : U.S. D e p a rtm e n t o f L a b o r r e d u c e d e n e r g y u s a g e ), o r th ro u g h c h a n g e s in p ro d u ctiv e e ffic ie n c y or ca p acity that are “ d is e m b o d ie d ,” that is, n ot a sso cia te d w ith ch a n g e s in the use o f p h y sica l inputs such as labor, capital or en ergy. E arlier studies have p r o v id e d an u n b ia sed estim ate o f the e ffe c t o f a rise in the relative p rice o f en e rg y on ou tput that supports the e n e rg y p rice / e c o n o m ic cap acity h y p o th e sis.2 B efore re-exam in in g 2An elaborate review o f other analyses o f energy price effects on the econom y is presented in Rasche and Tatom, “ Energy Price Shocks, Aggregate Supply,” pp. 16-33. A more recent critique o f the analysis here is Ernst R. Berndt, “ Energy Price Increases and the Productivity Slowdown in United States Manufactur ing,” Federal Reserve Bank o f Boston, The Decline in Produc tivity Growth, Conference Series No. 22 (June 1980), pp. 60-89. Berndt finds no effect o f higher energy prices on manufactur ing productivity, in contrast to the evidence in John A. Tatom, JANUARY 1982 FEDERAL RESERVE BANK OF ST. LOUIS the p rod u ction fu n ction estim ates, h ow ev er, it is u sefu l to re v ie w recen t revision s in the data series u sed to estim ate potential G N P. THE RECENT GNP AN D CAPITAL STOCK REVISIONS From tim e to tim e, the U.S. D ep artm en t o f C o m m e r c e a n n o u n c e s m a jor r e v is io n s in th e G N P accou n ts b a sed on n ew sou rce data, n e w estim ating p ro ce d u res and d efin ition al or con cep tu a l ch anges. T h e latest rev ision w as p u b lis h e d in D e c e m b e r 1980.3 T h e basis o f the recen t rev ision was n ew inform a tion from the 1972 inp u t-ou tpu t tables, the 1977 e c o n o m ic cen su ses o f various industries, and in for m ation from the 1973 and 1976 T axpayer C o m p li a n ce M ea su rem en t Program . In a ddition , G N P was r e d e fin e d to in c lu d e th e re in v e s te d ea rn in gs o f in corp ora ted foreig n affiliates o f U.S. d irect in v e s tors and to e x clu d e th ose o f in corp ora ted U.S. affil iates of foreign d irect investors. T h e red efin ition o f G N P prim arily affects th e m easu re o f in c o m e origin ating in the rest o f the w orld , w ith little effe ct o f the m ea su rem en t o f ou tp u t from the n ation ’ s private sector. A n oth er im portant part o f the revision was in gross private d o m e stic investm ent. T h e rev ision o f this m easu re was largely d u e to re v is e d estim ates o f p r o d u ce r d u rab le e q u ip m e n t in vestm en t. At the sam e tim e, a con cep tu a l ch an ge occu rred , shifting the ou tput and investm en t in h otels and m otels from the residen tial to the n on resid en tial sector. E x ce p t for the treatm ent o f re in v e s te d earn in gs abroad, h o w e v e r, the revision s of G N P prim arily affect data b e g in n in g in 1968. “ The Productivity Problem,” this Review (September 1979), pp. 13-14. Unfortunately, as Berndt notes, a major share of energy resources is classified as raw materials in his data set, and his analysis can be easily extended to show that most of the productivity decline he analyzes is due to an increase in the relative price of these “ raw materials.” Berndt also claims to show that an observed decline in the value of claims on existing physical capital relative to the replacement cost, as hypothesized above, is also not explained by energy price increases. His theoretical analysis is flawed by the omission o f a significant output effect that substantially raises the magnitude o f his esti mate o f the effect of higher energy prices on the value o f existing capital. 3For a discussion o f these revisions, see Keith M. Carlson, “ Re cent Revisions o f GNP,” this Review (March 1981), pp. 27-32; and “ The National Income and Product Accounts of the United States: An Introduction to the Revised Estimates for 1929-80,” Survey o f Current Business (D ecem ber 1980), pp. 1-26. Table 1 Revisions in Real Net Capital Stock and Private Business Sector Output (selected years)1 Year Upward revision of capital stock at beginning of year Upward revision of private business sector output 1950 2.1% 0.3% 1955 2.0 0.6 1960 1.8 0.5 1965 2.9 0.2 1970 2.9 0.2 1975 4.9 2.8 1Figu res are the percentage increase of 1980 revised data over data available in 1977. T h e n e w sou rce inform ation a ffected m easures o f the n ation ’ s capital stock as w e ll, e sp e cia lly after 1967. T h e reclassification o f h otel and m otel capital stocks is the prim ary sou rce o f ch an ges in the m ea sures p rior to 1967. W h ile the le v e l o f the n ation’ s n et n o n resid en tia l private capital stock (constant p rices) was raised b e ca u se o f these ch a n ges, the grow th rate was ch a n g e d very little p rior to 1973. F or e xa m p le, the re v ise d data sh ow a 4.2 p e rce n t annual rate o f grow th from 1948 to 1968, the same as earlier data. F rom 1968 to 1973, the re v is e d data indicate grow th o f the n et capital stock at a 4.4 p ercen t rate, up from 4.0 p e rce n t in the earlier data. From 1973 to 1978, the rev ised capital stock sh ow s that capital form ation s lo w e d to a 3.1 p ercen t rate. E arlier data sh ow the sam e exten t o f s lo w in g in capital form ation to a 2.7 p e rce n t rate from 1973 to 1978. As a result, the co n c lu s io n o f earlier research that capital form a tion slo w e d s u b se q u e n t to 1973, e sp e c ia lly w h e n m easu red relative to labor fo rce grow th, has b e e n u n affected b y the revisions. T h e rate o fg r o w th o fth e capital stock, h o w e v e r, has b e e n som ew h at faster sin ce 1968 than earlier estim ates s h o w e d ; this co u ld affect earlier estim ates o f inpu t-outpu t relationships. T a b le 1 sh ow s the exten t of b o th the u pw ard rev i sion o f the con stan t-dollar net stock o f fixed n o n re si dential private capital at the b e g in n in g o f the year and the private b u sin ess sector ou tput for data u sed in 1977 as co m p a re d w ith the recen t revision s. T h e capital stock has b e e n re v ise d upw ard relatively m ore than output. 5 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 THE PRODUCTION FUNCTION T h e basis of this Bank’ s poten tial ou tput estim ates is a p rod u ction fu n ction for private b usin ess sector (PBS) output that relates output to hours o f e m p lo y m ent, the utilization o f capital, and en ergy. A va il able m easures o f en erg y ten d to b e b rok en d o w n b y types o f users, such as residen tial, com m ercia l and industrial. N o en erg y m easures exist that are d e tailed b y p rod u ction vs. con su m p tion use b y h o u se h old s, or b y p ro d u cin g sectors lik e the m an ufac turing and private b u sin ess sector. S in ce en e rg y m easures co m p a tib le w ith existin g data on sectoral ou tput and em p loy m en t o f lab or and capital d o not exist, a “ first-order c o n d itio n ” for en erg y e m p lo y m en t is u s e d to elim in ate the quantity o f en erg y from th e p r o d u c t io n fu n c t io n , r e p la c in g it w ith th e relative p rice o f en ergy. Form ally, th e estim ated eq u a tion is o f the form , Table 2 Indirect Least Squares Estimation of a Cobb-Douglas Production Function Production Function: Xt = A h“ kf Et where X h k E A a, f3, y ert, = = = = = = output hours of all persons utilized capital stock energy input scale factor output elasticity of hours, capital, and energy, respectively = trend growth rate per year or per quarter = time period r t First-Order Condition for Energy: Pe/P = yX/E, ( 1 ) In X t = /8o + /8iln ht + y8oln kt + /J3I11 (Pe/P)t + /3-tln t, w h e re X t is PBS ou tput in p e rio d t, ht is hours o f all person s, kt is the u tilized n et n on resid en tial capital stock (constant p rices), the p rod u ct o f the F ed eral R eserve Board m anufacturing ca p acity utilization rate and the capital stock in p la ce at the e n d o f p e rio d t-1, and P e/P is the relative p rice o f en erg y, fou n d b y d eflatin g the p ro d u ce r p rice ind ex for fuel, p o w e r and related p rodu cts b y the im p licit p rice deflator for private b usin ess sector output. T h e t term is a tim e trend in ten d ed to capture the rate o f te ch n o lo g y ch a n ge. W h en eq u a tion 1 is d e riv e d from a C o b b D ou glas p rod u ction fu n ction , the /3s in eq u a tion 1 are related to the ou tput elasticities o f the inputs, as sh ow n in table 2. where Pe/P = the price of energy relative to the price of output Linear model: In Xt = In A - CL + t— 1-7 l-y In ht + B -r— l-y In kt + rt In (Pe/P), Estim ates o f the annual p ro d u ctio n fu n ction u sing the rev ised data for the p eriod s 1949-73, 1949-75 and 1949-80 are sh ow n in table 3.4 T h ere are three n o te w orth y revisions in the estim ates. First, the co e ffi cie n t on the relative p rice o f en erg y and estim ate o f the ou tp u t ela sticity o f e n e rg y are sm aller in a bsolu te value, th ou gh not in a statistically signifi cant sen se, w ith the n e w m easures o f ou tput and capital. In the earlier estim ation for 1949-73 and 1949-75, y is 11.7 p e rce n t (t = 1.92) and 12.0 p ercen t (t = 5.66), re sp e ctiv e ly . S e co n d , the autocorrelation adjustm ent, p, is sm aller than b e fo r e (0.63 for the 1949-75 p eriod ). F inally, the estim ates for the p e rio d 1949-73 are e v e n c lo s e r to th ose for the lo n g e r sam ple p e rio d s than they are w ith the earlier esti m ates. In the e a rlie r e s tim a tio n s, th ere are n o significant d iffe re n ce s in the co e ffic ie n t estim ates across p e rio d s, b u t r is 1.2 p e rce n t p er year and a is 58.9 p e rce n t in the 1949-73 sam p le p e rio d ; th ese are 1.6 p e rce n t and 64.9 p ercen t, re sp e ctiv e ly , in the earlier estim ation for the 1949-75 sam p le p e rio d . 4The ordinary least squares (OLS) estimates of the coefficients in table 3 are virtually identical but the Durbin-Watson statistics are 1.28, 1.35, and 1.37 for the 1949-73, 1949-75, and 1949-80 periods, respectively. To check whether this autocorrelated error pattern results from the omission of significant lagged input effects on output, one and two period lags on the input variables are added to the equations in table 3 and their OLS counterparts. When this is done, the coefficients are not significant, the Durbin-Watson statistic does not change and the estimate of p shown in table 3 is not reduced. An im portant h yp oth esis that was su p p orted in earlier w ork is re je cte d u sin g th e re v ise d data. A slo w in g in the tim e trend for te ch n o lo g ica l ch a n ge b e g in n in g in 1967 c o u ld not b e re je cte d earlier. F or all three sam ple p e rio d s in table 3, this h yp oth esis is rejected . A tim e-tren d variable w ith a va lu e o f zero to 1966, then in crea sed b y on e each year from 1966 on, was a d d ed to each eq u a tion estim ated in table 3. T h e t-statistics for the s lo w e r trend variable are FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 Table 3 Production Function Estimate for the U.S. Private Business Sector 1949-73 1949-75 1949-80 0o 1.3440 (3.78) 1.4663 (10.70) 1.4971 (13.25) 01 0.6885 (12.53) 0.7302 (13.25) 0.7201 (13.07) 02 0.3115 (5.67) 0.2698 (4.90) 0.2799 (5.08) 03 -0.07 04 (-1.00 ) -0.08 75 (-3 .9 6 ) -0.09 53 (-6 .2 2 ) 04 0.0168 (7.84) 0.0182 (9.69) 0.0177 (9.48) a 0.6432 (9.57) 0.6715 (14.30) 0.6574 (14.58) 0 0.2910 (5.27) 0.2481 (4.64) 0.2555 (5.31) y 0.0658 (1.07) 0.0805 (3.96) 0.0870 (6.81) t 0.0157 (6.11) 0.0167 (10.24) 0.0162 (10.36) R2 0.97 0.97 0.97 SE 0.0091 0.0099 0.0102 DW 2.03 1.93 1.93 P 0.44 0.39 0.39 - 1 .6 7 , - 1 .5 0 , and - 1 .4 1 , for the 1949-73, 1949-75, and 1949-80 p eriod s, resp ectiv ely . T h e slow in g is n ot statistically significant at a 5 p ercen t le v e l in any o f these p e rio d s .5 In ad d ition , a test for an optim al p o in t for a trend break u sin g a m in im u m standard error criterion fails to reveal a p oin t su perior to 1967. T h ere is n o e v id e n c e then for a s lo w d o w n in p r o d u c tivity grow th d u e to d is e m b o d ie d factors in flu en cin g the trend. T h e n ew estim ates d o n ot alter any o f the other earlier co n clu s io n s . In particular, the status o f a n u m b e r o f h yp oth eses tested earlier has b e e n u n ch a n g ed b e ca u se o f the ch anges in the private b u siness sector co n ce p ts and the n e w m easures. F or exa m p le, tests o f the C o b b -D o u g la s restriction y ie ld the rejection o f a translog sp ecifica tion o f the p rod u ction fun ction . ’ This result held before the recent revisions as well. See Rasche and Tatom, “ Energy