Tax Reforms, Free Lunches, and Cheap Lunches in Open Economies

Transcription

1 WP/8/227 Tax Reforms, Free Lunches, and heap Lunches in Open Economies Giovanni Ganelli and Juha Tervala

2

3 28 nernaional Moneary Fund WP/8/227 MF Working Paper MF nsiue Tax Reforms, Free Lunches, and heap Lunches in Open Economies Prepared by Giovanni Ganelli and Juha Tervala 1 Auhorized for disribuion by Enrica Deragiache Sepember 28 Absrac This Working Paper should no be repored as represening he views of he MF. The views expressed in his Working Paper are hose of he auhor(s) and do no necessarily represen hose of he MF or MF policy. Working Papers describe research in progress by he auhor(s) and are published o elici commens and o furher debae. This paper focuses on he macroeconomic and budgeary impac of ax reforms in a New Keynesian wo-counry model. Our resuls show ha boh income and consumpion unilaeral ax rae reducions do no consiue a free lunch, in he sense ha hey have negaive budgeary consequences for he counry which implemens hem. n addiion, he degree of self-financing implied by our model is in he 8½-24 percen range. Since he degree of self-financing esimaed in previous lieraure was larger, we conclude ha in our model no only he lunch is no free, bu is also no ha cheap. A comparison of alernaive (income-ax versus consumpion-ax based) fiscal simulus packages shows ha consumpion ax cus imply a larger shor-run impac on domesic oupu bu he income ax cus simulae he domesic economy more in he long run. We also look a he implicaions of a revenue-neural ax reform in which consumpion axes are increased o compensae for lower income ax collecion. JEL lassificaion Numbers: E62, F41, H2 Keywords: Tax us; Dynamic Laffer Effecs; Self-financing. Auhor s Address: gganelli@imf.org, juha.ervala@helsinki.fi 1 Mr. Ganelli is wih European Division of he MF nsiue and Mr. Tervala is wih he Deparmen of Economics, Universiy of Helsinki, Finland. We are graeful for commens o Enrica Deragiache, Lennar Erickson, Peri Haaparana, Juha Kilponen, Erkki Koskela, Tapio Palokangas, Panu Pouvaara, Ani Ripai and seminar paricipans a he MF nsiue and a he Universiy of Helsinki. Juha Tervala would like o hank he Yrjö Jahnsson Foundaion for financial suppor.

4 2 onens Page. nroducion...3. The Model...6 A. Households...6 B. The Governmen...8. Firms...9 D. The niial Seady Sae...1. Parameerizaion...1 V. The Domesic and nernaional Effecs of a u in he ncome Tax Rae...11 A. The mpac on he Domesic Economy...13 B. The nernaional Effecs...16 V. onsumpion Tax us...18 V. A Revenue Neural Tax Reform...21 V. Sensiiviy Analysis...23 V. onclusions...25 Appendix...27 References...28

