Public spending, prívate investment, and growth in O.E.C.D. countries: an empirical investigation

P.S.P.l/10-2-96 I I I ENCUENTRO DE ECONOMIA PUBLICA Depa amen o de Teo ía Económica y Economía Polí ica Uni e sidad de Se illa Se illa 9, 10 de eb e o de 1995 PONENCIA Public spending, p í a e in es men , and g ow h in O.E.C.D. coun ies: an empi ical in es iga ion. Isabel ARGIMON MAZA Banco de España José Manuel GONZÁLEZ-PÁRAMO Banco de España. Depa amen o de Hacienda Pública Uni e sidad Complu ense José Ma ía ROLDAN Eu opean Mone a y Ins i u e Summa y: This pape is conce ned wi h he empi ical ela ionship be ween go e nmen spending and p í a e in es men . A panel o 14 OECD coun ies is used. We p esen e idence which sugges s he exis ence o a signi ican c owding-in e ec o p í a e in es men by public in es men , h ough he posi i e impac o in as uc u e on p í a e in es men p oduc i i y. Mo eo e , go e nmen consump ion appea s o c owd ou p í a e in es men . The implica ions o hese esul e a e o o emos impo ance when i comes o iscal consolida ion. Dé ici educ ions enginee ed h ough cu e in public in es men could se e ely impinge upon p í a e capi al accumula ion and g ow h p ospec e. Resumen: Es e abajo analiza la elación empí ica exis en e en e el gas o público y la in e sión p i ada en un panel de 14 países pe enecien es a la OCDE. La e idencia p esen ada sugie e la exis encia de un signi ica i o e ec o "c owding-in" de la in e sión pública, que ope a a a és de la complemen a iedad en e las in aes uc u as y la in e sión p i ada. Po o a pa e, el consumo público pa ece eje ce un "e ec o expulsión" de la in e sión p i ada. La implicaciones de es os esul ados son de g an ascendencia en ma e ia de consolidación iscal. Reducciones del dé ici conseguidas median e eco es en la in e sión pública pod ían daña se e amen e la acumulación de capi al y las pe spec i as de c ecimien o. 1. INTRODUCTION: FISCAL POLICY, GROWTH, AND CROWDING OUT In his pape we in es íga e he ela ionship be ween p í a e in es men and go e nmen spending. Empi ical e idence is b ough o bea upon wo closely ela ed ques ions which lie a he co e o he deba e on he mac oeconomic e ec s o iscal policy: 1) Is public spending p oduc i e? 2) To wha ex en , i any, does go e nmen spending subs i u e o p í a e in es men ? The i s ques ion has s imula ed a conside able amoun o esea ch sin ce Rubinson (1977) and Ram (1986) ound a posi i e empi ical ela ionship be ween go e nmen size and GDP g ow h. Al hough his associa ion has al so been ob ained in mo e ecen wo k (see Lin, 1994a), a numbe o pape s ha e iden i ied an in e se associa ion be ween go e nmen spending and ou pu g ow h (e.g. G ossman, 1988, Mallow, 1986, Peden and B adley, 1989, and G ie and Tullock, 1989). In a ca e ul and comp ehensi e analysis o his issue, Dow ick (1993) concludes ha go e nmen size does no appea o ha e a sys ema ic and signi ican e ec on g ow h. This li e a u e has been c i icized on h ee main g ounds. Fi s , as shown by Hsieh and Lai (1994) and Dow ick (1993), he sign o he associa ion is qui e sensi i e o he choice o he sample o coun ies, he pe iod unde s udy, he econome ic echniques and he exis ence o measu emen p oblems. Second, as di e en ca ego ies o spending a e well documen ed o ha e di e se economic e ec s, analyses igno ing his ac a e no easy o in e p e . Eas e ly and Rebelo (1993) and