Price Shocks, Aggregate Supply,” p. 25. T h e ou tput elasticity o f hours d u rin g the three p eriod s is n ot significan tly d ifferen t from the share o f la b o r in total co sts d u rin g e a ch o f th e th re e p eriod s. T h is is extrem ely im portant as the C o b b D ou glas p ro d u ctio n fu n ction im p lies a p rice elastic ity for e n e rg y d em a n d that m ay b e b ia se d upw ard. W h ile this w o u ld n ot y ie ld a bias in the estim ated effe ct o f e n erg y p rices on output, it w o u ld y ie ld an u p w a rd -b ia sed estim ate o f yand a d o w n w a rd -b ia se d estim ate of a. T h ere is no e v id e n c e o f such a bias. T h e t-statistics for the e q u a lity o f the a estim ate and the actual share o f labor in ea ch p e rio d are —0.22, 0.27, and —0.08, re sp e ctiv e ly , so that the h yp oth esis that a is equal to the actual share o f lab or can n ot b e rejected . O th er factors that failed to add significan tly to th e p r o d u c tiv ity re la tio n s h ip s e stim a te d e a rlie r c o n t i n u e to b e i n s i g n i f i c a n t . T h e s e i n c l u d e adjustm ents for p o llu tio n abatem en t capital and the ch a n gin g p rop ortion s o f y o u n g p e o p le (age 16-19) or w o m e n in the labor force. F in ally, it rem ains the case that p re-19 74 p ro d u c tion fu n ction estim ates that om it e n erg y d e v e lo p m ents break d o w n after 1973. W h en the 1949-73 m o d e l is estim ated w ith ou t the relative p rice o f en erg y, the standard error o f the equ ation is iden tical to that sh ow n in table 3. W h en the sam ple p e rio d is ex te n d e d to 1975 and 1980, the standard error o f the equation w ith ou t e n erg y rises to 1.24 p e rce n t and 1.37 p ercent, re sp e ctiv e ly . T h e C h o w test indicates that a significant ch a n ge in the structure o f th e p ro d u ction fu n ction occu rs in each case w h e n e n erg y is om itted and the sam ple p e r io d is len g th en ed . As the stability o f the standard errors in table 3 indicates, such structural ch a n ges can b e rejected u sing the C h o w test w h e n e n e rg y p rice s are in clu d e d . An estim ate o f the p ro d u ctio n fu n c tio n u sin g quarterly data from 11/1948 to III/19 8 1 is: (2) hi X, = 1.4688 + 0.7.351 In h, + 0.2649 hi kt (21.03) (23.81) (8.58) - 0.0893 ln (P e /P )t + 0.0045 t (-8 .3 1 ) (16.94) R2 = 0.99 SE = 0.0074 DW = 1.96 p = 0.76 a =0.6748 = 0.2432 (10.22) y = 0.0820 f = 0.0041 (18.41) (26.09) (9.05) T h e estim ated co e fficie n ts are essen tially the sam e as th ose in table 3. T h is quarterly p ro d u ctio n fu n c- 7 FEDERAL RESERVE BANK OF ST. LOUIS tion is u sed b e lo w to d erive the rev ised poten tial ou tp ut series. T h e stability and all oth er p rop erties d iscu s s e d a b ov e for the annual eq u a tion s in table 3 a p p ly to the quarterly estim ates as w ell. T h e im pact o f a ch a n ge in the relative p rice o f e n e rg y on output, p rod u ctivity, real w a ges, and the capital stock can b e a ssessed u sin g the p rod u ctio n fu n ction estim ate in eq u a tion 2. F o r a giv en e m p lo y m en t o f la b or hours and capital services (the shortrun e ffect), a 10 p e rc e n t rise in the relative p rice o f e n e rg y re d u ce s PBS ou tp u t (X t) and p rod u ctivity b y 0 .8 9 p e rc e n t. T h e lo n g -r u n e la s tic ity o f ou tp u t, labor p rod u ctivity, real w a ges, and the capital stock is ( ~ y lot), or 0.122 in this ca se.6 T h u s, a 10 p ercen t rise in the relative p rice o f e n erg y leads to a long-run d e clin e in output that is 36 p e rce n t larger than in the sh ort ru n . In p a rticu la r, a 10 p e r c e n t in c r e a s e re d u ces output, p rod u ctivity and the capital-labor ratio b y 1.22 p ercen t. From the third quarter o f 1973 to th e third quarter o f 1974, and, again from the first quarter o f 1979 to the s e co n d quarter o f 1980, the relative p rice o f en erg y rose 40 p e rce n t.7 G iv e n the estim ates a b ov e, ea ch sh ock re d u ce d p ro d u c tivity and poten tial ou tput b y 3.6 p e rce n t in the short run and 4.9 p e rce n t after adju stm en t o f the m arket for capital good s. REVISED MEASURES OF POTENTIAL OUTPUT T o d eterm in e p oten tia l real G N P , m easures o f p oten tial em p lo y m e n t o f lab or and capital are u sed to con stru ct potential private b u sin ess sector output. O th er com p on en ts o f real G N P that are n ot sensitive to c y c lic a l m o v e m e n ts in o u tp u t a n d are in d e p e n d e n t o f the em p loy m en t o f labor are then a d d ed to obtain potential G N P . T h e latter co m p on en ts are the ou tp u t o rig in a tin g in th e rest o f the w o r ld , g e n e ra l g o v e rn m e n t, h o u s e h o ld s and n o n -p ro fit institutions. 6See John A. Tatom, “ Energy Prices and Capital Formation: 197277,” this Review (May 1979), pp. 2-11, for an explanation and derivation of this result. ’ Note that percentage changes are measured by the change in the logarithm of the relative price of energy. The exact magnitudes over the two periods are 40.7 percent and 40.3 percent, which measured as actual percentage increases are 50.2 and 49.6 per cent, respectively. The relative price of energy rose another 12 percent in the first half of 1981 due to the immediate effects o f domestic crude oil decontrol, but subsequent adjustments in the world market due to decontrol took 2.8 percentage points off this in the third quarter o f 1981 alone. Digitized for8 FRASER JANUARY 1982 T h e d evia tion o f actual from p oten tial e m p lo y m e n t o f th e n a t io n ’ s ca p ita l s to ck is b a s e d on an ob se rv a tio n that at peak p e rio d s in th e past, the F ed era l R ese rv e B oard ca p acity u tilization rate m easure has b e e n abou t 87.5 p ercen t. T h is b e n c h mark is u sed in th e private b u sin ess sector p r o d u c tion fu n ction for full e m p lo y m e n t.8 T h e p oten tia l inpu t o f hours o f all p erson s e m p lo y e d in the private b u sin ess sector is fo u n d b y d eterm in in g p oten tial hours p e r w ork er and p o te n tial e m p lo y m e n t. In ea ch case, actual m easures are related to a m easure o f slack in the lab or market. This slack m easure (U N ) is the u n e m p lo y m e n t rate o f the civ ilia n lab or fo rce (U), m inus the fu ll-e m p lo y m e n t u n e m p lo y m e n t rate o f the civ ilia n labor force (U F ), w h ich was p rep ared in 1977 for the C o u n c il o f E c o n o m ic A dvisers (U N = U — U F ).9 H ours p er w ork er in the private b u sin ess sector are fou n d from the regression o f hours p er w ork er on excess u n e m p lo y m en t in the current and past quarter, a shift variable (D ) to a ccou n t for the un u su ally h igh le v e ls o f hours p e r w ork er from III/1961 to 11/1967, and a tim e trend (t) to a ccou n t for a secu lar d e c lin e in hours p er w ork er. F o r the p e r io d 11/1948 to III/1 9 8 1 , this eq u a tion is: (3) In HPW = 0.797 - 0.496 UN, + 0.177 UN,.i (546.1) (-6.06) (2.16) - 0.0011 + 0.014 D (-57.06) (7.29) R2 =0.99 Si: = 0.0032 DW = 1.89 p = 0.62 T h is eq u a tion has not b e e n ch a n g e d sin ce 1977, e x c e p t for the a d d ition o f the significan t la g g e d slack 8It can be argued that, at these peaks, “ normal” operating con ditions for the nation’ s plant and equipment are not observed and that, if demand were sustained, firms would increase investment to lower operating rates to optima] levels. In this case an 87.5 per cent rate for the FRB capacity utilization rate overstates the “ natural rate” o f capacity utilization. This argument has been made in John A. Tatom, “ The Meaning and Measurement of Potential Output: A Com ment on the P erloff and Wachter Results,” in Karl Brunner and Allan H. Meltzer, eds., Three Aspects o f Policymaking: Knowledge, Data and Institutions, Camegie-Rochester Conference on Public Policy, volume 10 (1979), pp. 165-78. The benchmark is supported by comparative movements in “ excess” unemployment o f the civilian labor force and the capacity utilization rate. When the capacity utiliza tion is regressed on the excess unemployment rate described in the text below over the period I/1955-II/1981, the constant is 86.2 percent with a standard error o f 0.78 percentage points when a significant lagged unem ployment rate is included. 9This data series and its developm ent is described by Peter K. Clark, “ Potential Output in the United States 1948-80,” U.S. Productive Capacity: Estimating the Utilization Gap (Washing ton University: Center for the Study o f American Business, Decem ber 1977), pp. 21-66. FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 term .10 T h e sum o fth e slack term s, - 0 .3 2 , is virtually iden tical to the sin gle con tem p ora n eou s term in the earlier estim ates, so that o n ly the tim in g o ft h e c y c li cal e ffe c t has b e e n ch a n ged . Potential hours per w orker is fou n d from the p re d ic te d valu es of eq u a tion 3 w ith the slack variable set at zero in the current and past quarter. P oten tia l e m p lo y m e n t in th e private b u sin e ss sector is fou n d in a sim ilar m anner. In particular, the logarithm o f private b u sin ess sector e m p lo y m e n t (In E M t) is regressed on a constant, a tim e trend (T), e x cess u n e m p lo y m e n t in the current and past quarter, and a trend shift variable (T 2) to a ccou n t for a shift in the trend rate o f grow th o f the lab or force after 1964. T h is particular break in trend was ch ose n on the basis o ft h e lo w e st standard error o ft h e eq u a tion. A break in trend is in c lu d e d to im prove the e fficie n cy o ft h e estim ation o ft h e coefficien ts for the sla ck v a r ia b le s . T h e e q u a tio n fo r th e 11/1948III/1981 p e rio d is: (4) In EM = 3.94 + 0.002 T + 0.004 T2 (371.63) (7.21) (9.59) - 0.013 UN, - 0.003 UNu (-1 1.03 ) (-2 .6 3 ) “R- = 0.91 SE = 0.0046 DW = 1.87 p = 0.92 W h en this equation is d ifferen ced , the autoregressive distu rban ces d isap p ear (the D u rbin -W atson statistic w ith o u t first-order a u tocorrela tion adju stm en t is 1.89), and the coefficien ts for the trend, break in trend, and slack variables are virtually iden tical. T o find potential e m p lo y m e n t in the private b u siness sector, the actual le v e l o f em p lo y m e n t is cy clic a lly adju sted b y (0.13 U N t + 0.003 UN,.]) p ercen t, a cco r d in g to the lev el and first-d ifferen ce eq u a tion s.11 10This equation was explained in Robert H. Rasche and lohn A. Tatom, “ Potential Output and Rs Growth Rate — the D om i nance o f Higher Energy Cost in the 1970’s,” U.S. Productive Capacity: Estimating the Utilization Gap (Washington Uni versity: Center for the Study o f American Business, Decem ber 1977), pp. 76-77. The unusual shift in hours per worker in the '60s has also been noted by George L. Perry, “ Potential Output and Productivity,” Brookings Papers on Econom ic Activity (1:1977), pp. 11-47. Tests o f additional lagged values ofthe excess unemployment rate found them to be insignificant. “ The effect of a one percent rise in the excess unemployment rate on PBS employment should be roughly a percent decline equal to the ratio o fth e civilian labor force to PBS unemployment. This may be derived from the relation that PBS employment is (1 — U) LF — NE, where U is the unemployment rate ofth e civilian labor force (LF), and NE is non-PBS employment, measured by the difference in civilian employment and PBS employment. The actual ratio ofth e labor force to PBS employ- T his m e th o d o f d e te rm in in g p oten tial e m p lo y m en t differs from the o n e this Bank u sed earlier. U ntil re c e n tly , p o te n tia l PBS e m p lo y m e n t was fou n d b y subtracting the le v e l o f current e m p lo y m en t ou tsid e the private b u sin ess sector from p o te n tial civ ilia n e m p lo y m e n t [(1 — U f ) tim es the civilian labor force]. T h e form er was equated to the d ifferen ce in actual civ ilia n e m p lo y m e n t and PBS e m p lo y m e n t. T h is m eth od had tw o m in or sh ortcom in gs. First, p e r io d ic cen su s revision s and ch a n ges in sam plin g and estim ation m eth od s alter the civilia n lab or force and e m p lo y m e n t data, sligh tly a ffectin g a m easure such as the a b o v e and an a cco m p a n y in g m easure of poten tial output. S e co n d , this e m p lo y m e n t m easure was som ew h at cy clic a l, d e sp ite the a b se n ce o f any perm an ent cy clic a l effects on the civilian labor force m easure. T h e reason for this appears to b e that PBS e m p lo y m e n t and civ ilia n e m p lo y m e n t data are esti m ated b y d ifferen t m eth od s, and their d iffe re n ce is c y c lic a l.12 B e v ise d quarterly estim ates of poten tial real G N P are p resen ted in the a p p en d ix to this article. T h e se data as w e ll as actual real G N P are sh ow n in chart 2. T h e grow th rate o f this re v is e d p oten tia l ou tput series has b e e n u n ch a n g e d for past p eriod s. T h e average grow th rate o f poten tial ou tp u t was 3.7 p e r ce n t from 1949 to 1973, the same as in the original estim ates. T his rate has varied som ew h at in the past, h o w e v e r, largely re fle ctin g d iffe re n ce s in the grow th rate o f the labor force. F or exam p le, from 1951 to 1963, p oten tial ou tput grew at a 3.3 p e rce n t rate w h ile the civ ilia n labor fo rce grew at on ly a 1.2 p e r ce n t rate. Potential ou tp ut grow th a ccele ra te d to a 4.0 p e rce n t rate from 1963 to 1973, as labor force grow th a ccele ra te d to a 2.1 p e rce n t rate. S in ce 1973, the p oten tial grow th rate has averaged 3.1 p e r c e n t d e s p ite a la b o r fo r c e gro w th o f 2.4 p ercen t. T h e potential ou tput grow th rate has b e e n as high as 5.0 p e rce n t in 1977-78. T h e annual grow th rate o f poten tial ou tput in 1974 and 1980 was o n ly 2.0 p ercen t; in 1975, this grow th rate w as o n ly 2.6 p ercen t. T h e se rela tively s lo w rates reflect the im pact ment in the sample period has a mean ol 1.26. The remainder is due to cyclical variation in non-PBS employment that does not affect non-PBS output. 