The International Effects of Government Spending Composition

Transcription

1 W/5/4 The Inernaional Effecs of Governmen Spending Composiion Giovanni Ganelli

2 25 Inernaional Moneary Fund W/5/4 IMF Working aper Fiscal Affairs Deparmen The Inernaional Effecs of Governmen Spending Composiion repared by Giovanni Ganelli Auhorized for disribuion by Manmohan S. Kumar January 25 Absrac This Working aper should no be repored as represening he views of he IMF. The views expressed in his Working aper are hose of he auhor(s) and do no necessarily represen hose of he IMF or IMF policy. Working apers describe research in progress by he auhor(s) and are published o elici commens and o furher debae. This paper helps resolve a paradox in he lieraure, noiced by Alesina and eroi (995), which is ha, alhough governmen employmen is an imporan componen of public spending, he debae on he effecs of fiscal policy focuses almos exclusively on shocks o non-wage governmen consumpion. We incorporae he disincion beween spending for governmen employmen and spending for non-wage governmen consumpion in a new open economy macroeconomics model. Our resuls show ha a permanen reducion in public employmen in one counry reduces relaive privae consumpion and appreciaes he domesic exchange rae if i is mached by a reducion in axes. When he reducion in public employmen is used o finance increased non-wage governmen consumpion, he macroeconomic effecs resuls are ambiguous, and are affeced by he iniial level of he public wage bill. JEL Classificaion Numbers: E62, F4 Keywords: ublic spending composiion, fiscal policy, open economy Auhor(s) Address: gganelli@imf.org I would like o hank hilip Lane for his useful commens.

3 - 2 - Conens age I. Inroducion...3 II. The Model...4 A. The Iniial Seady Sae...7 B. Shor-Run Nominal Rigidiies...7 C. Log-linearizaion...7 III. The Effecs of Governmen Spending Composiion...8 A. Graphical Soluion...8 B. Effecs of a Reducion in Governmen Employmen...2 C. Long-Run Effecs...4 D. Effecs of a Shif in Governmen Spending Composiion...6 IV. Conclusions...7 References...8 Tables. The Log-linearized Domesic Economy...8 Figures. The Two Equilibrium Relaionships Effecs of a Reducion in Domesic Governmen Employmen...2

4 - 3 - I. INTRODUCTION Boh in ad hoc and in general equilibrium macroeconomic models governmen spending is generally inroduced, explicily or implicily, only as non-wage governmen consumpion of goods and services. 2 As Alesina and eroi (995) noice, however, spending for public employmen is an imporan componen of governmen spending, while spending for non-wage public consumpion ends o show he leas significan dynamics, ye paradoxically, governmen [non wage] consumpion ypically receives he mos aenion in he academic debae on he macroeconomic effecs of fiscal policy (Alesina and eroi, 995, p. 224). 3 In his paper we help resolve his paradox of he lieraure, by analyzing he effecs of governmen spending composiion in a wo-counry general equilibrium sicky-price model. Finn (998) has incorporaed he disincion beween spending for public employmen and for non-wage public consumpion in a closed economy, flexible-price Real Business Cycle (RBC) model. In ha conex, she showed how failing o pay aenion o such a disincion can lead o overesimaion of he governmen s impac on he economy cycle. Our analysis differs from Finn s on a number of poins (discussed in more deail in Secion III(C)). Mos noably, we use a wo-counry model belonging o he so-called new open economy macroeconomics srand of he lieraure raher han a RBC approach (see Obsfeld and Rogoff, 995, 996; and Lane, 2). This allows us o sudy he impac of changes in he composiion of public spending, no only on he long-run full-employmen equilibrium, bu also on he shor run, in which oupu is demand-deermined due o he presence of nominal rigidiies. Furhermore, he open economy dimension allows an analysis of he effecs of public spending composiion on variables, such as he exchange rae and he curren accoun. Lane and eroi (23) show in an open-economy reduced-form model how he composiion of public spending ineracs wih he exchange rae regime and has an empirical effec on macroeconomic oucomes. Since he new open economy macroeconomics is rapidly emerging as a new paradigm for he analysis of macroeconomic inerdependence, our conribuion also aims o provide furher insighs on he role of fiscal policy in his new framework, in addiion o hose ha can be derived from ad hoc models (Lane and Ganelli, 23, p. 329, sress he imporance of moving he research agenda in his direcion). Our analysis shows ha a permanen reducion in domesic governmen employmen ha is used o finance a reducion in general axaion implies an immediae increase in domesic privae consumpion compared o foreign. This resul is mainly due o he fac ha agens anicipae ha hey will have a lower lifeime ax burden, including in he new long-run equilibrium when he reducion in public secor employmen will be compensaed by increased employmen in he privae secor. Furhermore, under our assumpions, he secoral shif in employmen implies a 2 Two noeworhy excepions are Finn (998), and Lane and eroi (23), which are discussed below. 3 Ialics added by auhor.