Lin (1994b), o example, show ha spending ca ego ies ha p omo e human o physical capi al accumula ion a e posi i ely associa ed wi h g ow h, while o he spending i ems ha e nega i e o neu al e ec s. Finally, he single equa ion indings on he go e nmen - size/GDP- g ow h link could be in alid when his ela ionship uns h ough indi ec channels, i.e. p í a e in es men . These empi ical sho comings ha e a di ec bea ing on he second ques ion. Since capi al accumula ion is he engine o ou pu g ow h, any c owding-ou e ec s o p í a e in es men by public spending impinge upon p oduc ion expansión and wel a e p ospec e. The bulk o he empi ical li e a u e inds a signi ican ly nega i e e ec o public consump ion on g ow h while he e ec s o public in es men a e ound o be posi i e al hough less obus (Ba o, 1991; G ie and Tullock, 1989; Eas e ly and Rebelo, 1993). To wha ex en can hese esul e be aced o a c owding ou o p í a e in es men ? Aschaue (1989b) inds ha he di ec c owding-ou e ec o public in es men is ou weighed by a di ec c owding-in e ec associa ed wi h he ole o public capi al as a p oduc i e inpu and i s complemen an e0 wi h p í a e capi al (see Aschaue , 1989a). The e idence ga he ed in E enbu g (1993), who es ima es a simple mac omodel wi h a ional expec a ions, and in E enbu g and Woha (1995), ein o ces his conclusión, while he esul e in Bai am and Wa d (1993) suppo he c owding-ou hypo hesis. As o he impac o go e nmen consump ion, Aschaue (1989b) concludes ha i s c owding-ou e ec on p í a e in es men ca ies only a ma ginal explana o y powe , a esul ha could be in e p e ed as an indica ion ha public consump ion is a cióse subs i u e o p í a e consump ion. Howe e , ecen e idence ga he ed by Ka as (1994) o ce ully sugges e ha p í a e and go e nmen consump ion a e bes desc ibed as complemen a y (o un ela ed) goods(2), a ea u e ha ein o ces he c owding-ou e ec o p í a e in es men . The es o he pape is o ganized as ollows. Sec ion II ou lines he heo e ical a gumen e behind he c owding-ou hypo hesis. Fo exposi ional pu poses, we use a simple o e lapping-gene a ions model in which public and p í a e capi al a e complemen e, whe eas public and p í a e consump ion a e independen Sec ion III e alua es he impac on p í a e in es men and p í a e p oduc i i y o public spending. The empi ical esul e ob ained wi h a panel o 14 OECD coun ies a e p esen ed and b ie ly discussed. Finally, Sec ion IV d aws he main conclusions. 2. PRIVATE INVESTMENT AND PUBLIC EXPENDITURE The equilib ium app oach de eloped in Aschaue (1988) and Aschaue and G eenwood (1985), among o he s, assumes a compe i i e economy popula ed by a ional, iden ical, in ini ely li ed indi iduáis. In his con ex , he gene al equilib ium ela ionship be ween public spending and p í a e in es men may be exp essed in he ollowing wo equa ions: (1 Public and p í a e capi al a e "complemen a y" when he ma ginal p oduc i i y o he p í a e capi al inc eases as he quan i y o public capi al inc eases. (2) I public consump ion inc eases he ma ginal u ili y o p í a e consump ion, bo h a e said o be "complemen a y", and ice e sa. 3 i = i( k,ig,cg), i,>0, z2<0, z3^0 (1) k =A(Up, u<Q. ¿& (2) whe e i is p í a e spending, k is he ma ginal p oduc o capi al, ig and cg a e public in es men and public consump ion, espec i ely, k is p í a e capi al and kg is public capi al. Along neoclassical