12The difference between establisment-based payroll measures o f employment and households-sampling-based civilian employ ment measures is procyclical so that the old method resulted in a measure o f potential PBS em ploym ent that was inversely related to excess unemployment. This cyclical difference is discussed by Alexander Korns, “ Cyclical Fluctuations in the D ifference Between Payroll and Household Measures of Em ployment,” Survey o f Current Business (May 1979), pp. 14-44. 9 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 C h a rt 2 Real GNP B il li o n s of 1972 d o l l a r s Latest d a ta p lo tte d : 3 rd q u a rte r o f sharp in creases in the relative p rice o f en e rg y resou rces. T a b le 4 sh ow s the annual averages o f recen t le v e ls o f p oten tial G N P togeth er w ith recen t estim ates b y th e C o u n c il o f E c o n o m ic A d v isers (C E A ).13 T h e C E A estim ates range from 1.4 p e rce n t b e lo w to 1.0 p e rce n t a b ov e those p resen ted here. T h e p ercen tage d iffe r e n ce b e tw e e n the tw o m easures is virtually the same in 1973 as rep orted earlier, but the 1977 C E A estim ates for the p e rio d 1974-76 rose from 1.1 p ercen t larger to alm ost 3 p ercen t larger than this Bank’ s estim ates. T h e tw o estim ates are now extrem e ly clo s e , largely d u e to m ajor revision s in the C E A 13See Council of Econom ic Advisers, Economic Report o f the President, 1981. 10FRASER Digitized for B il li o n s of 1972 d o l l a r s Source: U.S. D e partm ent o f C om m erce estim ates re p o rte d in the Economic Report o f the President in 1977 and 1978. T h e se revision s p u s h e d the 1973 level b e lo w that estim ated b y this Bank, but then assu m ed a rou g h ly con stan t grow th rate that w as faster in 1974-76, then slo w e r in 1976-79, than that estim ated h ere. T h e C E A re p o rte d in 1981 that p oten tial ou tp ut was e x p e cte d to grow at a 2.9 p e r ce n t rate in 1979 and 1980, then return to a 3.0 p e r ce n t rate.14 14The CEA estimate is apparently based upon the expectation that the labor force will grow at a 1.75 percent rate, hours per worker w ill decline at a secular rate o f 0.5 percent and that potential productivity (output per hour) will rise at about a 1.75 percent rate. See Economic Report, 1980, pp. 89-90 and Economic Report, 1981, pp. 180-81. Such a rate o f productivity advance may appear optimistic in light o f the experience since 1978 or in 197.3-75. It should be noted, however, that potential FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 Table 4 Table 5 Recent Measures of Potential GNP (billions of 1972 dollars) Recent Productivity Developments (compound annual rates) 1973 Potential GNP CEA estimate Ratio $1,254.8 $1,234.9 98.3% 1974 1,279.7 1,277.5 99.8 1975 1,313.0 1,320.6 100.6 1976 1,351.8 1,365.1 101.0 1977 1,400.6 1,411.4 100.8 1978 1,470.0 1,459.3 99.3 1979 1,526.2 1,504.6 98.6 1980 1,556.2 1,548.5 99.5 T h e prim ary d iffe r e n ce sh ow n in table 4 is that the C E A grow th rate o f poten tial ou tput o f 3.4 p erce n t in 1973-76 e x ce e d s the 2.5 p e rce n t rate estim ated h ere, w h ile its grow th rate for 1976-79 o f 3.3 p erce n t is less than the 4.1 p e rce n t rate estim ated here. It appears that, in recen t years, the C E A has sm ooth ed its p oten tial ou tp ut series to capture the sharp su p p ly shock effects on poten tial output b y lo w e rin g the grow th rate o f potential ou tput over several years. As a resu lt, the le v e ls o f p oten tia l ou tp u t h ave not d iffe re d substantially. This d iffe re n ce is to an extent intentional, as the C E A has always e m p lo y e d a given grow th rate for lo n g p eriod s. T h is te n d e n c y has b e e n te m p e red in recen t years, as can b e seen b y the slight variability in the C E A annual grow th rate sh ow n in table 4. It m ay b e that the 1980-81 p rod u ctiv ity losses that re s u lt from e n e r g y s h o ck s w ill b e la r g e ly re flected in the C E A ’ s use o f too slow a rate o f p o te n tial grow th for the early 1980s. RECENT ACTUAL AN D POTENTIAL PRODUCTIVITY DEVELOPM EN TS T h e sharp drops in poten tial output grow th in 1974-75 and 1980 reflect the e ffe c t o f m ajor en e rg y p rice ch an ges on actual and p oten tial prod u ctivity. output per hour, discussed in the next section, rose over fiveyear periods at no less than a 2.5 percent rate from 1948-73. Following the implementation o f accelerated depreciation and corporate tax cuts, the pace o f capital formation rose sharply so that it surged to the post-World War II peak rate of 3.2 percent from 196.3 to mid-1970. Even during 1978 potential productivity growth had risen to over a 2 percent rate as the adjustment to the prior energy shock was apparently approaching completion. A repeat of that pattern and recent supply-side policies suggest a more rapid pace o f productivity growth from 1982-85 than that projected by the CEA. IV/1948IV/1973 IV/1973IV/1980 Difference Growth of output/hour 2.87% 0.67% 2.20 Potential growth rate 2.82 0.94 1.88 Cyclical factors 0.09 -0 .2 0 0.29 -0 .0 4 -0 .0 7 0.03 Residual factors Contribution to potential growth rate of: Capital accumulation (Growth in high-employment capital-labor ratio) 0.92 (3.52) Energy price changes 0.07 0.28 (1.04) -1 .1 3 0.64 (2.48) 1.20 S in ce 1973, p ro d u ctiv ity ’ s abysm al p erform a n ce has b e e n a m ajor co n ce rn for p olicym ak ers. Th u s, it is u sefu l to detail the factors in flu e n cin g such grow th o v e r the last seven years.15 A n analysis o f th e actual and p oten tial p rod u ctivity d e c lin e for the private b u sin ess sector appears in table 5, w h ere grow th rates and the con trib u tion o f various factors are co m p a re d for tw o p eriod s: 1948 to the en d of 1973, and 1973 to the e n d o f 1980. O utput p er h ou r grew at a 2 .87 p e rce n t rate from IV /1948 to IV /1973, th en s lo w e d to a 0 .67 p e rce n t rate o v e r the next seven years. This grow th can b e an a ly zed in tw o w ays. T h e first is to lo o k at the con trib u tion o f the factors en terin g eq u a tion 2: the actual ch a n ges in the grow th o f e m p lo y e d capital relative to labor, the relative p rice o f en erg y, the p a ce o f te ch n o lo g ica l ch a n ge, and residuals d u e to random errors o f fitting the equ ation at the e n d points o f the p e rio d . T h e se co n d , sh ow n in the top p an el o f table 5, is to break d o w n actual p rod u ctivity grow th in ea ch p e rio d into ch an ges d u e to the grow th o f p oten tia l p rod u ctivity, ch a n ges d u e to cy clic a l variations in the e m p lo y m en t o f capital and lab or at the b e g in n in g and e n d p e rio d s, and d iffe re n ce s in the residu al or random error co m p o n e n t o f equation 2. 15In contrast to Edward F. Denison, “ Explanations o f Declining Productivity Growth,” Survey o f Current Business, (August 1979, part 2), pp. 1-24, the analysis here o f post-1973 productiv ity developments fully explains the productivity “ puzzle,” while other explanations do not. See Tatom, “ The Productivity Problem” or especially D enison’s paper for a discussion o f these other factors. The puzzle is presumably all the more challenging to other analysts due to the post-1978 cessation o f productivity growth. 11 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 C h a rt 3 C ap ital Labor Ratio (Private Business Sector) 1972 d o l l a r s per w o r k e r 1972 d o l l a r s per w o r k e r Sources: U.S. D epartm ent o f Lab or, a n d B o ard o f G overnors o f the Federal Reserve System The p o te n tia l r a tio is th e c a p ita l sto c k a d ju s te d fo r an 8 7 .5 % c a p a c ity u t iliz a t io n r a te d iv id e d b y p o te n tia l h o u rs o f e m p lo y m e n t in th e p r iv a te b u sin e ss s e cto r. The a c tu a l r a tio uses th e a c tu a l c a p a c ity u t iliz a t io n ra te a n d h o u rs o f e m p lo y m e n t in th e p r iv a te bu s in e s s s e c to r. L a te s t d a t a p lo tte d : 3 rd q u a rte r T h e con trib u tion o f cy clic a l factors accou n ts for the d iffe re n ce b e tw e e n the p rod u ctiv ity effe ct o f the grow th o f the p oten tial PBS ca p ita l-lab or ratio (the con trib u tion o f the cap ital-labor ratio to poten tial grow th) and the e ffe c t o f the actual grow th in the utilization o f capital p er hour. T h e d iscrep a n cy b e tw e e n the tw o arises from the cy clic a l variability o f the ca p ital-labor ratio sh ow n in chart 3. T h e sum o f the “ cy clic a l factor” and the con trib u tion o f “ capital a ccu m u la tion ” to poten tial p rod u ctiv ity grow th in d i cates the estim ate o f the actual im pact o f m ovem en ts in the o b se r v e d ratio o f u tilized capital to la b or hours on the o b s e r v e d p rod u ctivity grow th. 12FRASER Digitized for M ost o f the 2.2 p e rce n ta g e -p o in t d e c lin e in p ro d u ctivity grow th o v e r the last se v e n years has b e e n d u e to factors that s lo w e d p oten tia l p ro d u ctiv ity grow th. F or the particular com p a rison sh ow n , c y c li cal d iffe re n ce s b e tw e e n p ro d u ctiv ity m ovem en ts in the tw o p e rio d s o r residu al errors a cco u n t for o n ly 0.3 p ercen tage points o f the o b s e r v e d slow in g. In the lo w e r part o f the table, the factors co n trib u tin g to the p oten tial p ro d u ctiv ity grow th s lo w d o w n are sh ow n . W hat is om itted in the lo w e r part o f the table is th e trend grow th o f total factor p rod u ctivity w h ich con trib u ted 1.82 p ercen ta ge poin ts to the rate FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 C h a rt 4 Potential and A ctual Output per Hour (Private Business Sector) 1972 d o l l a r s L a te st d a t a p lo tte d : 3 rd q u a rte r of p rod u ctivity grow th in both p eriod s. T h e d irect e ffe c t o f en erg y p rice shocks o v e r the last seven years has b e e n to reverse the slight p ositive con tri b u tion o f en e rg y p rice d e clin e s o v e r the p rior 25 years, so that 1.20 p ercen ta g e p oin ts o f th e 1.88 p e rcen ta g e-p oin t-p er-y ea r d e clin e in p oten tial p ro d u ctivity grow th has b e e n d u e to this factor. T h e re m a in d er has b e e n d u e to a s lo w in g in capital form ation. As n oted in parentheses, the grow th rate o f the capital stock relative to p oten tia l hours o f e m p lo y m en t was 3.52 p e rce n t o v e r the 25 years e n d in g in 1973; su b seq u en tly , this grow th slo w e d to about o n e p ercen t. This s lo w in g re d u ce d the con trib u tion o f capital form ation from a 0.92 p erce n t rate to a 0.28 p e rce n t rate o v e r the last seven years. W h en this 1972 d o l l a r s S o u rc e : U .S . D e p a rtm e n t o f L a b o r result is c o m b in e d w ith the e ffe ct on p rod u ctivity grow th o f cy clic a l m ov em en ts in the capital-labor ratio, th e resu lt is that cap ital form a tion , w h ich a d d e d 1.01 p ercen tage p oin ts (0.92 + 0.09) to the actual p a ce o f p rod u ctivity grow th from the e n d o f 1948 to the e n d of 1973, o n ly co n trib u te d 0.08 p e r ce n ta g e p oin ts to the actual rate o f p ro d u ctiv ity grow th from the e n d o f 1973 to the e n d o f 1980. Im p lic itly , c y c lic a l d iffe r e n ce s b e tw e e n th e e n