5 - 4 - more producive use of domesic resources, herefore easing curren accoun pressures and allowing a more appreciaed shor-run exchange rae. When he reducion in public employmen is used o finance an increase in governmen nonwage consumpion, he macroeconomic effecs are ambiguous since in ha case domesic agens do no gain from he reducion in he ax burden, bu hey benefi from he expansion in demand implied by higher public consumpion. Our analyical resuls show ha, following such policy, a sronger pressure for reducion in domesic relaive privae consumpion emerges in cases in which he iniial share of privae income derived from employmen in he public secor is larger. The paper is organized as follows: he nex secion inroduces he model, Secion III discusses he effecs of wo macroeconomic policies wih he help of a simple graphical apparaus; and Secion IV concludes. II. THE MODEL There are wo counries in he world, which we label Home and Foreign. The world is populaed by a coninuum of infiniely-lived agens. Toal world populaion is normalized o. Home agens are on he inerval [, n], while foreign agens are on he inerval [ n, ]. A measure of firms also exis in he world. Each firm has some monopolisic power in he producion of a differeniaed good and n firms are based in he domesic counry while -n firms are based in he foreign counry. Agens derive uiliy from consumpion of a composie good, leisure and real balances. 4 Home agens supply labor in he perfecly compeiive domesic labor marke, receive shares of profis from domesic firms, and pay lump-sum axes. 5 The domesic agen s opimizaion problem can, herefore be expressed as σ M k U = β C + L, () s= subjec o he real budge consrain σ s σ s 2 [ χlog ] s s σ s 2 M M W B + = ( + π + r) B + + L C τ +, (2) where β is he discoun facor, σ is he ineremporal elasiciy of consumpion and χ and k are posiive parameers. The only inernaionally raded bond is a riskless real bond denominaed in 4 Since he model is symmeric, we will only presen he opimizaion problem of he domesic agen. The foreign agen opimizaion problem is analogous. 5 We assume ha domesic profis are uniformly disribued across domesic agens (see Hau, 2, for a similar assumpion).