lines, changes in in es men -which a e desc ibed by he pa ial de i a i es o he abo e unc ions- a e he esul o in e empo al smoo hing o consump ion by p í a e agen s. Aschaue s discussion o he c owding-ou issue is heiuis ic, due o he complexi y o his analy ic amewo k. None heless, his quali a i e conclusions ca y o e in much simple neoclassical models. Suppose ha ou economy is popula ed by o e lapping gene a ions o equal size. Each household li es o wo pe iods. Households ea n labou income only in he i s pe iod. Labou supply is ixed. Consump ion o a ep esen a i e household when young in ime is: c,1 = w, - , - s, , (3) whe eas consump ion when oíd a ime +1 can be w i en as: <Á=í,(K+1), (4) whe e w, is he wage a e, s is sa ings, +I is he in e es a e in he second pe iod and , is a ax le ied on he young. The go e nmen uses ax e enues o inance public consump ion o a public good na u e -which en e s household's u ili y unc ion- and public in es men -which is a p oduc i e inpu in p í a e p oduc ion. Suppose ha he u ili y unc ion o a ep esen a i e agen is: U = hLnc¡ + (l- yMC +_L[5I/ic,2+1 + (1-8)1^, (5) whe e cg is public consump ion, 6 ep esen e ela i e p e e ence o p í a e consump ion and p is he a e o ime p e e ence. Maximiza ion o (5) subjec o (3) and (4) yields he sa ings unc ion: s = o(w, - ), a=_L . (6) Fo simplici y, assume ha bo h p í a e and public capi al ully dep ecía e in e e y pe iod. The p oduc ion unc ion in pe wo ke e ms is: y = Ai,ai¡ , a+p<l , (7) whe e i is p í a e capi al (and p í a e in es men ) and ig is public capi al (and public in es men ). Compe i ion in ac o ma ke e implies: l4 =CL4/;%p -<¿ (8) w -(l-a)Ai?i¡ -(l-c y . (9) Capi al a ailable o p oduc ion in +1 equals sa ings o he young in . This gi es he momen a y equilib ium condi ion: ^ = 0(1 -a)M?i¡ ~ o , . (10) This equa ion, oge he wi h he go e nmen budge cons ain + i, • CID de ine he equilib ium o he economy. A pe manen ax- inanced inc ease in public consump ion educes bo h i s - pe iod p í a e consump ion and sa ings, and hus in es men in he nex pe iod. In he long- un, a e d opping ime subsc ip e, we ob ain: 5 di_ de l-oa(l-cO¿/a-1// < O , (12) g p o ided ha he economy is no oo a away om he golden ule (no e ha when =0, by equa ion 8 he denomina o in 12 educes o l-o(l-a)). As his c owding-ou e ec gene a ed by public consump ion is associa ed wi h he e ec o axes upon sa ings, a ax- ínanced inc ease in public in es men mus also p oduce a di ec c owding-ou e ec o p í a e in es men Howe e , public in es men may ha e an addi ional e ec p í a e in es men is c owded-in whene e public capi al aises he p oduc i i y o p í a e capi al. In ac , i he ini ial le el o public capi al we e well below i s op imal (i.e. ou pu -maximizing) le el, he c owding-in e ec would ou weigh he nega i e in luence o axes upon sa ings. Mo e gene ally, he s eady-s a e e ec o an inc ease in public in es men upon p í a e capi al accumula ion is: In he mo e elabó a e Aschaue model his ambigui y eme ges as well. The di e ences a e jus a ma e o de ail. When public and p í a e capi al a e equally p oduc i e, an inc ease in public in es men c owds ou an equi alen amoun o p í a e in es men , wi h no e ec s upon weal h and consump ion. Howe e , an addi ional c owding-ou e ec would a ise i public in es men we e mo e p oduc i e: as li e ime weal h inc eases so does consump ion