5 3. NTRODUTON The debae on wheher ax cus can pay for hemselves is ofen associaed wih he idea, popularized in he 198s, of he Laffer curve. The original Laffer argumen was ha here is, a any given poin in ime, a hump-shaped relaionship beween he ax rae and acual revenue collecion. Laer academic and policy discussions have exended his concep in a dynamic sense. While he precise meaning of a dynamic Laffer curve is open o inerpreaion and various definiions have been used in he lieraure, a minimum necessary condiion for dynamic Laffer effecs o happen is ha a ax cu oday will increase growh and, a some poin in he fuure, deliver higher ax revenues in he absence of oher policy changes. 2 n realiy, he idea ha his migh be possible pre-daes he Laffer debae and goes back a leas o Keynes, who saed ha: Nor should he argumen seem srange ha axaion may be so high as o defea is objec, and ha, given sufficien ime o gaher he fruis, a reducion of axaion will run a beer chance han an increase of balancing he budge. (Keynes 1933; p.5). n more recen years, Auerbach (25) has sressed ha he mehodology used by he US Join ommiee on Taxaion (JT) o forecas he revenue impac of legislaion changes is a parial equilibrium one, in he sense ha i akes nominal GDP and oher macro aggregaes as given. Auerbach (25) argues ha such pracice, by ruling ou he possibiliy of a posiive response of economic aciviy o ax reducions, biases he legislaive process agains ax cus. To overcome his problem, he suggess ha he JT should adop a general equilibrium mehodology, in which feedback effecs from axes o oher macroeconomic variables are aken ino accoun. 3 Ouside he US, as noed by Keen, Kim and Varsano (26), much of he rheoric in counries which have implemened so called fla ax reforms has been concerned more wih he rae reducion aspec of he reform han wih flaness iself. From his poin of view, a main drive behind he recen wave of fla axes has been he idea ha ax cus would provide a free lunch by self-financing hemselves. 4 Beside he quesion of he budgeary impac, anoher imporan aspec of ax reforms is relaed o heir open economy dimension. n heir Harry G. Johnson Lecure, Frenkel and Razin (1989) argued ha, due o he increased inegraion of world capial markes and is 2 f his minimum necessary condiion is no saisfied, none of he various definiions of dynamic Laffer effecs used in he lieraure (see, for example, reland (1994), p. 563; Novales and Ruiz (22), p. 188) can be saisfied. 3 The parial equilibrium mehodology used by he JT is also referred o as saic scoring, while he alernaive general equilibrium mehodology is also referred o as dynamic scoring. 4 While he erm fla ax has been used loosely and he various versions which have been adoped (mos noably by Russia and by oher counries in enral and Easern Europe) vary widely, common feaures have ofen been boh a reducion of he number of income ax brackes and a subsanial reducion in ax raes. See Keen, Kim and Varsano (26) for an ineresing analysis of recen fla ax experiences.

6 4 effecs on policy inerdependence beween counries, a proper analysis of he implicaions of ax reforms should be carried ou wihin a global open-economy framework. This saemen, which was wihou doub already rue when he Lecure was delivered, is all he more valid in relaion o oday s highly globalized world economy, in which ax compeiion and highly mobile facors of producion need o be facored in by naional auhoriies in heir ax policy decisions. n his paper we aim a joinly analyzing he wo imporan dimensions of ax reforms discussed above, by focusing on he domesic and inernaional budgeary impac of unilaeral income and consumpion ax cus. n doing so, we use a New Keynesian wocounry model wih imperfec compeiion and nominal rigidiiies. Our model falls in o he so-called New Open Economy Macroeconomics srand of he lieraure. 5 Our resuls show ha, for a sandard parameerizaion, dynamic Laffer effecs do no emerge in our model. Boh income and consumpion unilaeral ax raes reducions have negaive budgeary consequences for he counry which implemens hem. n addiion o sudying wheher ax cus can pay for hemselves and be a free lunch for he budge, we also sudy wheher hey can be largely self-financing in he sense of Mankiw and Weinzierl (26) and herefore resul a leas in a cheap lunch. We find ha he degree of self-financing of income ax cus in real erms is abou 17 percen in our benchmark parameerezaion and is in he percen range in he sensiiviy analysis ha we carry ou. The degree of selffinancing of consumpion ax cus in real erms is 11.5 percen in our benchmark and in he range in our sensiiviy analysis. Since he magniudes ha we derive for selffinancing are a he boom of he range calculaed in previous lieraure, we conclude ha in our framework ax cus no only do no deliver a free lunch, bu ha he lunch is also no ha cheap. n addiion o he budgeary impac, he ax reforms ha we sudy also have imporan implicaions for domesic and foreign macroeconomic variables. A reducion in he domesic labor income ax rae generaes a domesic boom, in which boh oupu and consumpion increase. The foreign economy is affeced boh in he shor and in he long run hrough various ransmission channels (an expendiure swiching effec, a erms of rade effec, and a rade surplus/defici effec). Anoher conribuion of our paper is a comparison beween alernaive (income-ax based versus consumpion-ax based) fiscal simulus packages. We wiev his as an imporan issue, since he policy makers of several indusrial counries have shown, in recen years, a renewed ineres in fiscal policy as a couner-cyclical ool. For example, ax cus have been used o 5 Following he seminal paper by Obsfeld and Rogoff (1995, 1996), imporan conribuions o his lieraure include, bu are no limied o, Bes and Devereux (2, 21), orsei and Peseni (21), and Obsfeld and Rogoff (2, 22). Surveys of his lieraure are provided by Lane (21), Sarno (21), ouinho (25), and orsei (27).