6 - 5 - erms of he composie consumpion good which we denoe wih B ; r is he real ineres rae on his bond beween and. B also accouns for inernaional shifs in wealh. M denoes nominal money balances held a he beginning of period. 6 Agens also supply labor in a perfecly compeiive labor marke, receive profis shares from domesic firms Π, and pay lump-sum axes τ. C is a privae consumpion index and he corresponding price index (boh described more in deail below), while W denoes he nominal wage. The firs wo argumens of he uiliy funcion are a real privae consumpion index (see equaion (4)) and real money balances, while L is oal labor supply and he erm k L 2 capures he negaive effec on uiliy deriving from leisure loss. Equilibrium in he labor 2 s marke implies where L and G L L L = +, (3) G L are he amoun of labor demanded, respecively, by he domesic privae and public secors (see also Finn, 998). The privae consumpion index aggregaes across he differeniaed goods produced by all firms in he economy boh a home and abroad. Firms enjoy a cerain degree of monopolisic compeiion, specified by he magniude of he parameer θ >. Denoing wih z a generic represenaive firm, wih cz ( ) he consumpion of he differeniaed oupu of his firm by he represenaive agen, and wih pz ( ) represening he domesic currency price of his oupu, he aggregae privae consumpion index and he corresponding price index ake he following forms: θ θ θ θ C = [ c( z) dz], (4) ( ) [ ( ) θ ] = p z dz θ. (5) We assume ha he law of one price holds. This implies ha he home and foreign consumer prices indexes are linked by he purchasing power pariy () relaionship = E, where E is he nominal exchange rae, defined as he price of he foreign currency in erms of domesic currency. Maximizaion of equaion () subjec o equaion (2) yields he following firs-order condiions: 6 Noe ha we adop Obsfeld and Rogoff (996) iming convenion, M herefore denoes money beween period and period +, while B denoes bonds beween period and.

7 - 6 - C C M + σ = [ β ( + r )], (6) + + i =, (7) σ + χ C i+ C σ W = kl, (8) where i + is he nominal ineres rae beween and +. Equaions (6) o (8) are, respecively, an Euler equaion, a money demand equaion, and a labor-leisure rade-off equaion. In order o keep our seup as simple as possible, while a he same ime allowing a meaningful analysis of he effecs of composiion of fiscal policy, we assume ha labor is he only facor of producion used by firms, according o he simple producion funcion Y = L. (9) We also assume ha he labor supplied o he public secor is no used for producive purposes, bu he public secor pays he same wage deermined in he perfecly compeiive domesic labor marke. Furhermore, domesic agens are indifferen beween working for he domesic governmen and working for domesic firms. This implies he following relaionship beween he price of a differeniaed produc and he nominal wage: θ p( z) = W. () θ The real-erm budge consrain faced by he governmen in each period is given by: W M M L + G = τ +. () G Equaion () expresses he fac ha in every period he governmen can use lump-sum axes and seignorage in order o finance public spending, boh for non-wage governmen consumpion and for governmen employmen. Governmen consumpion is defined as an aggregae across he differeniaed goods z, wih he same elasiciy of subsiuion, and is herefore analogous o he privae consumpion aggregae defined in equaion (4). In he framework of his model he world demand for he represenaive differeniaed produc z akes he form: pz ( ) θ W W Y( z) = [ ] ( C + G ), (2) W W where C is world privae consumpion and G is world public consumpion.

8 - 7 - A. The Iniial Seady Sae We consider an iniial seady sae where governmen spending for goods and ne foreign asses are zero for boh counries, we accordingly denoe his seady sae wih a zero subscrip ( G = G = B = B = ). Noice ha in our iniial seady sae governmen spending for employmen is no necessarily zero. This will allow us o carry ou a meaningful analysis of he macroeconomic effecs of reducions in he size of public employmen. B. Shor-Run Nominal Rigidiies We inroduce shor-run nominal rigidiies in he form of one period prese prices in he currency of he producer, as in Obsfeld and Rogoff (995, 996), and Ganelli (23, 24). Domesiccurrency prices of domesic goods and foreign-currency prices of foreign goods are herefore rigid in he period in which he shock his, bu hey are free o adjus o heir new seady-sae level in he period afer he shock, when he economy reaches is new full-employmen seady sae. In erms of noaion, in he log-linearized version of he model, we will denoe shor-run variables using ildes, and long-run variables using has. Our assumpion on shor-run rigidiy will imply ha ph ( ) = p( f) = in he log-linearized version of he model. C. Log-linearizaion Table summarizes he log-linearizaion of he model around he iniial seady sae defined above. Given he symmery of he model, only he domesic equaions, he equaion, and he world goods marke equilibrium equaion are repored. Equaions (3) o (24) are loglinearized versions of he price index, he Euler equaion, he shor-and long-run money demand equaions, he labor-leisure supply equaion, he equilibrium condiion in he domesic labor marke, he demand equaion, he shor-and long-run curren accoun equaions, he governmen budge consrain (excluding seignorage), he, and he world goods marke equilibrium G L condiion. In Table, L and L are, respecively, he iniial share of labor used by he privae L G WL and public secors as shares of he oal iniial labor supply, and H = is a measure of he Y iniial public wage bill as a share of GD.