oday. These wo sho - un e ec s could be compensa ed by he c owding-in e ec which ope a es h ough an inc ease in p í a e capi al p oduc i i y. As o he consequences o an inc ease in public consump ion, Aschaue admi s he possibili y ha i s e ec could be nil. I public consump ion is a pe ec subs i u e o p í a e consump ion o he ma ginal p opensi y o consume ou o weal h emains cons an o e ime o e e y agen , p í a e in es men changes would no be needed o smoo h ou consump ion. None o hese assump ions a e made in ou simple exposi ional model(3). (' These models a e no di ec ly compa able. No e ha we a e assuming ha public and p í a e consump ion a e nei he complemen a y ño subs i u es. On he o he hand, wi h income acc uing only in he i s pe iod and a ixed p opensi y o sa e, c owding ou ob ains e en i he ma ginal p opensi y o consume emains cons an . In such case, which implies = p, p í a e in es men is c owded-ou by [l-a(l-p) ]"' uni s pe di a[l-P(l-aMí°í/"'] < (13) l-oaO-cOA--1// > 6 3. DATA, METHODOLOGY AND EMPIRICAL RESULTS The empi ical analysis is ca ied ou using he annual ime-se ies-c oss-sec ion da a om Summe s and Hes on (1991). Only he ou een OECD coun ies wi h a ailable da a on in as uc u e capi al we e included o he pe iod 1979-88(4), which esul ed in an unbalanced panel da a se Da a de ini ions a e gi en in he Appendix. Ou analysis o he c owding-ou hypo hesis is based on he es ima ion o he ollowing pai o equa ions: +^ o**+Vln** +4* (15) whe e subsc ip s e e o ime and j o coun y, i and ig a e, espec i ely, p í a e and public in es men in ela ion o p í a e p oduc i e capi al s ock (p í a e and public in es men a es), cg is public consump ion, in ela ion o p í a e p oduc i e capi al s ock, k is ma ginal p í a e capi al p oduc i i y (which is p oxied by he a io o g oss ope a ing su plus o p í a e p oduc i e capi al s ock), k is he s ock o p í a e p oduc i e capi al and kg is in as uc u e capi al, so ha no esiden ial in es men is in ol ed(5). We would expec 2^ o be posi i e, and bo h a3 and a4 o be non-posi i e unde he c owding-ou hypo hesis. On he o he hand, gi en he s anda d assump ion o dec easing ma ginal p oduc i i y, b2 is expec ed o be nega i e, and b3 posi i e i in as uc u es and p í a e p oduc i e capi al a e complemen e. uni o addi ional go e nmen consump ion spending. (4) The coun ies and he pe iods a e he ollowing ones: Aus alia (1979-87), Aus ia (1979-87), Belgium (1979-87), Canadá (1979-88), Denma k (1979-88), Fede al Republic o Ge many (1979-88), Finland (1979-87), F ance (1979-88), I eland (1979-87), No way (1980-86), Spain (1979-86), Sweden (1979-87), Uni ed Kingdom (1979-87) and Uni ed S a es (1980-87). (5) Un o una ely, ou da a do no allow p i a e and public in as uc u e o be sepa a ed ou . I may be a gued ha bo h ypes o in as uc u e should ha e oughly he same e ec s on p oduc i i y. In any case, his ac has o be bo ne in mind when in e p e ing he esul s. 7 The econome ic analysis o he c owding-ou issue equi es he use o echniques well sui ed o deal wi h coun y he e ogenei y. Le us i s e e o he mos ec ic i e e sión o (14) and (15). Assume ha a^, = am and b^, = bm, Vj V : ha is, he esponse o p í a e in es men and p í a e capi al p oduc i i y o he igh -hand side a iables does no change o e ime ño ac oss coun ies. The esul e o such a es ic i e model a e p esen ed in column (1) o Table 1. In columns (2) o (6) we elax hese es ic ions so as o ake in o accoun he he e ogenei y o he da a. In column (2) we epo he esul e o le ing a, and b, change ac oss coun ies while aking he emaining coe íícien e o be he same ac oss coun ies. The es ima es co espond o a íxed e ec s model wi h coun y dummies, and a e he e o e a wi hin es íma e. The dummy coe icien e a e no epo ed. In o de o choose be ween he mos es ic i e model and he wi hin es ima es, an F es is ca ied ou , whose esul e a e epo ed in he lines headed by F^. In column (3) we epo he es ima es o he model in i s di e ences. Such an app oach no only akes in o accoun íxed e ec s, which d op ou when he model is speci ied in i s di e ences, bu may be ega ded as a way o dealing wi h non-s a iona i y when his p oblem is p esen í Column (4) epo e he esul e ob ained om a andom e ec s model. The main di e ence wi h he wi hin es ima o is ha he coun y speci ic e ec s a e now ea ed as andom and unco ela ed wi h he eg esso s. A Hausman es is cons uc ed o check o his es ic ion, wi h he esul e shown in he ow labeled %2 in able l^. (6) This es is o mula ed as: SSRR - SSRU/(ku - k ) _ p SSRU/(T _ o) ~ ( e -M. (T- u) whe e U and R s and o he un es ic ed and he es ic ed model, espec i ely, SSR is he sum o squa ed esiduals, k is he numbe o es ima ed coe icien s and T is he o al numbe o obse a ions. ^ The Hausman es is o mula ed as: whe e pw is he ec o o he k wi hin-g oup es ima es, PGLS is he co esponding gene alized leas squa es es íma e o he andom e ec s model, and Vw and VGLS a e hei espec i e es ima ed a iance-co a iance ma ices. Unde he nuil hypo hesis o no có ela ion ( he andom e ec s 8 B. INSTRUMENTAL VARIABLES ESTIMATES TABLE Al. CROWDING-OUT EFFECTS INSTRUMENTAL VARIABLE ESTIMATION (1979-88) (1) (2) (3) li 0.80 (4.92)*' 0.74 (4.68)" 0.80 (5.00)** eq(l) -0.07 (0.13) -0.62 (1.36) -1.02 (2.28)** -1.05 (2 66V* eq (z) ln(k/l)j -0.56 (A A<C ** (4.ÜO) 1.07 (6.29)'* o eq (1) eq (2) 0.024 0.041 0.024 0.024 xj 112 112 112 a: s anda d e o o he eg ession The ins umen s used a e aA, c,., and in equa ion (1) and ln (k/1),., and ln (kg/1), in equa ion (2). Coun y dummies whe e included in bo h equa ions as ins umen s. 15 • :; ¿ : ." i w BIBLIOTECA C. PRODUCTION FUNCTION EQUATION The es ima ed equa ion is gi en by ln (yll) = a + (1 -a-p) n[kll + p ln[£//] whe e so ha again he es ic ion o cons an e ums o scale has been imposed(10). The esul e a e p esen ed in able A2. The inclusión o a ime end, , aiming a cap u ing echnological change, esul e in he es ima es shown in columns (2), (4), and (6). The andom e ec s model seems o be he one ha i e be e he da a. The posi i e and s a is ically signi ícan sign o he in as uc u e a iable seems o be p esen in all models, excep unde column (4). Howe e , in ha case, he end a iable is no s a is ically signi ícan ei he , so ha he speci ica ion seems o be ejec ed by he da a. The ac ha he andom e ec s model canno be ejec ed s a is ically may be a e lec ion o he ac ha he echnological he e ogenei y among he di e en coun ies is no di ec ly ela ed o he di e en ini ial s a es o he echnology. In ac , i could be iden ical o all coun ies, and he indi idual e ec s could be ela ed o o he aspec o o he economic en i onmen . Looking a columns (5) and (6) i may be concluded ha , al hough he ou pu elas ici y wi h espec o in as uc u e is smalle ha he elas ici y wi h espec o p i a e capi al, he esul e imply again ha he e is c owding in o public in es men , h ough i s posi i e impac on p i a e p oduc i i y. This esul ein o ces hose p esen ed in Table 1. The es s o he hypo hesis o cons an e ums o scale a e no comple ely sa is ac o y, e en when he wi hin es ima o is conside ed. Howe e , gi en he speci ica ion o he p oduc i i y unc ion i seems mo e easonable o ha e cons an e u ns o scale. 