7 5 simulae he economy in he US more han once in he las decade, while Japan also ried o boos growh in he 199s hrough an expansionary fiscal policy. n Europe, he Sabiliy and Growh Pac has been recenly re-inerpreed in a way ha faciliaes he counercyclical use of fiscal policy. The imporance of fiscal policy as a ool o simulae he economy in a downurn was also recenly sressed by he MF s Managing Direcor, who saed Bu in a sense, medium-em fiscal policy is all abou saving for a rainy day. is now raining (Srauss-Kahn 28). Our resuls show ha, if a given reducion in public spending and oal revenue collecion is achieved by a consumpion (raher han income) ax rae reducion, he impac on domesic oupu is larger in he very shor run bu smaller in he medium and long run. The imporance of he issues on which we focus has obviously no escaped previous conribuions. Mos of he papers in he ax reform lieraure use closed-economy endogenous growh or neoclassical models o ascerain he exisence of dynamic Laffer effecs. reland (1994) finds ha in an endogenous growh model an income ax reducion from an original rae of 2 percen generaes a dynamic Laffer effec as long as he new rae is greaer han 7.6 percen. Pecorino (1995) criicizes reland (1994) for no aking ino accoun ha reurns from human capial accumulaion are less highly axed compared o oher income sources, and shows ha if his is aken ino accoun dynamic Laffer effecs only emerge for iniial levels of ax raes of he order of 6 percen. Novales and Ruiz (22) exend reland (1994) and Pecorino (1995) by explicily aking ino accoun ransiional dynamics beween balanced growh pahs. They show ha dynamic Laffer effecs can arise for reducions of income ax raes of up o 5 percenage poins saring from an iniial rae of 23 percen. Bruce and Turnovsky (1999) sress ha reland s resul rely on an implausibly high (greaer han uniy) ineremporal elasiciy of subsiuion and ha dynamic Laffer effecs can be ruled ou under more realisic parameerizaions. Dynamic Laffer effecs are likely o emerge in lieraure discussed above because, under he assumpion of endogenous growh, he growh impac of ax cus are large. This is no necessarily he case in neoclassical Real Business ycle (RB) models, in which he issue is raher how much of he ax cu pays for iself. Accordingly, Mankiw and Weinzierl (26) have recenly shifed he research focus, in a neoclassical seing, from dynamic Laffer effecs o he degree of self-financing, in he sense of how much of he parial equilibrium revenue loss is paid for by growh (see secion V.A below for he formal definiion). Using his mehodology, hey calibrae a neoclassical growh model o he US, finding ha 17 percen of a labor income ax cu is self-financing for sandard parameer values, a magniude close o he one (19 percen) calculaed by Traband and Uhlig (26) in a similar exercise. ompared o he above menioned closed-economy papers, one innovaion of our conribuion is ha of joinly analyzying he macroeconomic and budgeary implicaions of ax reforms in an open economy framework. This allows us, in he spiri of Frankel and Razin (1989), o sudy he domesic and inernaional impac of unilaeral ax reforms in a

8 6 globalized economy. A paper which akes a similar open economy view of ax reforms is Mendoza and Tesar (1998). Using a dynamic neoclassical wo-counry model, hey sudy he domesic and inernaional implicaions of a US ax reform. One difference of our conribuion compared o heirs is ha hey do no explicily focus on dynamic Laffer effecs and on he degree of self-financing, since he ax reforms hey analyze are such ha any revenue loss caused by reducions in income ax raes is perfecly offse by an increase in consumpion axes. 6 Our choice of using a New Keynesian model is anoher imporan difference compared o all he paper reviewed above, which allows us o ake ino accoun how ax reforms inerac wih marke imperfecions such as he degree of nominal rigidiy in he economy. The paper is organized as follows. Secions and respecively inroduce he model and our benchmark parameerizaion. Secion V and V presen and compare he resuls of income ax and consumpion ax raes reducions. Secion V focuses on a revenue-neural ax reform. Secion V presens some sensiiviy analysis. Secion V concludes.. THE MODEL We use a sandard NOEM model, similar o he one developed by Bes and Devereux (2). ompared o he laer, here are wo main differences. The firs is he inroducion of income and consumpion axes insead of lump-sum ones. The second is ha nominal rigidiies ake he form of saggered price seing as in alvo (1983), raher han one-period fixed prices. Bes and Devereux (2) assume ha a fracion of firms fix prices in he currency of he consumer. Their model herefore allows for boh Local urrency Pricing (LP) which implies deviaions from Purchasing Power Pariy (PPP) and Producer urrency Pricing (PP). n his paper we focus our aenion on he PP case, absracing from deviaions from PPP. The model conains wo counries. Firms and households are indexed by z [,1]. A fracion n of households and firms are locaed in he domesic counry, while 1 n are locaed in he foreign counry. n he presenaion of he model below we will inroduce domesic equaions. Unless equaions for he foreign counry are explicily discussed, hey can be assumed o be symmeric o he equaions for he domesic counry. A. Households Households gain uiliy from privae consumpion and real balances, and experience disuiliy from supplying labor. Their uiliy funcion is herefore given by 6 Alhough we mosly focus on policy experimens in which governmens adjus ransfers o compensae changes in ax collecion, in Secion V we also look a he implicaions of a revenue neural exercise similar in spiri o he one carried ou by Mendoza and Tesar (1998).