9 - 8 - Table. The Log-linearized Domesic Economy p = np ( h) + ( n)[ e+ p ( f)] (3) c = c + σ ( β ) r (4) r p p (5) m p = c σ + δ δ m p = c (6) σ l = p( z) p c (7) σ G p L g L (8) l = l + l L L l p = y = θ[ p p( h)] + c W + g W (9) b = y c ( n) e g (2) c = δ b + y+ p( h) p g (2) ( G H ) w+ l + g = τ (22) e = p p (23) pw, W W W l = y = c + g (24) Source: Auhor s calculaions. III. THE EFFECTS OF GOVERNMENT SENDING COMOSITION In his secion we will look a he inernaional macroeconomic effecs of wo differen policies. Under one policy, he domesic governmen asymmerically reduces he amoun of spending for public employmen, maching his reducion wih a reducion in axes. Under he second policy, he domesic governmen changes he composiion of is spending, by reducing public employmen and using his reducion o finance an increase of spending on non-wage public consumpion. During our analysis we keep moneary policy consan and we, herefore, absrac from seignorage. A. Graphical Soluion In order o sudy he effecs of he policy ha we wan o analyze, i is useful o combine he various equaions in order o obain wo basic relaionships, which can be used o provide a simple and inuiive graphical soluion of he model. Combining he equaion, he Euler equaion, he money demand equaion, and heir foreign counerpars, i is possible o derive he following relaionship beween shor-run relaive privae consumpion and he exchange rae:

10 - 9 - ( ) e = c c (25) σ Equaion (25) is sandard in a framework in which money demand depends on consumpion. I expresses he fac ha an increase in domesic consumpion relaive o foreign, by increasing money demand for a given level of he money supply, requires an appreciaion of he domesic currency (a decrease in e ) in order o resore equilibrium in he money marke. In Ganelli (24), his mechanism is referred o as he money demand effec. A second relaionship beween he exchange rae and relaive consumpion can be derived by combining he oher equaions of he model. In paricular, combining he domesic and foreign long-run demand equaions and curren accoun equaions, and assuming symmeric iniial L L shares of labor in he privae and public secors beween he wo counries ( = and L L G G L L = ) we can derive he following equaion: L L G L L g g [ ( ) c δ θ c = b + l l + ( l l )] ( g g ) n θ L L (26) Making use of he domesic and foreign labor-leisure equaions we can derive G L g g L l l = ( l θ l ) ( c c ) θl L + σ θl + L (27) Combining equaion (26) and equaion (27), we can derive an expression for long-run ne foreign asses as follows: G n n L L g g n b θ = K( c c ) + ( l l ) + ( g g ) δ δ θ L ( θl + L) δ (28) σθ ( L + L) + ( θ ) L where K = is a posiive consan, which depends on he degree of θ L + L monopolisic disorion in he economy, on he ineremporal rae of subsiuion of consumpion, and on he share of labor in he privae secor a he iniial seady sae. 7 In order o eliminae b we can make use of he shor-run curren accoun equaions. Subracing from equaion (2) is foreign counerpar and making use of he fac ha he hypohesis of one period price sickiness and he domesic and foreign demand equaions imply y y = θ e, we ge: 7 In order o prove ha K is posiive, i is enough o remember ha θ > by definiion.