d o f 1973 and 1980 o ffs e t the e ffe c t o f grow th in capital p er h our, so there was virtually n o ch a n ge in the actual e m p lo y m e n t ratio. T h e sm all ch an ges in table 4 b e c o m e quite large w h e n c o m p o u n d e d o v e r the seven -y ear p e rio d . F or exam p le, the slo w in g in the p oten tial grow th rate o v e r the seven -y ea r p e rio d re d u ce s private sector 13 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 C h a rt 5 Real GN P per C ivilia n W orker 1 948 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 L a te st d a ta p lo tte d : 3 rd q u a rte r ou tp ut b y 13.8 p erce n t b y the en d o f 1980. T h e d irect im pact o f en erg y p rice increases alon e o v e r these seven years red u ces ou tp ut b y 8.3 p ercent. M o reov er, a large share o f the red u ction in capital form ation sin ce 1973 has b e e n d u e to en erg y p rice d e v e lo p m e n ts . T h e in clu sion o f the en erg y p ricein d u ce d s lo w in g in the d e sire d ca p ita l-lab or ratio leads to an 11.5 p e rc e n t loss in output. T h e rem ain in g loss in p oten tial p rod u ctiv ity is a ssociated w ith a n on -en ergy-rela ted s lo w in g in capital form ation .16 16Factors responsible for the cessation o f growth in the utilized capital-labor ratio besides the decline in the productivity o f capital due to energy price changes and minor cyclical influ ences, include such factors as higher expected inflation, inflation uncertainty, and riskier returns due to an increased probability o f governmental intervention through regulatory initiatives. See Tatom, “ The Productivity Problem.” Patric H. Hender- Digitized for 14FRASER 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 8 0 1981 Sources-. U.S. D e p a rtm e n t o f L a b o r a n d U.S. D e p a rtm e n t o f C o m m e rc e Fin ally, it sh ou ld b e n oted that table 5 presents a sum m ary v ie w o f the e ffe c t o f e n e rg y p rice ch an ges on actual and p oten tia l p ro d u ctiv ity that d oes not reflect the actual pattern o f even ts. In particular, the “ s lo w d o w n ” d e s c r ib e d in table 5 is n ot con tin u ou s. A ssocia ted w ith each e n erg y sh ock is a on ce -a n d -fo rall d e c lin e in b o th m easures o f p rod u ctivity, w ith shott, “ The D ecline in Aggregate Share Values, Taxation, Valuation Errors, Risk, and Profitability,” American Economic Review (D ecem ber f9 8 f), pp. 909-22, discusses these and other factors that contribute to the slowdown in capital formation and argues that inflation alone has had little impact on the decline in share values and, implicitly, capital formation. Instead he claims that a change in risk premiums (attributed to increased uncertainty about price and regulatory changes) in equity and bond yields and reduced pretax profitability have been the reason for about half the decline in share values. Hendershott does not assess the role o f a higher relative price o f energy in reducing the pretax real profitability o f the corporate capital stock. FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 tem porarily slo w e r grow th as the cap ital-labor ratio is adju sted tow ard a lo w e r d esired lev el. T h is pattern is m ore clea rly apparent in chart 4, w h ich show s poten tial and actual m easures o f private b u siness sector ou tput p er hour. D ev ia tion s in the tw o are p red om in a n tly d u e to the b u sin ess c y c le . T h e pat tern o f poten tial p rod u ctivity d ev e lo p m e n ts in chart 4 show s virtually n o grow th from m id-19 7 3 to m id1975 and relatively slow exp an sion from m id -19 7 5 to m id-1977. F o llo w in g the s e co n d e n e rg y shock, potential p rod u ctivity fell, then was virtually un ch a n g ed until the e n d o f 1980. T h e s e co n d phase o f a r e la t iv e ly s lo w p a c e o f p o te n tia l p r o d u c tivity expan sion is apparent in the first three quarters o f 1981. Chart 5 sh ow s an alternative m easure of p ro d u ctiv ity, real G N P p er civ ilia n w orker, again m easu red on b oth an actual and a h ig h -em p loy m en t b asis.17 T h e prim ary d iffe re n ce from chart 4 is the secular rate o f d e c lin e in hours p er w orker. Both actual and poten tial real G N P p er w ork er have flattened out tw ice relative to the p rior trend grow th , w ith a resu m p tion of grow th from early 1977 until early 1979. At the en d o f 1980, poten tial real G N P per 17The high-employment measure of civilian employment is found by regressing changes in the logarithm o f the civilian labor force on a constant, a shift for faster labor force growth after 1964 and current and one-lagged changes in the excess unemployment rate. Additional lags are not statistically significant. Moreover, the constraint that the effect of slack is zero after two quarters could not be rejected. The effect o f a one percentage-point in crease in slack is to increase the labor force by 0.2 percent (t = 2.34) in the current quarter and this is offset in the subsequent quarter. To find the high-employment civilian employment, these cyclical effects are added back to the observed civilian labor force and high-employment unemployment (UF) (LF) is removed. w orker stood o n ly 5.7 p e rce n t h igh er than at the e n d o f l9 7 3 , so that five years w orth o ft h e prior trend grow th (2.2 p e rce n t rate) has b e e n lost du rin g the past seven -y ear p eriod . SUMMARY R e ce n t revision s in the m easures o f the nation’ s ou tput and capital stock, as w e ll as m in or ch an ges in p ro ce d u re s, have altered this Bank’ s m easures of p oten tial output. T h e m ajor co n clu s io n s o f earlier Bank studies, h o w e v e r, have b e e n u n affected b y th ese ch anges. In particular, the grow th o f potential ou tput has b e e n sharply re d u ce d b y the 1973-74 and 1979-80 e n erg y shocks and s u b se q u e n t adjustm ents in the d e sire d capital intensity o f p rod u ction . T h e se effects have b e e n con firm ed b y the re-estim ation of ea rlier p ro d u ctio n fu n ction co e fficie n ts, and, m ore im portant, the confirm ation o f the p rior em p irical e s tim a te s in th e la te s t r o u n d o f e n e r g y p r ic e increases. T h e d e c lin e in the grow th o f potential ou tput sin ce 1973 has, in recen t years, b e e n a ck n o w le d g e d b y the C o u n c il o f E c o n o m ic A dvisers, b u t through a trend r e d u c t io n rather than th rou g h sharp te m p o ra ry d e clin e s in 1974-75 and 1979-80 as im p lie d here. N on eth eless, the lev e l o f poten tial ou tp ut estim ated b y the C E A in recen t years is little d ifferen t from this Bank’ s estim ate. T h e slo w in g in p oten tia l ou tput masks a sharper red u ction in the grow th o f p ro d u c tivity in re ce n t years. A d eta iled analysis o f p ro d u c tivity d e v e lo p m e n ts show s a m arked deterioration in grow th relative to past trends. In the m easu rem en t o f potential output, this d eterioration has b e e n par tially o ffse t b y a m ore rapid grow th o f b oth poten tial and actual em p loy m en t. (See appendix on next page) 15 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 Appendix Potential GNP (in billions of dollars)1 I II IV 1948 NA $ 489.1 $ 502.9 $ 506.1 1949 $ 515.0 523.6 530.1 532.0 1950 536.1 540.7 546.9 555.7 1951 566.3 576.6 582.9 586.7 609.3 1952 592.3 591.4 599.3 1953 612.8 615.6 617.5 621.6 1954 630.1 634.3 641.0 643.7 1955 648.6 653.5 662.9 672.4 1956 672.3 679.7 683.5 687.8 1957 691.2 698.0 707.7 708.5 1958 710.1 720.8 730.4 731.3 1959 732.0 739.8 746.3 756.0 1960 757.7 763.6 768.4 774.5 1961 782.8 789.2 797.7 806.1 1962 813.7 824.7 828.8 828.9 1963 836.9 843.5 853.7 862.7 1964 870.9 877.1 881.2 884.7 1965 895.9 908.5 918.6 930.7 1966 942.8 955.6 963.2 973.1 1967 981.4 985.3 997.9 1,010.2 1968 1,017.1 1,033.1 1,041.4 1,051.3 1969 1,069.3 1,081.3 1,092.0 1,097.3 1970 1,109.9 1,117.8 1,128.6 1,136.6 1971 1,150.2 1,160.8 1,166.4 1,176.3 1972 1,188.9 1,203.1 1,216.8 1,226.2 1973 1,236.9 1,247.6 1,262.3 1,272.4 1974 1,273.5 1,275.0 1,280.1 1,290.3 1975 1,300.3 1,307.3 1,316.8 1,327.5 1976 1,335.3 1,345.7 1,355.6 1,370.5 1977 1,384.8 1,394.5 1,405.5 1,417.7 1978 1,437.8 1,460.7 1,479.5 1,502.2 1979 1,521.7 1,520.9 1,529.3 1,533.1 1980 1,545.6 1,544.4 1,560.9 1,573.9 1981 1,584.0 1,580.4 1,595.0 'Prepared using data available through November 30, 1981. 16FRASER Digitized for III JANUARY 1982 FEDERAL RESERVE BANK OF ST. LOUIS The Role of Fiscal Policy in the St. Louis Equation R. W. HAFER T H E “ St. L ou is eq u a tio n ” relates the grow th of n om inal in com e (G N P ) to b oth the grow th o f m o n e y and h ig h -em p loy m e n t gov ern m en t exp en ditu res. In o th e r w o r d s , it attem p ts to e x p la in ch a n g e s in G N P b y ch a n ges in m on etary and fiscal actions. O n e con sisten t result o f estim ating the St. L ou is eq u a tion for the U.S. e c o n o m y is that m on eta ry action s have a stron g and lastin g im p a ct o n the grow th of G N P , w h ile fiscal actions have on ly a w eak and transitory im pact; fiscal effects essen tially wash out o v e r four or five quarters. T h e p u rp ose o f this article is to reassess the role o f fiscal actions within the framework o f the St. L ou is e q u a tion .1 As a result o f u pd a ted estim ates and tests, strong e v id e n c e is p resen ted reaffirm ing that fiscal actions are in con seq u en tia l in d eterm in ing G N P grow th, o n ce the effects o f m o n e y grow th are taken into a ccou n t. IN-SAMPLE ESTIMATES T h e St. L ou is eq u a tion is the fou n d ation for the sm all-scale m onetarist m o d e l associated with the F ed era l R eserve Bank o f St. L ou is. T h e nature o f the eq u a tion (as w e ll as the m o d e l) is m onetarist b eca u se G N P grow th is d e term in ed prim arily b y the grow th o f the n om inal m o n e y stock. A lth ou gh the eq u a tion re co g n izes and e m p irica lly captures the short-run e ffe cts o f stim ulative fiscal a ction s, p re v io u s re1Alternative specifications of the St. Louis equation provide other frameworks in which the role o f fiscal actions could be tested. Our purpose, however, is not to test for the impact of fiscal actions across a broad class o f models, but to focus attention on the widely recognized St. Louis specification. An alternative specifi cation may be found in John A. Tatom, “ Energy Prices and ShortRun Econom ic Performance,” this Review (January 1981), pp. 3-17. search has dem on strated that fiscal actions have no lastin'' e ffe c t o n G N P g ro w th in th e St. L o u is s p e c ific a tio n .2 N ot su rp risin g ly , this fin d in g has b e e n the so u rce o f a co n tin u in g d e b a te .3 T h e St. L ou is e q u a tio n ty p ica lly is w ritten as 4 (1) Y(t) = a« + 1 in, M (t-i) + i= 0 4 i e, E (t-i) + e(t), i= 0 w h e re Y is n om inal G N P , M is the m o n e y su p p ly and E is the m easure o f fiscal p o licy . T h e dots a bove the 2See, for example, Leonall C. Andersen and Jerry L. Jordan, “ Monetary and Fiscal Actions: A Test o f Their Relative Impor tance in Econom ic Stabilization,” thisReview (November 1968), pp. 11-24; Leonall C. Andersen and Keith M. Carlson, “ A Mone tarist M odel for Econom ic Stabilization,” this Review (April 1970), pp. 7-25; Keith M. Carlson, “ Does the St. Louis Equation Now Believe in Fiscal Policy?” this Review (February 1978), pp. 13-19; Keith M. Carlson, “ Money, Inflation, and Econom ic Growth: Some Updated Reduced Form Results and Their Impli cations,” this Review (April 1980), pp. 13-19; and Keith M. Carlson and Scott E. Hein, “ Monetary Aggregates as Monetary Indicators,” this Review (November 1980), pp. 12-21. 3Earlier works include Richard G. Davis, “ How Much Does Money Matter? A Look at Some Recent Evidence,” Federal Reserve Bank o f New York Monthly Review (June 1969), pp. 119-31; E. Gerald Corrigan, “ The Measurement and Importance o f Fiscal Policy Changes,” Federal Reserve Bank o f New York Monthly Review (June 1970), pp. 133-45; and Edward M. Gramlich, “ The Useful ness of Monetary and Fiscal Policy as Discretionary Stabilization Tools,” Journal o f M oney, Credit and Ranking (May 1971), pp. 506-32. More recent analyses include Alan S. Blinder and Stephen M. Goldfeld, “ New Measures o f Fiscal and Monetary Policy, 1958-73,” American Economic Review (D ecem ber 1976), pp. 780-96; Benjamin M. Friedman, “ Even the St. Louis M odel Now Believes in Fiscal Policy,” journal o f M oney, Credit and Ranking (May 1977), pp. 365-67; Yash P. Mehra, “ An Empirical Note on Some Monetarist Propositions,” Southern Economic Journal (July 1978), pp. 154-67; and Yash P. Mehra and David E. Spencer, “ The St. Louis Equation and Reverse Causation,” Southern Economic Journal (April 1979), pp. 1,104-20. 