9 7 ν + 1 s χ M s 1 ε ls( z) U = β [logs + ( ) ] (1) 1 ε P ν + 1 s= s where < β < 1 is he discoun facor, s is a composie good represening privae consumpion and Ps is he price index associaed wih i. M s denoes nominal money balances and l s (z) he household s supply of labor; ε > is he inverse of he consumpion elasiciy of money demand, ν is he elasiciy of he marginal disuiliy of producing oupu wih respec o oupu, and χ is a posiive parameer. The composie privae consumpion good is defined in he following equaion as an aggregae across he individual goods produced by firms 1 θ 1 θ θ θ 1 = [ c ( z) dz] (2) where θ is he elasiciy of subsiuion beween any pair of individual goods. The associaed price index is n θ * 1 θ 1 θ = [ p ( z) dz + ( E p ( z)) dz] n P (3) where p (z) is he price of good z expressed in domesic currency, p * ( z ) is he foreign currency price of foreign good z and E is he nominal exchange rae, defined as he price of he foreign currency in erms of he domesic currency. The budge consrain of he domesic represenaive household is given by M + + δ D = D 1 + M 1 + (1 τ ) wl ( z) (1 + τ ) P + π P T (4) where D denoes he household s holding of nominal bonds. Bonds are denominaed in he currency of he domesic counry and accoun for inernaional shifs in wealh, δ is he price of a bond (he inverse of one plus he nominal ineres rae), w is he nominal wage paid o he household in a compeiive labor marke, π is he household s share of profis received from firms, τ andτ are he ax raes on household income and consumpion, and T denoes real ransfers from he governmen. 7 Given ha bonds are denominaed in domesic currency, he budge consrain of he foreign represenaive household is 7 The fac ha bonds are denominaed in domesic currency does no inroduce any asymmery across counries, since we assume open capial markes so ha nominal ineres raes are equalized inernaionally. Furhermore, PPP holds under PP.

10 8 M D D * * * * * 1 * * * * * * * * + δ = + M + (1 τ ) w l ( z) (1 ) P P T 1 + τ + π + (5) E E where foreign variables are denoed by aserisks. A global asse-marke clearing condiion nd (1 n) D * = also holds. + Domesic households maximize (1) subjec o (4), and an analogous opimizaion problem holds for foreign households. The resuling firs order condiions are δ P ) P (6) ( 1+ τ ) = β (1 + τ + 1 * * δ + τ β τ ) P E (7) * * * * ( 1 ) P E+ 1 = ( w ν = (8) (1 + τ ) P l l * ν M P M P (1 τ ) * (1 τ ) w = (9) * * (1 + τ ) P * * * 1 ε χ(1 + τ ) = ( ) (1) 1 δ * * 1 χ(1 + τ ) ε = ( ) (11) δe E Equaions (6) and (7) are he Euler equaions for opimal domesic and foreign consumpion including axes, hey reduce o sandard Euler equaions if he ax rae on consumpion is kep consan. Equaions (8) and (9) are he domesic and foreign opimal labor supply equaions, which equae he disuiliy of supplying an exra uni of labor wih he marginal uiliy of he exra privae consumpion ha can be bough due o he marginal increase in labor supply. Equaions (8) and (9) show ha higher labor or consumpion axes reduce labor supply for given levels of he real wage and consumpion. Finally, equaions (1) and (11) show ha households opimal money demand is an increasing funcion of privae consumpion (including axes) and a decreasing funcion of he ineres rae. B. The Governmen We assume ha all governmen spending is for public ransfers o households, which can be financed hrough income and consumpion axes or seignorage. 8 We herefore absrac from governmen spending for public consumpion and invesmen. Taking ino accoun symmery across agens, he governmen budge consrain in per-capia erms can herefore be wrien as 8 n wha follows we will keep money supply consan, herefore absracing from seignorage in pracice.