11 - - b = ( n)[( θ ) e ( g g ) ( c c )] (29) Using equaion (4) and is foreign counerpar, i can be easily shown ha in his framework relaive consumpion follows a uni roo process (i.e., c c = c c ). 8 Combining equaion (28) wih equaion (29), and making use of he uni roo resul, we can eliminae b and obain a second relaionship beween sor-run consumpion and exchange rae: G K L L g g e θ = ( + )( c c ) + ( l l ) + ( g g ) θ δ δ θ L ( θl + L ) θ + ( ) δθ ( ) g g. (3) Unlike equaion (25), equaion (3) also depends on some policy variables such as governmen employmen and non-wage governmen consumpion. Since K is a posiive consan, equaion (3) expresses a posiive relaionship beween e and c c. Equaion (3) describes anoher mechanism a work in he model, which goes in he opposie direcion compared o he money demand effec. For a given governmen policy sance, an increase in domesic consumpion relaive o foreign worsens he curren accoun of he domesic counry, hus creaing a pressure for depreciaion of he domesic currency (an increase in e ). Figure provides a graphical represenaion of equaions (25) and (3) for given governmen policies (i.e., assuming no shocks o governmen employmen and non-wage consumpion in G boh counries: l = l G = g = g = g = g = ). The inuiion for he upward sloping shape of equaion (3) is he curren accoun effec, which we have jus described, while he money demand effec explains he downward slope of equaion (25). This graphical apparaus can be usefully employed in order o analyze he macroeconomic effecs of policy changes. 8 This is a well-known resul in his framework (see Obsfeld and Rogoff, 996).

12 - - Figure. The Two Equilibrium Relaionships e (3) c c (25) Source: Auhor s calculaions.

13 - 2 - B. Effecs of a Reducion in Governmen Employmen Figure 2 illusraes he effecs of a reducion in domesic governmen employmen compared o foreign. Since equaion (25) is no influenced by changes in governmen policies, he downward sloping curve is he same as in Figure. Figure 2. Effecs of a Reducion in Domesic Governmen Employmen (3) EQ pre-shock c c EQ pos-shock (25) Source: Auhor s calculaions.

14 - 3 - An analysis of equaion (3), on he oher hand, shows ha, for given non-wage governmen consumpion ( g = g = g = g = ), an asymmeric decrease in domesic governmen G G employmen relaive o foreign ( l l < ) will imply a lower level of e (a domesic appreciaion) for any given ( c c ). In graphical erms, such policy shifs equaion (3) righward (Figure 2). The new equilibrium, herefore, implies a higher level of relaive privae consumpion and a more appreciaed exchange rae. 9 An inuiion for he resul ha lowering governmen spending on employmen implies an increase in domesic consumpion relaive o foreign, can be found in he fac ha his policy reduces he amoun of privae resources used by he governmen. A secoral reallocaion of labor involving a shif ou of he public and ino he privae secor akes place. This implies a reducion of he ax burden on he economy, hus increasing he oal amoun of privae wealh available in he domesic economy, and allowing domesic agens o consume more compared o foreigner. I is imporan o noice ha i is he fuure componen of he policy ha impacs consumpion decisions. In oher words, if he reducion in public employmen was only emporary ( l l < wih l = l = ), here would be no consumpion reallocaion. This is because in he shor run, due o he nominal rigidiy, he reduced demand for labor in he public secor would no be compensaed by an increased demand in he privae secor. The aggregae consumpion behavior of he economy would no be affeced in ha case, because he gain associaed wih he reducion in he ax burden would be offse by he loss of labor income semming from public secor rerenchmen. If he reducion in governmen employmen is permanen, on he oher hand, agens anicipae ha hey will be able o benefi from lower axes even in he long run, when he economy will be in a full employmen equilibrium. Furhermore, in he long run he reducion in he ax burden will no be offse by lower income, because he reduced public employmen will be compensaed by an increase in employmen in he privae secor. Due o he presence of raional expecaions, domesic agens anicipae he lifeime increase in wealh and raise consumpion immediaely. The money demand effec, which we have described above, implies ha an appreciaion of he domesic exchange rae is necessary in order o reesablish equilibrium in he money marke. This is due o he fac ha he increase in relaive consumpion increases domesic money demand relaive o foreign for a given moneary policy sance. Addiionally, under our assumpions (see equaion (9)), he secoral shif in employmen from he public o he privae secor implies a more producive use of labor, herefore increasing he oal amoun of resources 9 This can also be shown by subsiuing equaion (25) direcly ino equaion (3) and noicing ha his implies a negaive relaionship beween g and c c l l.