17 FEDERAL RESERVE BANK OF ST. LOUIS variable s y m b o liz e ra tes-of-ch ange, ao, nii and e; are param eters to b e estim ated, a nd e(t) is a random error term .4 T h rou gh the years, various em p irical m easures o f m on etary and fiscal actions have b e e n u sed to esti mate the St. L ou is equ ation. R e cen t red efin ition s o f the m on eta ry aggregates rek in d led the d eba te abou t w h ich aggregate b est explains G N P grow th . Since recen t e v id e n c e suggests that M 1B is the p referred aggregate, it is u sed in this article.5 T h e con sen su s c h o ic e o f the fiscal m easure has n ot yet b e e n resolved . T h e h ig h -em p loy m en t fe d eral exp en ditu re m easure o f fiscal p o lic y is clo s e ly associated w ith the textbook in com e-ex p en d itu re m o d e l o f the e c o n o m y .6 In that m o d e l, gov ern m en t purch ases of g ood s and services plus transfers are c o n s id e r e d a d irect stim ulus to nom inal aggregate d em an d . C on seq u en tly , an increase in this m easure represents an increase in the g ov ern m en t’ s in flu en ce o n n om inal d em a n d and, therefore, nom inal in co m e . A n alternative, the h ig h -em p loy m en t federal surplus variable, captures the net in flu en ce o f exp en ditu res and receip ts d u e to the ch a n gin g patterns in federal g ov ern m en t p urch ases and transfers, and in tax rates. A lth ou gh oth er less co m m o n ly u sed m easures exist, th ese tw o are u sed in this article to assess the in flu en ce o f fiscal actions on the grow th of G N P. E q u a tion 1 was estim ated u sing ordinary least squares for th e p e r io d I/1 960-IV /1980; the regres sion results are rep orted in table 1. E qu ation s A and B sh ow the results ob tain ed u sin g the grow th o f t h e h ig h -e m p lo y m e n t fe d e r a l e x p e n d itu r e s and the ch a n g e in le v e l o f the h ig h -e m p lo y m e n t fed era l su rplus, r e s p e c tiv e ly .7 T h e results co n fo rm w ith 4The rate-of-change specification was introduced by Carlson, who demonstrated that the original first-difference specification ot Andersen and Jordan introduced econometric difficulties when estimated into the mid-1970s. See Carlson, “ Does the St. Louis Equation Now Believe in Fiscal Policy?” 5For evidence o fth e superiority o f M1B, see Carlson and Hein, “ Monetary Aggregates as Monetary Indicators;” R. W. Hafer, “ Selecting a Monetary Indicator: A Test o fth e New Monetary Aggregates,” this Review (February 1981), pp. 12-18; and R. W. Hafer, “ Much Ado About M2,” this Review (October 1981), pp. 1.3-18. G“ High-employment” measures are used to reduce the distortion introduced from the impact of the level o f econom ic activity on actual government receipts and expenditures. See Keith M. Carlson, “ Estimates o f the H igh-E m ploym ent Budget and Changes in Potential Output,” this Review (August 1977), pp. 16-22. 7The changes in the high-employment federal surplus variable scaled by income also was used as an alternative measure of Digitized for 18FRASER JANUARY 1982 p re v io u sly rep orted findings: T h e su m m ed im pact o f m o n e y grow th is not statistically d iffe re n t from unity at the 1 p e rce n t le v e l, and the cu m u lative im pact o f fiscal actions is n ot statistically d ifferen t from zero. T h e grow th o f M 1B has a significant, sus tain ed in flu en ce on the grow th o f G N P ; fiscal actions, m ea su red eith er as the grow th o f h ig h -e m p lo y m e n t fe d e ra l ex p e n d itu re s or th e ch a n g e in the h ig h e m p lo y m e n t federal surplus, d o not. T h e cu m u lative im pa ct o f the m onetary and fiscal variables is statistically co m p a ra b le w ith p reviou s results. In d e e d , th e lag pattern for the m o n e y grow th variable is essen tially the sam e as C arlson rep orted for a variety o f sam p le p e r io d s .8 As in p re v io u s studies, the im pact of m o n e y grow th on G N P grow th is quite rapid: T h e rep ercu ssion s o f a ch a n ge in the grow th o f m o n e y o c c u r w ith in tw o quarters and fade away b y the third and fourth la g g e d quarters. T h e lag patterns for the fiscal variables, in contrast, d o not com p a re as w e ll to p re v io u sly rep orted fin d ings. W h e n c o m p a re d w ith th e results for oth er sam ple p e rio d s, the m a gn itu d e o f t h e in d ivid u al lag co e fficie n ts and their sign ifican ce ch an ge dramati ca lly for the fiscal variables. In d e e d , it appears that, in a d d ition to h avin g no lasting im pact on G N P g ro w th , th e fisca l v a ria b les exert n o sig n ifica n t e ffe c t in any quarter. A lthough e co n o m e tric d ifficu l ties p re c lu d e a firm co n c lu s io n b a se d o n the signifi ca n ce o f in d ivid u al lag estim ates, the sen sitivity o f e stim a te s o f th e fisca l m e a su re s o v e r d iffe r e n t sam ple p e rio d s suggests that a G N P -fiscal actions link is d u b io u s. T o further exam ine this issu e, three alternative eq u a tion s w e re estim ated. T h e s e e q u a tions isolate the relative explanatory p ow ers o f t h e m o n e y and fiscal variables on G N P grow th. E sti m ates from th ese specification s also are rep orted in table 1. E qu a tion C reports the estim ation o f eq u a tion 1 u sin g o n ly the m o n e y grow th variable to explain G N P grow th. N ot surprisingly, the overall explan a tory p o w e r o f the eq u a tion is not d im in ish e d su b stantially b y the e x clu s io n o f eith er fiscal p o licy m easure. M o re o v e r, the cu m u lative im pact o f m o n e y grow th on G N P grow th is n ot statistically differen t from unity o v e r the cou rse o f o n e year. T h is suggests that the m o n e y -G N P link in the St. L ou is equ ation is q uite robust. T h e sam e can n ot b e said, h o w e v e r , o f fiscal actions. The conclusions reached in the article were not altered by this change. 8See Carlson, “ Does the St. Louis Equation Now Believe in Fiscal Policy?” and Carlson, “ Money, Inflation, and Economic Growth.” JANUARY 1982 FEDERAL RESERVE BANK OF ST. LOUIS Table 1 Regression Results for Alternative St. Louis Equations, 1/1960-1V/1980 Equations Tested1 Coefficients A B Constant mo mi m2 m3 rri4 2mj 2.46 (2.17) 0.401 (3.40) 0.393 (5.07) 0.225 (2.21) 0.062 (0.85) -0.013(0.11) 1.067 (5.53) 2.69 0.359 0.349 0.200 0.061 -0.001 0.969 eo ei e2 e3 e4 -e. 0.062 (1.46) 0.020 (0.62) -0.021 (0.58) -0.018(0.54) 0.014 (0.33) 0.056 (0.59) -0.006(0.13) -0.031 (0.84) -0.05 8(1 .4 7) -0.072(1.94) 0.057 (1.33) -0.22 5(1 .6 3) R2 SE DW 0.349 3.493 2.02 (2.69) (2.82) (3.92) (1.87) (0.74) (0.00) (3.50) 0.361 3.460 2.07 C D 2.80 (2.84) 0.422 (3.73) 0.397 (5.28) 0.218(2.16) 0.065 (0.90) 0.007(0.06) 1.109(6.30) 6.02 (5.42) 0.086(1.70) 0.047(1.22) 0.016(0.38) 0.039(1.02) 0.076(1.54) 0.265 (2.50) 0.355 3.477 2.02 0.055 4.209 1.52 E 8.52 (18.13) -0.031 -0 .0 0 6 -0 .0 5 9 -0.09 0 -0 .0 7 3 -0 .1 9 6 (0.61) (0.13) (1.23) (2.00) (1.39) (1.18) 0.017 4.291 1.47 1All equations estimated using a fourth-degree Almon polynominal lag with both end points constrained to zero. Equations A and D use high-employment federal expenditures; equations B and E use high-employment federal surplus. M1B is the money measure used throughout. Absolute values of t-statistics appear in parentheses; R2 is the coefficient of determination corrected for degrees of freedom, SE is the regression standard error and DW is the DurbinWatson statistic. the em p irical relation sh ip b e tw e e n the fiscal actions and G N P. E qu ation s D and E in table 1 report the results of re g re s s in g G N P g row th on th e grow th o f h ig h em p lo y m e n t exp en ditu res and changes in the lev el o f h ig h -em p loy m en t surplus, resp ectiv ely . In e q u a tion D , the estim ated co e fficie n ts are n o tice a b ly d ifferen t from th ose in eq u a tion A. T h e first lag co e ffic ie n t is m ore than d o u b le the estim ate from eq u a tion A, and th e s e co n d and third lag terms are p o s i t i v e . M o r e im p o r ta n t, th e c u m u la t iv e im pact for the exp en d itu res variable is positive and statistically significan t at the 5 p ercen t lev el. U nfor tunately, th ese results are m arred b y the existen ce o f significant first-order serial correlation a m on g the residu als, e v id e n c e d b y the lo w D u rb in -W a tson (D W ) statistic. T h e p re s e n ce o f significant serial correlation in eq u a tion D p rov id es im portant inform ation . O n e potential cau se o f serial correlation is the om ission o f an im portant explanatory variable w h ich is c y c li cal. T h e c o n s e q u e n c e o f such an om ission is that the error term w ill absorb the cy clic a l pattern o f the variable, and the su cce ssiv e error terms w ill n ot b e random . It is clea r from a com pa rison o f equ ation s A and D that the e x c lu d e d variable is the grow th o f M 1B : A d d in g m o n e y grow th to eq u a tion D e lim i nates first-order serial correlation . I f o n e assum es that equ ation A is the “ co rre ct” sp ecifica tion , the m o d e l re p re se n te d b y eq u a tio n D (b e ca u se it is m issp eeified ) yie ld s co e ffic ie n t estim ates that may b e seriou sly b ia se d and sig n ifica n ce tests o f q u e s tion ab le efficacy. Th u s, the e v id e n c e p ro v id e d b y e q u a tion D d o e s n ot su p p ort the e x is te n ce o f a significant, lasting affect on G N P grow th o f fiscal actions cap tu red b y the grow th of h ig h -e m p lo y m e n t exp en ditu res. E q u a tio n E in ta ble 1 p resen ts th e results o f re g ressin g G N P grow th on ch a n ges in the high e m p lo y m e n t federal surplus. In lin e w ith a priori reason in g, the results suggest that an in crease in the size o f the h ig h -e m p lo y m e n t surplus retards G N P grow th. A lth ou gh the su m m ed im pact o f the surplus variable is o f the appropriate sign, it is not statis 19 JANUARY 1982 FEDERAL RESERVE BANK OF ST. LOUIS tica lly d ifferen t from zero. T h is variable d oes not exert a significant, lasting effe ct on G N P grow th, a fin d in g con sisten t w ith eq u a tion B. T h e p re s e n ce o f serial correlation , h o w e v e r, again suggests the p o s sib ility o f a m issp ecifica tion . This is e v id e n c e d by com p a rin g equation s D and B: T h e add ition o f the m o n e y grow th variables rem oves the serial correla tion p rob lem . T his result indicates that eq u a tion E, like eq ua tion D , is m iss p e cifie d w ith ou t the m o n e y va riab le.9 A n y u n ce rta in ty a b o u t th e re la tiv e im p a ct o f m onetary and fiscal actions on G N P grow th can b e d is p e lle d further b y a com p a rison o f equ ation s A, B and C in ta ble 1. T h is co m p a riso n a llow s us to address the q u estion “ G iv en the in flu en ce o f m o n e tary actions on G N P grow th, d oes the inform ation con ta in ed in the fiscal variables significantly im p ro v e u p on m o n e y ’ s explanatory p o w e r ? ” T o answ er this qu estion , o n e n e e d o n ly exam ine th e re p o r te d t-statistic for th e re le v a n t su m m ed variables. F or exa m p le, eq u a tions A and B in d ica te that adding the fiscal variables to the G N P m o n e y regression d oes not significantly im p rov e the e x planatory p o w e r: T h e t-statistics for the re sp ectiv e Seis are n ot greater than n orm ally a ccep ta b le critical values. T h e s e results in d icate that, o n c e the influ en ce o f m o n e y grow th is a ccou n ted for, the a dd ition o f the fiscal variables d oes not statistically im p rove th e e x p la n a tio n o f G N P g r o w th . T h e s e resu lts further p o in t to the statistical d om in a n ce o f m o n e y g ro w th o v e r eith e r fiscal m ea su re in e x p la in in g ch an ges in G N P grow th. results are rep orted in table 2 .10 T h e C h o w test results in d ica te that the stability h yp oth esis is re je cte d only for the reg ression s o f G N P grow th on the tw o fiscal variables, that is, e q u a tions D and E : T h e ca lcu la te d F-statistics e x c e e d the releva n t 5 p e rce n t critical valu e. T h e ca lcu la ted F-statistics for the equations that in clu d e m on ey grow th or use m o n e y grow th alon e to exp la in G N P grow th in d ica te that th ey p r o v id e structurally stable p a r a m e te r e s tim a te s a cro s s th e I /1 9 6 0 -I V /1 9 8 0 p e rio d . T h e s e findings im p ly that the rela tion sh ip b e t w e e n G N P g r o w th a n d m o n e y g r o w th has r e m a in e d r e la t iv e ly s t a b le a cro s s th e 2 0 -y e a r sam ple p e rio d . O n the oth er hand, th ey su ggest that th e e ffe c ts o f fisca l a ctio n s o n G N P g ro w th