16 REFERENCES ASCHAUER, D.A. (1988). "The equilib ium app oach o iscal policy"; Jou nal o Money, C edi and Banking, 20, pp. 41-62. ASCHAUER, D.A. (1989a). "Is Public Expendi u e P oduc i e?"; Jou nal o Mone a y Economics, 23, pp. 177-200. ASCHAUER, D.A. (1989b). "Does public capi al c owd-ou p i a e capi al?"; Jou nal o Mone a y Economics, 24, pp. 171-188. ASCHAUER, DA. and GREENWOOD, J. (1985). "Mac oeconomic e ec s o iscal policy"; Ca negie-Roches e con e ence se ies on public policy, 23, pp. 91-138. BAIRAM, E. and B. WARD (1993). "The ex e nali y e ec o go e nmen expendi u e on in es men in OECD coun ies"; Applied Economics, 25, pp. 711-716. BARRO, R. (1991). "A c oss-coun y s udy o g ow h, sa ing and go e nmen "; en D. Be nheim y J. Sho en (eds.): Na ional sa ing and economic pe o mance, NBER, Uni e si y o Chicago P ess, Chicago, Illinois, pp. 271-301. DOWRICK, S. (1993). "Go e nmen consump ion: I s e ec s on p oduc i i y g ow h and in es men "; in N. Gemmell (ed.): The g ow h o he public sec o . Theo ies and e idence, Edwa d Elga , Alde sho , pp. 136-152. EASTERLY, W. and REBELO, S. (1993). "Fiscal policy and economic g ow h: An empi ical in es iga ion"; 32, Jou nal o Mone a y Economics, pp. 458-493. ERENBURG, S.J. (1993). "The eal e ec s o public in es men on p i a e in es men "; Applied Economics, 25, pp. 831-837. ERENBURG, S.J. and M.E. WOHAR (1995). "Public and p i a e in es men : a e he e causal linkages?"; Jou nal o Mac oeconomics, Vol. 17, 1. GRIER, K.G. and TULLOCK, G. (1989). "An empi ical analysis o c oss-na ional economic g ow h, 1951-1980"; Jou nal o Mone a y Economics, 24, pp. 259-279. 17 GROSSMAN, P.J. (1988). "Go e nmen and economic g ow h: a non linea ela ionship"; Public Choice, pp. 193-200. HSIEH, E. and LAI, K.S. (1994): "Go e nmen spending and economic g ow h: he G-7 expe ience"; Applied Economics, 26, pp. 535-542. KARRAS, G. (1994). "Go e nmen spending and p i a e consump ion: Some in e na ional e idence"; Jou nal o Money, C edi , and Banking, 26, pp. 9-22. LIN, S. (1994a). "Go e nmen spending and economic g ow h"; Applied Economics, 26, pp. 83-94. LIN, S. (1994b). "Alloca ion o go e nmen spending and economic g ow h"; Economic No es by Mon e dei Paschi di Siena, Vol. 23, 1. pp. 130-141. MALLOW, M.L. (1986). "P i a e sec o sh inkage and he g ow h o indus ialized coun ies"; Public Choice, 49, pp. 143-154. PEDEN, E. and BRADLEY, M. (1989). "Go e nmen size, p oduc i i y, and economic g ow h"; Public Choice, 61, pp. 229-245. RAM, R. (1986). "Go e nmen size and economic g ow h: A new amewo k and some e idence om c oss-sec ion and ime-se ies da a"; Ame ican Economic Re iew, 76, pp. 191-203. RUBINSON, R. (1977). "Dependency, Go e nmen e enue, and economic g ow h"; S udies in Compa a i e In e na ional De elopmen , 12, pp. 3-28. SUMMERS, R. and HESTON, A. (1991). "The Penn Wo ld Table (ma k 5): an expanded se o in e na ional compa isons, 1950-1988"; Qua e ly Jou nal o Economics, May. 18