15 - 4 - available in he domesic economy and reducing he impac of he curren accoun effec. This conribues o explain why he money demand effec dominaes he curren accoun effec in deermining he shor-run impac on he exchange rae. Since we know ha y y = θ e, we can conclude ha he appreciaion of he exchange rae also implies a decrease in domesic oupu compared o foreign. This is due o he expendiure swiching effec relaed o he appreciaion of he domesic currency, which shifs demand from domesic goods o foreign ones. C. Long-Run Effecs In order o assess he long-run effecs of a domesic reducion in public employmen, we can sar by noicing ha he uni roo resul ( c c = c c ) implies ha he effec of he policy change on relaive consumpion in he long run will be he same as in he shor run. A reducion in domesic public employmen, herefore, implies an increase in domesic consumpion relaive o foreign boh in he shor and in he long run. In order o beer undersand he long-run implicaions of he policy ha we are considering, i is also useful o derive is effecs on he long-run ne foreign asses posiion of he domesic counry b. A possible mehod o solve for b is o combine equaion (25) and equaion (3) in order o express c c as a funcion of policy variables only. The nex sep is o use he uni roo resul ( c c = c c ) in order o find an expression which can be used o eliminae c c in equaion (28), yielding he following reduced form for b K where K2 = + ( + ) >. σ θ δ G n L L K g g b θ = ( + )( l l ) + δ θ L ( θl + L) K2 nk n K + ( g g ) + ( + )( g g ) δ K θ δ K δ( θ ) 2 2 From equaion (3) i is clear ha a reducion in domesic governmen employmen has a negaive effec on he ne foreign asse posiion of he domesic counry. The inuiion for his resul is relaed o he increase in relaive consumpion boh in he shor and in he long run which, ogeher wih he shor-run reducion in relaive oupu, creae pressures for a worsening of he long-run financial posiion relaive o he foreign counry. In order o have some insighs on he impac on long-run oupu, i is possible o combine he demand funcion (9), he long-run curren accoun funcion (2), and heir foreign counerpars, his yields (omiing for simpliciy erms independen of changes in governmen employmen): (3)