are uncertain and may b e q u ite d ifferen t across e c o n o m ic en viron m en ts. T h e s e co n d test exam ines the statistical causal o rd e rin g b e tw e e n the m on etary and fiscal variables and G N P grow th. T h is test p ro ce d u re , d e v e lo p e d b y G ranger, in v o lv e s estim ating a set o f eq u a tion s for ea ch p air-w ise test.11 T o test for G ran ger causality, it is assum ed that the inform ation releva n t to the p re d ictio n o f the re s p e ctiv e variables is co n ta in e d s o le ly in the data series Y and X (e.g., G N P and m o n e y or fiscal variables). T h e test p ro ce d u re c o n sists o f estim ating the equ ation s n n (2) Y(t) = 1 T h e results o f tw o additional statistical tests are rep orted in this section . T h e first test exam ines the stability properties o f the alternative specification s rep orted in table 1. T h e s e con d test p rovid es som e e v id e n c e about the causal ord erin g o f the m onetary and fiscal variables w ith resp ect to G N P grow th. 1 ftY(t-j) + € (t) j= l and (3) X (t) = in i i= l STABILITY AN D CAUSALITY TESTS « , X(t-i) + i= l K, X (t-i) + m 1 Sj Y(t-j) + 7 ,(t). j= l It is assu m ed that e(t) and r)(t) are u n correla ted error se rie s. U n id ir e c tio n a l ca u s a tio n from X to Y is p re s e n te d in ta ble 1, the I/1 9 6 0 -IV /1 9 8 0 sam p le p e rio d was split at 11/1970 (the m id -p oin t o f the sam ple) and the C h o w test was a p p lied . T h e test 10The mid-point was chosen to maximize the pow er o f the test. On this point, see John U. Farley, Melvin Hinich and Timothy W. McGuire, “ Some Comparisons o f Tests for a Shift in the Slopes o f a Multivariate Linear Time Series M odel,” Journal o f Econo metrics (August 1975), pp. 297-318. The Chow test may be inappropriate given the existence o f serially correlated errors, as evidenced in equations D and E. Even so, the stability test results in table 2 are based on the ordinary least squares estimation found in table 1 as a matter o f consistency. Using GLS estimates o f equations D and E in table 1 indicates that the stability hypothesis still is rejected at the 5 percent level: The F-statistics are 3.02 and 3.44 for the GLS estimates o f equations D and E, respectively. 9The reader may note that equation C, the specification em ploy ing only the m oney growth variable, is not beset by the problems o f serial correlation. This, along with the textual evi dence, strongly suggests that the misspecification problems affect only the regressions o f GNP growth on fiscal variables. 11C. W. J. Granger, “ Investigating Causal Relations by E cono metric M odels and Cross-Spectral Methods,” Econometrica (July 1969), pp. 424-38. For a useful comparison o f various causal tests, see Edgar L. Feige and Douglas K. Pearce, “ The Casual Causal Relationships Between M oney and Income: Some Caveats for Time Series Analysis,” Review o f Economics and Statistics (November 1979), pp. 521-33. T o test the tem p ora l stability o f the eq u a tion s Digitized for 20FRASER JANUARY 1982 FEDERAL RESERVE BANK OF ST. LOUIS Table 2 Table 3 Stability Test Results for GNP Equations, 11/1970 Break Point Granger Test Results Equation1 Calculated F-statistic Reject Stability2 A 1.63 No B 1.62 No C 1.53 No D 2.83 Yes E 3.62 Yes 'See table 1 for description of alternative equations. Exogeneity test1 Calculated F-statistic M —» Y 2.682 Y -» M 0.56 Reject exogeneity of policy variable No Y 1.10 Y -» HEGE 0.24 AHEGS - * Y 0.66 Y - » A HEGS 2.922 HEGE Yes 2Based on 5 percent critical value (F(7,70) = 2.14). Yes im p lie d if the estim ated coefficie n ts on the la gg ed X variable in eq u a tion 2 are statistically d ifferen t from zero as a grou p and th e set o f estim ated coefficien ts on the la g g ed Y variable in equation 3 is n ot statisti ca lly d ifferen t from zero. C on v ersely , u n id irection al causation from Y to X exists i f the set o f lagg ed X co e fficien ts in equ ation 2 is not statistically d ifferen t from zero and the set o f lagg ed Ys in eq u a tion 3 is statistically n on -zero as a group. F e e d b a ck is su g gested w h en the sets o f X and Y coefficie n ts are statistically significan t from zero in b oth equations 2 and 3. In d e p e n d e n c e occu rs w h en the sets o f X and Y co e fficie n ts are not significant in eq ua tion s 2 and 3. T h e G ranger test results, b a s e d on the 1/1960IV /1980 sam ple, are rep orted in table 3 .12 T h e first pair-w ise test is d o n e w ith m o n e y grow th and G N P grow th (M ,Y ). T h e ca lcu la ted F-statistics suggest that there is u n id irection a l causation from m on e y to G N P .13 In the s e c o n d bivariate test, w ith the grow th o f h ig h -e m p lo y m e n t g ov ern m en t e x p e n d i tures (H E G E ), n eith er F -valu e surpasses the 5 p e r cen t critical value. T h e s e results su ggest that the grow th o f h ig h -e m p lo y m e n t exp en ditu res and G N P grow th are in d e p e n d e n t. T h e last test, w h ich uses the ch a n g e in the h ig h -e m p lo y m e n t gov ern m e n t 12The estimated equations consist of four lagged dependent vari ables and four lagged “ causal” variables. Because the Granger test requires the data to exhibit stationary characteristics, a linear time trend and seasonal dummies were entered into the estim ated regressions. M oreover, the fu ll-p eriod results are based on the finding o f no structural break in the estimated test equations. For a further discussion and application o f the Granger test, see Hafer, “ Selecting a Monetary Indicator.” 13This finding is supported by previous empirical research. See Christopher A. Sims, “ Money, Income, and Causality,” Ameri can Economic Review (September 1972), pp. 540-52; Mehra and Spencer, “ The St. Louis Equation and Reverse Causation;” and Hafer, “ Selecting a Monetary Indicator.” Contrary findings are presented in Feige and Pearce, “ Casual Causal Relationships.” 1The arrow indicates hypothesized direction of causality. Significant at 5 percent critical value (F(4,71) = 2.50). surplus variable (A H E G S ), indicates that u n id ire c tional causation from G N P grow th to ch an ges in the surplus m easure ca n n ot b e rejected. T o sum m arize, the statistical e v id e n c e suggests th at m o n e y g r o w th c a u s e s G N P g r o w th . T h e fin d in g s p r e s e n t e d h e re p r o v id e n o e m p ir ic a l su pp ort for the “ reverse cau sation ” argum ent. T h e e v id e n c e a b o u t th e r e la tio n s h ip b e t w e e n fiscal m e a su re s and G N P g ro w th , h o w e v e r , su g g e sts that G N P causes fiscal actions. T h is is true sp e cifi ca lly for the surplus m easu re; the o u tco m e for the exp en ditu res variable indicates n o causal relation ship in eith er d ire ctio n (in d e p e n d e n ce ). T h e s e find ings indicate that there is no reliable relation sh ip b e tw e e n G N P grow th and the fiscal m easures u sed in this article. T h e p r e c e d in g C h o w test results su pp ort this co n c lu s io n b y sh o w in g that o n ly the relation sh ip b e tw e e n m o n e y and G N P is statistically stable across the entire sam ple p eriod . POST-SAMPLE ESTIMATES T h e final p ie c e o f e v id e n c e u sed to assess the relation sh ip b e tw e e n G N P grow th and m onetary/ fis ca l a c t io n s is a c o m p a r i s o n o f th e r e la t iv e p o s t-s a m p le p r e d ic t iv e a b ilit ie s o f s e le c t e d equations fou n d in table 1. T o this en d, each equation in itia lly w as e stim a te d from 1/1960 th rou gh IV / 1970. Four-quarter-ahead estim ates o f G N P grow th w e r e th en g e n e r a te d u sin g actual v a lu es o f the m o n e y and fiscal m easures o v e r this p e rio d . T h e 21 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 Table 4 Post-Sample Prediction Errors of GNP Growth1 Sample period from 1/1960 to fourth quarter: Equation Money only Money growth and highemployment expenditures Money growth and highemployment surplus 1970 0.22% 0.20% 1971 2.06 1.58 1972 1.35 2.07 1.10 1973 —1.18 -1 .8 3 -0 .1 8 1974 2.28 2.01 0.69 1975 0.41 0.65 1.70 1976 1.23 1.14 1.36 1977 2.58 2.80 2.94 1978 -1 .8 3 -1 .9 4 -0 .4 8 1979 -0.11 -0 .6 0 -1 .4 8 Average Error 0.70% 0.61% -0.31 % 1.98 0.73% ’ Errors are actual less predicted and are calculated for the four quarters following the sample period. estim ation p e rio d was then ex te n d e d to in clu d e the fo u r quarters o f 1971, and four-quarter-ahead fo re casts for 1972 w e re con structed . T his p ro ce d u re was co n tin u e d ea ch yea r th rou gh 1980. T h e average errors in p re d ictin g G N P grow th are p resen te d in table 4. S in ce the p u rp ose of these com parison s is to determ in e if the alternative fiscal p o lic y m easures add anything to the estim ates o f G N P grow th once money growth is known, results for three equations are p r e s e n t e d : m o n e y o n ly , m o n e y a n d h ig h e m p lo y m e n t exp en d itu res, and m o n e y and h igh e m p lo y m e n t surplus. T h e results in table 4 in d icate that, on average, m o n e y alon e (equ a tion C ) results in G N P grow th p re d ictio n errors that are less than 1 p ercen tage point. T h e addition o f the exp en d itu res m easu re o f fiscal p o lic y (equ ation A) leads to on ly a m arginal red u ction in the average error, w h ile the addition o f the surplus m easure (equ ation B) leads to a m arginal increase in th e average foreca st error. A lth o u g h th ese ch a n ges m ay appear rather substantial, tests rev ea l that the average errors for the alternative specification s are not statistically differen t. In fact, the ca lcu la ted t-statistics com p a rin g the averages are all less than u n ity.14 B ased on the eq u a tion s’ relative p re d ictiv e abilities, th erefore, on e w o u ld d o ju st as 14For instance, the t-statistic comparing the average error term from the “ money only” regression to that using money and the expenditures variable is 0.14. The statistic comparing money only and the regression using the surplus measure is -0 .5 0 . Digitized for22 FRASER w e ll, on average, to use o n ly m o n e y grow th to p re d ict future G N P grow th as u sin g m o n e y grow th and eith er o f the tw o fiscal p o lic y m easures exa m in ed in this article. CONCLUSION T h is paper has re -exa m in ed the role o f fiscal p o lic y in the con tex t o f the St. L ou is equ ation . T h e results are b roa d ly con sisten t w ith p reviou s findings. S p e cifically , fiscal actions exert n eith er a significan t n or lasting im pa ct on the grow th o f G N P . A u g m e n tin g p re v io u s w ork, h o w e v e r, the results h ere p ro v id e further e v id e n c e against the relian ce on fiscal p o lic y m easures to explain m ovem en ts in G N P . T h ree co n clu s io n s can b e draw n from th ese test results: T h e fiscal p o lic y m easures 1) d o n ot signifi cantly increase the explanatory p o w e r or forecastin g a b ility o f an e q u a tio n that a lrea d y in co rp o ra te s m o n e y grow th ; 2) d o n ot e xh ib it stable statistical re la tio n s h ip s w ith G N P g ro w th ; a n d 3) are n o t e x o g e n o u s w ith re sp e ct to G N P grow th. T h e e v i d e n c e suggests that fiscal actions are in e ffe ctiv e for stabilization p u rp oses. M o re o v e r, the results add in creasin g stature to the u se o f m onetary p o lic y as a tool in stab ilizin g fluctuations in e c o n o m ic activity. Comparing the mean forecasts generated by the two fiscal equa tions yielded a t-statistic o f -0 .1 9 . Food and Agriculture — Current Situation and Prospects for 1982 CLIFTON B. LUTTRELL T A H IS article is largely a sum m ary o f the U nited States fo o d and agricultural ou tlook for 1982 as d is cu ssed at the A nnual N ational A gricultural O u tlook C o n fe r e n ce in N o v e m b e r 1981. T h e overall ou tlook for this year is for in crea sed p rod u ction and lo w e r p rices for m ost farm products. Retail fo o d p rices, w h ich rose abou t 8 p ercen t last year, are lik ely to rise at an e v e n slo w e r rate in 1982. In a dd ition to the p rosp ectiv e im pact o f larger farm com m od ity su p p lies, agricultural p rices w ill also b e a ffe c te d b y a s lo w in g in the grow th of overall fo o d d em an d this year. R estrictive d om e stic m onetary p o licie s d e s ig n e d to re d u ce inflation w ill ten d to re d u ce the grow th in d o m e stic dem an d . Sim ilar restrictive m on etary p o lic ie s in a n u m ber of foreign coun tries p oin t to a slu ggish foreign d em an d for U.S. farm prod u cts. D e s p ite the e x p e c te d relatively lo w prices o f farm p rod u cts and the rising farm p rod u ction costs, the quantities o f m ost co