16 - 5 - y y = θ δ ( ) b θ ( c c ) θ n + θ (32) Equaion (32) illusraes he main economic channels ha impac relaive oupu in he long run. The negaive relaionship beween b and y y illusrae he fac ha higher ne foreign asses would make domesic agens richer, herefore promping hem o consume more leisure and work less. On he oher hand, he posiive relaionship beween relaive consumpion and relaive oupu illusrae he need o work more in order o finance higher goods consumpion. Since we have already shown ha a reducion in public employmen implies a reducion in b and an increase in ( c c ), i is clear from equaion (32) ha he oupu of he domesic counry relaive o foreign will increase in he long run following a reducion in public employmen. Taking ino accoun he simple relaionship ha we have posulaed beween labor inpu and producion in he privae secor, i is sraighforward o verify ha he formal resuls on he oupu dynamics confirm our inuiive explanaion on he secoral shif from public o privae employmen, which implies an increase of privae employmen in he long run. We believe ha his paper is he firs conribuion in he new open economy macroeconomics framework o differeniae beween governmen spending for public employmen and for nonwage public consumpion. As already sressed in he inroducion, however, his problem was analyzed in he RBC radiion by Finn (998). Our conribuion differs from Finn s in a leas hree aspecs. Firs, since in our seup we incorporae some ypically Keynesian feaures, such as imperfec compeiion and nominal rigidiies, we can carry ou an analysis ha differeniaes he effecs of changes in public employmen beween he period in which he economy is characerized by underuilizaion of resources (he shor run) and he full employmen period (he long run). In comparing our resuls wih Finn s, i is imporan o bear in mind ha she considers an expansion in public employmen as he benchmark exercise (Finn, 998, p. 642), while we have focused on he effec of public employmen rerenchmen. Our analysis shows ha he resul regarding he effecs of changes in governmen employmen on privae consumpion idenified by Finn is likely o be broadly consisen wih he one ha we find. Regarding oupu dynamics, however, our analysis implies ha a fall in oupu is no always he necessary oucome of an increase in governmen employmen, conrary o wha Finn finds. In our model, if governmen employmen were increased, he oupu of he domesic counry would increase compared o he foreign in he shor run. Admiedly, he counry compeiiveness facor would play a role in his resul as would he effecs of iniial underuilizaion of resources. This brings us o he second imporan difference beween our model and Finn s, namely our open economy focus, which allows us o look a he effecs of changes in public employmen on variables like he exchange rae and he exernal posiion of he counry.

17 - 6 - Finally, we solve our model by deriving analyical soluions and using a simple graphical apparaus, raher han using numerical simulaions. Alhough his prevens us from carrying ou a quaniaive analysis of he impac of he various componens of public spending on he cycle, which is an exremely ineresing feaure of Finn s paper, we also believe ha our mehod presens some advanages. In paricular, he simple graphical apparaus ha we presen is helpful in providing an inuiive explanaion for he resuls. Deriving analyical soluions also allows us o sudy he effecs of a change in he composiion of governmen spending (an increase in governmen consumpion financed by public employmen rerenchmen). We now urn o he analysis of such a policy. D. Effecs of a Shif in Governmen Spending Composiion In Secion III C we focused on he consequences of a domesic reducion in public employmen for consan non-wage governmen consumpion. As we have already sressed, when such policy is implemened, domesic agens benefi from a reducion in axes. In his secion we analyze he effecs of a slighly differen policy, in which he reducion in governmen employmen is used o finance an increase in non-wage governmen consumpion. From he log-linearized version of he governmen budge consrain equaion (and from is foreign equivalen) i follows ha he following relaionship holds: G G g g = H ( ) l l (33) where he bar superscrip denoes permanen policies (i.e., g = g = g,ec.). This implies ha equaion (3) can be rewrien as follows K θ L L + δ G g g e = ( + )( c c ) + [ ( ) H]( l l ) (34) θ δ δ θ L ( θl + L) θ δ where he impac of he policy ha we are considering will be capured by he expression in he square bracke. Since, as in he case of he previous policy, equaion (25) will sill be independen of governmen policies, equaion (34) will deermine he impac on shor-run relaive consumpion and exchange rae. Since he expression is he sum of a posiive and of a negaive erm, i is clear ha a reducion in public employmen, which is used o finance an increase in non-wage public consumpion, will have an ambiguous effec on he various macroeconomic variables. If he posiive erm in he square bracke is dominan, equaion (34) implies ha, following he reducion in public employmen, he same level of relaive consumpion will be consisen wih a lower exchange rae (equaion (34) will shif righward, as equaion (3) does in Figure 2), while he opposie (lefward shif) will be rue if he negaive erm dominaes. The inuiion for he ambiguous resul can be found by noicing ha when he reducion in public employmen is used o finance an increase in governmen spending, agens do no benefi from he reducion in he ax burden. Since he reducion in labor is only absorbed by he privae