m m od ities available, such as cereals and bakery p rod u cts, b e e f and vea l, and fresh fruits and veg eta b les, are lik ely to rise further in 1982. A record 1981 fe e d grain cro p b o o s te d 1981-82 fe e d grain su pp lies to 280.1 b illio n tons, 11 p ercen t a b o v e last year. L arge fe e d grain su p p lies and low e r fe e d p rices w ill in d u ce farmers to m aintain livestock p ro d u ction at a rela tively h igh le v e l. W h eat and rice su pp lies are at re co rd levels and a large potato crop was h arvested. O f th e m ajor crop s, o n ly the 1981 fruit cro p was som ew h at sm aller than a year ago. Th u s, the overall farm and fo o d o u tlo o k for this year is for in crea sed ou tp ut o f farm prod u cts, e s p e c ia lly liv e stock p rodu cts, and relatively large fo o d su pp lies. N et farm in co m e after in ven tory adjustm ents rose in 1981 to about $22 b illio n or a bou t $2 b illio n a bove the 1980 total. N et in co m e in 1982 w ill b e d e te r m in ed ch iefly b y returns on crop s to b e p la n ted later this year. H o w e v e r , it m ay d e c lin e from last yea r’ s le v e l, e sp e c ia lly i f th ere are further increases in crop in ven tories in the 1982-83 cr o p year. OU TLO O K FOR AGRICULTURE E c o n o m ic co n d itio n s in the farm sector in 1982 are lik ely to ch a n ge v e ry little from 1981, e sp e c ia lly for the next several m onths. T h e su p p ly o f farm prod u cts relative to d em a n d is e x p e cte d to rem ain relatively high and, d esp ite further overall inflation, farm c o m m od ity p rices w ill p ro b a b ly increase little, if any. P rodu ction ex p e n se s, w h ile risin g at a s lo w e r rate than a year ago, are e x p e cte d to con tin u e to increase, resu ltin g in little ch a n ge in n et cash receip ts through m id-vear. 23 FEDERAL RESERVE BANK OF ST. LOUIS T h e su ccess o f 1982 crop s, h o w e v e r, w ill b e a m ajor factor in d eterm in in g con d ition s in the last h a lf o f the year. B eca u se the d em an d for crop s is relative ly in elastic in the short run, another large crop w ill result in a further d e c lin e in relative p rices o f farm p rod u cts and lo w e r real farm in co m e . A sm all crop , h o w e v e r, co u ld lead to an in crea se in farm p rices and in com es. Recent Production and Price Movements T o som e extent, the overall ou tlook for fo o d and agriculture this year is related to the relative p rices o f th ese co m m o d itie s vs. oth er com m od ities. This rela tion sh ip is cle a rly o b s e r v a b le in the grow th rates o f cro p and livestock p rod u ction d urin g recen t d e c a d e s . P ro d u c tio n o f all farm p rod u cts in the U n ited States rose at abou t 2.5 p ercen t p er year th ro u g h th e 1 970s, f o ll o w in g a r e la tiv e ly s lo w grow th rate o f abou t 0.6 p ercen t p er year in the last h a lf o f the 1960s (table 1). T h e h igh er grow th rate for farm prod u cts in the 1970s fo llo w e d rela tiv e ly h ig h er farm p rod u ct p rices, e sp e c ia lly in the first h a lf o f the d eca d e. D u rin g the 1965-70 p eriod , the p rice o f all farm prod u cts rose on average o n ly 2.4 p ercen t p er year, co m p a red w ith an average in crease o f 4.2 p ercen t p e r year for the con su m er p rice index. T h is d e clin e in relative p rice o f farm p rod u cts, e s p e c ia lly for crop s, re d u ce d the in cen tiv es for p rod u ction . T h e increase in exp ort d em an d for farm p rodu cts in the early 1970s, c o u p le d w ith the re d u ce d stocks o f crop s h e ld b y the g ov ern m en t in p rice support operation s, le d to a sharp turnaround in the p rices o f farm p rod u cts com p a red w ith gen eral p rices. D u r ing th e 1970-75 p eriod , the average p rices o f all farm p ro d u cts rose sharply (11 p e rc e n t p e r year) and p ro d u ction q u ick ly in crea sed in resp on se. O verall farm output, as sh ow n in table 1, rose at a 2 .45 p er cen t rate in the first h a lf o f the 1970s, co m p a re d w ith a 0.6 p ercen t rate du rin g the p reviou s five years. C ro p p rod u ction grow th was greater than liv e stock p rod u ction d u rin g th e 1970s, reflectin g the relatively larger rise in cro p p rices, e sp e c ia lly dur in g th e first h a lf o f th e d e ca d e . D u rin g the 1970-75 p e rio d , crop p rod u ction rose at a 3.89 p e rce n t rate co m p a red w ith - 0 . 7 7 p e rce n t for livestock p ro d u c tion. In contrast, d u rin g the last half of the 1960s, livestock ou tp ut in crea sed at a 2.02 p erce n t rate c o m pared w ith a 0.20 p ercen t rate o f in crease for crops. T h e g ov ern m en t h e ld large in ven tories of crops a cq u ire d in p rice su p p ort op era tion s d u rin g the 24FRASER Digitized for JANUARY 1982 Table 1 Rates of Change of Production and Prices of Farm Products Percent changes, annual rates1 1965-70 1970-75 1975-81 Prices All farm products Livestock Crops Consumer prices 2.4% 4.7 -0 .6 4.2 11.0% 7.8 15.0 8.0 5.5% 6.7 4.5 8.0 Production All farm products Livestock Crops 0.60 2.02 0.20 2.45 -0 .7 7 3.89 2.47 2.04 3.41 1Beginning year average to ending year average. SOURCE: U.S. Department of Agriculture, A g ricu ltu ra l Statistics (1980), and A g ricu ltu ra l O utlook (1981). 1960s. As inflation a cce le ra te d in the late 1960s, h o w ever, gov e rn m e n t p rice supports for th ese p rod u cts fa ile d to k e e p p a ce w ith gen eral inflation, and g o v e rn m en t-h eld stocks w e re re d u ce d . T h e valu e o f farm com m o d itie s o w n e d b y the g o v e rn m e n t (large ly crop s) d e c lin e d from $4.1 b illio n at the e n d o f 1965 to $1.6 b illio n at the e n d o f 1970.1 Partly as a result o f this red u ction in gov e rn m e n t stocks, cro p p rices d e c lin e d both in n om inal term s ( —0.6 p e rce n t p er year) and relative to oth er p rices. F o llo w in g the sharp increase in exports (largely crop s) and the d e clin e o f stocks in the early 1970s, cro p p rices rose at a 15 p e rce n t rate and livestock p ro d u ct p rices at a 7.8 p e rce n t rate.2 D u rin g this p e rio d , the annual rate o f inflation as m ea su red b y the co n su m e r p rice ind ex was 8.0 p ercen t. A siza ble ch a n ge in the relative p ro d u ctio n o f crop s and liv e stock fo llo w e d : C rop ou tp ut rose at a 3.89 p e rce n t rate w h ile livestock ou tp ut d e c lin e d at a 0 .7 7 p e r cen t rate. This d e clin e re fle cte d th e h ig h e r-p rice d fe e d and the slo w e r grow th in d e m a n d for livestock produ cts. D u rin g m ore re ce n t years, 1975-81, the grow th rate o f exp ort d em a n d for crop s has d e c lin e d , and agricultural p rices, w h ich rem ain ed rela tively h igh 'U.S. Department o f Agriculture, Agricultural Statistics (1980). 2For an analysis o f the rise in export demand, see Clifton B. Luttrell, “ Rising Farm Exports and International Trade Poli cies,” this Review (July 1979), pp. 3-10. FEDERAL RESERVE BANK OF ST. LOUIS into the last h a lf o f the d e ca d e , n ow have d e c lin e d relative to oth er p rices. G row th in the average p rice o f all farm p rodu cts d e ce le ra te d d urin g the 1975-81 p e rio d to a 5.5 p ercen t rate com p a red w ith the g e n eral inflation rate o f 8 p ercen t as m ea su red b y the c o n sum er p rice index. T h e p rice o f liv estock produ cts, h o w e v er, rose at a faster rate than crop s, and liv e stock p rod u ction a ccelera ted som ew h at from the 1970-75 rate. N everth eless, partly as a c o n s e q u e n c e of the favorable 1981 w eather, crop ou tput rose at a faster rate than liv estock output. Incentive f o r Farm Production A gricultural p rices in late 1981 w ere relatively low w h e n com p a red w ith the gen eral p rice level. F or exa m p le, late in the fourth quarter, farm p rod u ct p rices w e re 8 p erce n t b e lo w levels o f a year earlier and o n ly 26 p ercen t a b ov e their 1977 average. In contrast, the con su m er p rice in d ex was abou t 10 p ercen t a b ov e a year earlier and 54 p ercen t a b o v e its 1977 average. M oreov er, p rices p aid b y farmers for p ro d u ction item s w e re abou t the same as a year earlier and 46 p e rce n t a b ov e the 1977 average. W h ile there is less in cen tiv e for in crea sed overall farm p rod u ction this year than last, su p p ly and d em an d factors are m ore en cou ra g in g for livestock p rod u cers than for cro p p rod u cers. Large fe e d su p p lies and a ccom p a n y in g rela tively lo w fe e d p rices m ean lo w e r fe e d cost for livestock p rod u cers and, co n se q u en tly , som e im p rov em en t in livestock re turns relative to those for crop s. A verage crop prices in the fourth quarter o f 1981 w e re d ow n 15 p ercen t from a year earlier, w h ile liv estock prices w ere d ow n o n ly 10 p ercen t.3 Sim ilarly, a com p a rison o f recen t p rices w ith the average for 1977 sh ow s that, d urin g this p e riod , the average p rice o f liv estock p rodu cts rose 33 p ercen t w h ile that o f crop s in crea sed on ly 19 percent. Meat Animals and Poultry A relatively high le v e l o f m eat and p ou ltry p ro d u c tio n is e x p e c t e d later this y e a r f o llo w in g a slight red u ction in ou tp ut d u rin g the first quarter. JANUARY 1982 p e cte d to rem ain large, and p lacem en ts of cattle in fe e d lo ts are lik ely to exp an d from the relatively lo w n u m b er cu rren tly o n feed . A lth ou gh ))ork p ro d u ctio n in early 1982 w ill lik ely b e greater than was p ro je cte d last autum n, slaughter for the year is e x p e cte d to b e 4 p e rce n t to 6 p ercen t less than a year ago. B ased on in ten d ed farrow ings, h o w e v e r, the slaughter le v e l in late 1982 m ay equal that o f late 1981. D e sp ite the m ore favorable fe e d p rice s this year, broiler p rod u cers are p la n n in g o n ly a sm all exp an sion for 1982. T h is fo llo w s an 11 p e rce n t increase last year, w h e n sizable losses w e r e in cu rred o n m uch o f the ou tput as the larger than e x p e cte d pork su p p lie s d e p r e ss e d poultry p rice s.4 B roiler ou tp u t w ill lik ely con tin u e a b o v e 1981 le v e ls in the first part o f this year but m ay d e c lin e to 1981 levels in the fourth quarter. T u rk ey p rod u ction in 1982 is lik ely to b e lo w e r than 1981 p ro d u ctio n as a result o f re d u c e d p rices and rela tively lo w net returns last year. G iv e n the p ro sp e ctiv e su p p ly of and d em an d for m eat and poultry, 1982 average p rices o f th ese p ro d ucts are not e x p e cte d to greatly e x c e e d the average o f 1981. C h o ic e b e e f p rices are e x p e cte d to increase on ly m od era tely through the first half of the year and m ay average o n ly 1 p e rce n t to 2 p ercen t a b o v e the range o f the past tw o years. H o g p rices in the first half o f the year are e x p e c te d to average about 10 p ercen t a b o v e the 1981 le v e l b u t are lik ely to b e o n ly m o d erately h ig h er in the s e co n d half of the year. B roiler p rice s are e x p e cte d to average slightly h igh er than last year, lo w e r in the first h a lf and h ig h er in the s e co n d . T u rk ey p rice s are e x p e cte d to rise to about 70 cents p er p o u n d in the s e co n d h a lf o f the year, up from 55-56 cen ts p er p o u n d in late 1981. Dairy Products and Eggs R eflectin g the relatively h igh g ov ern m en t p rice supports ($13.10 p er h u n d re d w e ig h t (cw t.) for 3.67 p e rce n t m ilk) and lo w e r fe e d p rices, m ilk p ro d u ctio n is e x p ected to increase, at least in the first h a lf o fl9 8 2 . T h e U.S. D epartm en t o f A gricultu re projects that dairy output m ay d e c lin e from year-earlier le v e ls in the s e co n d h a lf o f the year, but eq u a l last year’ s ou t put for the year as a w h o le . A n industry report given B eef p rod u ction is lik e ly to increase 2 p ercen t to 4 p ercen t. T h e sla u gh ter o f n o n fe d cattle is ex3U.S. Department o f Agriculture, Agricultural Outlook (D ecem ber 1981). 4Pork is a good substitute for poultry. Hence, with large pork supplies and relatively low pork prices, consumers tend to substitute pork for poultry in their diets. This results in reductions in the demand for and price o f poultry as well. 25 FEDERAL RESERVE BANK OF ST. LOUIS at the ou tlook co n fe re n c e , h ow ev er, p oin ts to a further rise in m ilk p rod u ction o f 2 p erce n t from the re co rd 1981 le v e l.5 In either case, p rod u ction of dairy prod u cts is e x p e cte d to e x c e e d d o m e stic con su m p tion plus ex ports in the first h a lf o f 1982, w h ich w ill result in further a dd ition s to g ov ern m en t stocks through p rice support operations. G ov ern m en t purch ases in p rice support operation s on a m ilk eq u iv a len t basis totaled abou t 11 p erce n t o f p rod u ction in the first three quarters o f 1981 and such purchases w ill rem ain large. Industry estim ates o f the cost to taxpayers o f such purchases are as h igh as $2.5 b illio n . M ilk p rice s for all o f 1982 are e x p e c te d to average about 1 p ercen t a b ov e 1981. W ith favorable fe e d costs in p rosp ect, dairy farmers sh ou ld realize h ig h er n et returns. E g g p rod u cers have b e e n red u cin g the n u m b e r o f re p la cem en t p ullets as a result o f u n favorable p rice relationships sin ce early 1980. E g g p rod u ction , h o w ever, has b e e n m ain tain ed b y p rod u cers k eep in g th eir hen s in p rod u ction lon ger. P rodu ction in late 1981 was d o w n sligh tly from 1980 lev els, and m ay con tin u e b e lo w year-earlier lev els through m id1982. W ith som ew h at h ig h er e g g p rices in p rosp ect, p ro d u ctio n is lik ely to rise in the s e co n d h a lf o f the year and eq u a l last yea r’ s le v e l for the en tire year. E g g p rices are e x p e cte d to average abou t 74 cents p er d o z e n for the year, u p from 70 cen ts p er d o ze n in 1981. C rops As a result o f record crop s o f w h eat, coa rse grain and rice this m arketing year, w o r ld su p p lies o f m ost crops are w e ll a bove their 1980/81 lev els. T h e se larger su p p lies have le d to lo w e r average p rices and less in ce n tiv e for p rod u ction grow th in 1982. N e v e r th eless, w ea th er is a m ajor factor in cro p p rod u ctio n , and w eather con dition s as the crop d ev elop s w ill b e a m ajor factor in d eterm in in g both the size of the 1982 crop and the average p rices r e c e iv e d b y farmers. Food Grains (Wheat and Rice) T h e 1981 record wheat cro p o f 2.75 b illio n b u sh els, u p 16 p e rce n t from 1980, c o m b in e d w ith a rela tively large June 1981 carryover, p rov id e s a 1981/82 m arket su pp ly o f 3.74 b illio n b u sh els. This is 14 p e rce n t a b ov e the p rev iou s record o f 1980/81. 5E. Linw ood Tipton, Dairy Outlook, Production Consumption Estimates 1982 (Government Printing Office). Digitized for26 FRASER JANUARY 1982 E xport d em a n d for U.S. w h ea t is e x p e cte d to rise as a resu lt o f rela tiv ely sm all cro p s in the S oviet U nion and Eastern E u rop e. W h eat exports are p ro je c t e d at a re co rd 1.9 b illio n b u sh e ls, u p from 1.5 b illio n last year and an average o f 1.05 b illio n fo r the 1970/79 d e ca d e . In cre a se d sh ipm ents are in p ro s p e ct to the S oviets, India, Brazil, Iran and M o r o cc o , b u t C h in a is e x p e cte d to rem ain the largest U.S. w h ea t cu stom er. H e n c e , the average farm p rice for 1981/82 w h ea t m ay rem ain near that o f last year, d esp ite the in crea sed short-run su pp ly. Stocks o f w h e a t at yea r-en d are e x p e cte d to b e d o w n to abou t 908 m illio n b u sh els from 991 m illion b u sh els last year, reflectin g th e in crea sed exports. N everth eless, th e Secretary o f A gricu ltu re last S ep tem b er a n n o u n ce d plans to im p le m e n t a re d u c e d a creage program for the 1982 w h ea t crop . Rice p rod u ction in 1981 was estim ated at 178.8 m illio n c w t ., 12 p e r c e n t a b o v e th e p r io r 1980 re co rd le v e l. T h is p lu s ca rryov er stocks results in a total rice su p p ly o f 195.4 m illio n cw t. for the 1981/ 82 m arketing year, or 14 p e rce n t m ore than the p rev iou s 1980/81 record . R ice exports for 1981/82 are p ro je cte d at 79.0 m illio n cw t., d o w n from 91.4 m illio n last year, w ith d o m e stic use at 56.5 m illio n cw t., u p 3.8 p e rce n t from last year. H e n ce , yea r-en d stocks are lik e ly to total 55 m illio n to 6 0 m illio n cw t., w e ll a b o v e last y ea r’ s carryover. B ecau se o f reco rd p ro d u ctio n , carryover stocks and th e e x p e cte d d e c lin e in exp ort d em a n d , the season -average p rice o f rice is e x p e cte d to d e c lin e to abou t $10 p er cw t., $2 p e r cw t. b e lo w last year’ s p rice. H e n c e , w ith a target p rice (b e lo w w h ich d e fic ie n c y paym ents are m ade to e lig ib le farmers) o f $10.68 p er cw t. in effect, d e fic ie n c y paym ents b y the g o v e rn m e n t to e lig ib le p rod u cers are e x p e cte d . Feed Grains (Corn, Oats, Barley, Sorghum Grain) T h e U.S. fe e d grain su p p ly o f 2 80.2 m illio n m etric tons is abou t 10 p e rce n t m ore than a year ago. A re co rd cr o p o f 2 45 m illio n tons, 3 p e rc e n t a b o v e the p reviou s 1979 re co rd cro p and 24 p e rc e n t a b o v e the 1980 cro p , w as p ro d u ce d . C o rn p ro d u ctio n , a bou t four-fifths o f total fe e d grain, was up 22 p ercent. W ith the large cro p , the su p p ly o f fe e d grain in creased faster than d em an d , and the p rice d e clin e d . C o m p rice s are e x p e c te d to average o n ly about JANUARY 1982 FEDERAL RESERVE BANK OF ST. LOUIS Table 2 Changes in Per Capita Consumption of Major Food Products Percent change in Food Group All foods Cereals and bakery products Beef and veal Pork Poultry Eggs Dairy products Fresh fruits Fresh vegetables Processed fruits and vegetables Sugar and sweets Fats and oils Nonalcoholic beverages 1980 consumption in pounds (retail weight) 1978 1408 -0 .4 % 1979 0.8% 1980 1981 (Preliminary) -0 .9 % -1 .4 % 1982 (Forecast) 1% 150 78 68 61 35 308 84 207 -1 .1 -5 .7 0.2 4.7 1.8 0.4 -1 .8 -0 .7 3.8 -1 0 .9 14.1 8.4 2.0 - 0 .6 3.4 1.9 -0 .5 - 2 .3 7.1 0.2 -2 .0 -3 .0 4.0 0.9 0.7 0.4 -7 .0 2.0 - 3 .4 0.0 - 3 .0 -3 .0 1 2-3 -6 0 -2 1 1 5 142 133 55 11 0.9 0.7 3.4 7.7 1.5 1.5 2.5 3.7 -2 .8 -2 .7 0.4 -5 .0 -1 .5 - 2 .0 -0 .8 3.6 -5 1-2 1-2 1 SOURCE: U.S. Department of Agriculture, The Current Food S ituation and O utlook fo r 1981. $2.60-$2.90 p e r b u s h e l for the year, d o w n from a bou t $3.10 p er b u s h e l last year. E xports are lik ely to b e u p som ew h at, 74 m illio n m etric tons com p a red w ith 69 m illio n in 1980/81, a nd d o m e stic use m ay in crea se from 147 m illion to 156 m illion m etric tons. N everth eless, a 50 m illio n m etric ton carryover is estim ated for this year, w e ll a b o v e the 35 m illio n last year. Fats and Oils (Soybeans and Cottonseed) Soybean p rod u ction o f 2.1 b illio n b u sh els last year was 415 m illion b u sh els m ore than in 1980. C o u p le d with relatively large b e g in n in g stocks, the total quantity available for 1981/82 was b o o ste d to 2.4 b illio n b u sh els. T h is is 13 p ercen t a b ov e a year ago and on ly 1 p ercen t b e lo w the record 1979/80 le v e l. W ith the large su p p ly available and n o m ajor increase in d em an d , p rices to farmers have d e c lin e d from year-ago levels and are e x p e cte d to average from $5.50 to $7.00 p er b u sh el for 1981/82, w e ll b e lo w the $7.61 estim ated for 1981. Cottonseed p rod u ction m ay reach a re c o rd 6.2 m illio n short tons. W h en a d d ed to last yea r’ s very sm all carryover o f 0.4 m illion short tons, su pp lies this year w ill total a rela tiv ely large 6.6 m illion short tons. P rices for co tto n se e d a veraged $105.50 a ton d u rin g the early autum n or $10 p er ton less than a year earlier. T h e season average p rice is forecast at $110 a ton, 12 p e rce n t less than last year, d u e to the rela tiv ely large su p p ly o f co tto n se e d and soybean s, w h ich is a g o o d substitute for co tto n se e d . Cotton T h e rela tively large 1981 cotton cro p o f 15.5 m il lio n bales (480 p ou n d s p e r bale) plu s a 2.7 m illion b a le carryover p rovid es 18.2 m illion bales for d o m e s tic con su m p tion and export during 1981/82. W ith d o m e stic cotton co n su m p tio n estim ated at 6.2 m il lio n bales and exports at 7.0 m illio n b a les, carryover stocks at the e n d o f 1981/82 are p ro je c te d to be rela tively h igh at 5.0 m illio n bales. C otton p rod u cers re c e iv e a d e ficie n cy p a y m e n tb y the govern m en t if the 1981 calen d ar year average p rice o f u pland cotton is less than the target p rice o f 70.87 cents p er p ou n d . S u ch d e fic ie n c y paym ents may b e relatively large for the 1981 cro p as the S ep tem b er p rice for u pla n d cotton was o n ly abou t 63 cents p er p o u n d . Tobacco T o b a c c o p ro d u ctio n in 1981 was estim ated at 2.01 b illio n p ou n d s, up 13 p e rce n t from a year earlier. T h e large cr o p and som e in crea sed ca rryover in d i 27 FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 Table 3 Changes in the Consumer Price Index of Food Percent change in Component Food Food away from home Food at home Cereals and bakery products Beef and veal Pork Other meats Poultry Fish and seafood Eggs Dairy products Fresh fruits Fresh vegetables Processed fruits and vegetables Sugar and sweets Fats and oils Nonalcoholic beverages Other prepared foods Relative importance in food CPI 1978 1979 100.0% 30.7 69.3 10.0% 9.0 10.5 10.9% 11.2 10.8 8.7 9.8 4.7 3.0 2.3 2.3 1.3 9.3 2.4 2.8 8.9 22.9 12.9 17.8 10.3 9.5 -5 .5 6.7 19.4 7.9 10.1 27.3 1.5 14.7 5.0 9.8 9.5 11.6 12.4 2.9 11.9 5.7 -3 .4 3.8 5.1 9.2 -1 .8 9.8 6.2 8.9 10.2 1.8 9.8 5.0 5.5 8.9 7.8 7.5 5.3 20.1 7-8 5-7 8-11 6-7 3-5 8-9 4-5 4-5 8-9 -1 - 0 4.5 2.9 1.9 7.6 5.8 10.5 12.2 9.5 5.7 8.0 8.6 7.8 8.0 5.0 10.1 7.0 22.9 6.6 10.6 10.8 12.2 8.2 11.5 4.1 10.5 9-10 2-3 5-6 2-3 9-10 1980 8.6% 9.9 8.0 1981 (Preliminary) 8.2% 9.3 7.8 1982 (Forecas 7% 8 6 SOURCE: U.S. Department of Agriculture, The Current Food Situation and O utlook fo r 1981. cate that total su pp lies for the 1981/82 m arketing year w ill b e abou t 5 p e rce n t larger than a year ago. G o v e rn m en t p rice su pp ort lev els for to b a cco are 12 p e rc e n t h ig h er than a year a go so that cash receip ts to fanners from sales o f the 1981 cr o p are u p about 25 p ercent. T h e form u la for gov ern m en t p rice su p ports for e lig ib le to b a c co requ ires that the support p rice for the 1982 cro p b e in creased 11 p e rce n t from the 1981 lev el. T h u s, w ith an average to b a cco harvest, cash returns from to b a cco sales w ill rise 11 percent. FOO D OU TLO O K L e d b y sizable gains in the ou tp ut o f fresh veg eta b le s, b e e f and veal, overall fo o d su p p lies this year are forecast su fficien tly h igh to p ro v id e a 1 p ercen t in crease in con su m p tion p er capita. In creases are also in p rosp ect for cereals and bakery p rod u cts, dairy p rod u cts, and fresh fruit (table 2). S om e re d u c tion in con su m p tion p er capita is in p ro s p e ct for pork, egg s, and p ro c e ss e d fruit and v eg eta b les. C ru d e fo o d s tu ff p rices are not e x p e cte d to rise m u ch at the farm le v e l this year, on ly abou t 1 p ercen t Digitized for28 FRASER to 4 p ercen t. As in d ica ted earlier, b e e f, poultry, dairy p ro d u ct and e g g p rice increases w ill b e m in im al. I f h o g p ro d u ctio n d e clin e s relative to d e m a n d as ex p e cte d , pork p rice increases m ay e q u a l the inflation rate. F resh fruit p rices m ay also rise at the inflation rate. Farm le v e l p rice s for m ost cereals and dry e d ib le b ean s, h o w e v e r , w ill b e lo w e r than in 1981. C o n se q u e n tly , m ost o f the in crease in fo o d costs this year w ill o c c u r in the p ro c e ss in g and m arketing sector. T h e average p rice in crease for fo o d at grocery stores is estim ated to b e about 6 p e rce n t (table 3). N o m ajor p rice ch a n ges are forecast for any fo o d grou p. P rice increases for pork, p ro c e ss e d fruit and v e g eta b les, and oth er p rep ared food s such as cereals and bak ery p rod u cts, w h e re m ost o f the costs rep re sent p ro ce ss in g and m arketing services, w ill lik e ly app roach the rate o f inflation. SUMMARY T h e U.S. D ep artm en t o f A gricu ltu re forecasts that there w ill b e n o im p ro v e m e n t in n et farm in com es FEDERAL RESERVE BANK OF ST. LOUIS JANUARY 1982 in 1982. E xp ectation s are for risin g fo o d su p p lies and relatively m oderate fo o d p rice increases. receip ts to farmers c o u ld b e d o w n from $1 b illio n A bu n d an t quantities of m ost farm p rodu cts are available for d o m e stic use and exp ort fo llo w in g the reco rd 1981 cr o p harvest. W h ile exports are e x p e cte d to rise further, su p p lies o f m ost crop s and livestock p rod u cts w ill b e su fficien t to lim it p rice increases to less than the rate o f inflation. Cash b e greater. to $3 b illio n and the d e c lin e in n et in co m e co u ld Larger quantities o f fo o d are in p ro sp e ct for c o n sum ers, and the fo o d co m p o n e n t o f the con su m er p rice in d ex w ill lik e ly increase at less than the rate o f inflation. 29