18 - 7 - secor in he long run, his policy, herefore, implies a ne loss of labor income for domesic agens, which should depress domesic consumpion compared o foreign consumpion. A he same ime, however, he increased governmen consumpion simulaes demand, herefore increasing profis and conribuing o increasing privae wealh, which has a posiive effec on privae wealh and herefore on consumpion. Inspecion of he expression in square bracke in equaion (34), suggess ha he case of a lefward shif of he curve is more likely he bigger he share of public wage bill on GD a he iniial equilibrium, H. This is consisen wih our inuiive explanaion for he ambiguous effec of such a policy, since a siuaion in which a larger iniial share of income is derived from employmen in he public secor will imply a sronger negaive wealh effec, and herefore, a sronger pressure for reducion in consumpion, when public employmen rerenchmen happens. IV. CONCLUSIONS In his paper we inroduced a disincion beween governmen spending for public employmen and for non-wage public consumpion in a new open economy macroeconomics model. This allowed us o help overcome a paradox of economic lieraure, noed by Alesina and eroi (995), which implies ha he academic debae on he macroeconomic effecs of fiscal policy refers almos exclusively o he impac of governmen consumpion. Since he new open economy macroeconomics framework is emerging as he new paradigm for he analysis of macroeconomic inerdependence, our conribuion offers some insighs on he effecs of public spending composiion in his new framework. We have shown ha a reducion in domesic employmen used o reduce he ax burden implies boh, an immediae increase in domesic consumpion compared wih foreign consumpion, and an appreciaion of he domesic exchange rae. When he reducion in public employmen is used o finance an increase in domesic non-wage governmen consumpion, he resuls are ambiguous and can be affeced by he iniial share of privae income derived from employmen in he public secor. The research presened in his paper could be exended in various direcions. In paricular, inroducing governmen invesmen ino he model would allow analysis of a wider range of policies. Since in he framework presened in his paper Ricardian equivalence holds and governmen deb would no have real effecs, we have limied our analysis o reducions in governmen employmen mached by changes eiher in consumpion spending or in axaion. Exending he model o a framework in which governmen deb can have real effecs (e.g., along he lines of Ganelli, 24) would allow an analysis of how he composiion of governmen spending ineracs wih various financing alernaives.

19 - 8 - REFERENCES Alesina, Albero, and Robero eroi, 995, Fiscal Expansions and Adjusmens in OECD Counries, Economic olicy, pp Finn, Mary G., 998, Cyclical Effecs of Governmen Employmen and Goods urchases, Inernaional Economic Review, Vol. 39, pp Ganelli, Giovanni, 23, Useful Governmen Spending, Direc Crowding-Ou and Fiscal olicy Inerdependence, Journal of Inernaional Money and Finance, Vol. 22 (), pp , 24, The New Open Economy Macroeconomics of Governmen Deb, forhcoming in he Journal of Inernaional Economics. Hau, Harald, 2, Exchange Rae Deerminaion: he Role of Facor rice Rigidiies and Nonradeables, Journal of Inernaional Economics, Vol. 5 (2), pp Lane, hilip R., 2, The New Open Economy Macroeconomics: A Survey, Journal of Inernaional Economics, Vol. 54 (2), pp , and Giovanni Ganelli, 23, Dynamic General Equilibrium Analysis: The Open Economy Dimension, Elemens in Dynamic Macroeconomic Analysis, ed. by S. Alug, J. Chadha, Lane hillip, and C. Nolan (Cambridge: Cambridge Universiy ress)., and Robero eroi, 23, The Imporance of Composiion of Fiscal olicy: Evidence from Differen Exchange Rae Regimes, Journal of ublic Economics, Vol. 87, pp Obsfeld, Maurice, and Kenneh Rogoff, 996, Foundaions of Inernaional Macroeconomics, Chaper, Cambridge, Massachuses: MIT ress., and Kenneh Rogoff, 995 Exchange Rae Dynamics Redux, Journal of oliical Economy, Vol. 3, pp