11 9 w M M τ l (12) 1 T = + τ + P P where T denoes real ransfers. Technology. Firms Each firm produces a differeniaed good according o he simple producion funcion y ( z) = l ( z) (13) where y (z) is he oupu of firm z and l (z) he labor inpu used by firm z. Profis We assume ha each firm enjoys a cerain degree of monopolisic power in he producion of is differeniaed good. Under his assumpion, demand for he oupu of firm z is given by p ( z) y ( z) (14) d where θ = ( ) P W W is world aggregae consumpion given by W = n + (15) * ( 1 n) Profis are defined as π ( z) = p ( z) y ( z) wl. Using (13), (14) and (15) profis can be wrien as p ( z) θ * π ( z) = ( p ( z) w )( ) ( n + (1 n) ) (16) P Price Seing n he absence of price rigidiies, he profi maximizaion process would imply ha he price of each differeniaed good is given by a simple mark-up over wages, according o he formula θ p ( z) = w (17) θ 1 However, following alvo (1983), we inroduce nominal rigidiies by assuming ha each firm reses is price wih a probabiliy 1 γ in each period, independenly of oher firms and independenly of he ime elapsed since he las adjusmen. Each firm has o ake ino accoun, when seing is profi-maximizing price, ha in every subsequen period here is a probabiliy < γ < 1 ha i will no be able o revise is price seing decision. When seing a new price in period, each firm seeks o maximize he presen value of profis weighing

12 1 fuure profis by he probabiliy ha he price will sill be effecive in ha period. Thus he represenaive home firm seeks o maximize s max V ( z) = ζ, s p ( z) s= γ π ( z) (18) s whereζ is he domesic discoun facor beween period and period s, defined as, s ζ pricing rule 1, s = Π j= s (1 + i j ) θ p ( z) = ( ) θ 1 s= s=, where i he domesic nominal ineres rae. The resul is he following s W 1 θ γ ζ, s ( s )( ) w Ps s W 1 θ γ ζ, s ( s )( ) P s (19) All firms in a counry are symmeric and every firm ha changes is price in any given period chooses he same price and oupu consisenly wih (19). The srucure of price seing implies ha each period a fracion of 1 γ of firms ses a new price and he remaining fracion keeps heir price unchanged. D. The niial Seady Sae n he policy exercises which we carry ou below, we log-linearize he model around a symmeric seady sae. We consider he special case in which iniial ne foreign asses are zero ( D = ). Under his assumpion we have, using he zero subscrip o denoe he iniial seady sae 1 1 τ θ 1 ν + 1 = = [( )( 1+ τ θ y )] (2). PARAMETERZATON The benchmark parameerizaion of he model mosly follows Bes and Devereux (2). The elasiciy of subsiuion beween differeniaed goods θ is se o 11. The discoun facor β is assumed equal o.94, implying a seady-sae ineres rae of abou 6 percen. The consumpion elasiciy of money demand 1/ε is se o 1. The disuiliy parameer ν is assumed o be uniary. The counries are assumed o be of equal size, implying n=.5. The price rigidiy parameer is se a γ=.5. niial income and consumpion ax raes of respecively 2 and 8 percen are assumed ( τ =.2; τ =.8). Our goal is o presen simple numerical examples in order o illusrae he heoreical mechanisms underpinning he domesic and inernaional impac of ax reforms. We herefore do no calibrae he model o any paricular

13 11 counry. n any case, we believe our iniial income ax raes o be wihin he ballpark of he range of plausible esimaes for many indusrial and emerging marke economies. Baxer and King (1993) and reland (1994), for example, useτ =. 2 as a benchmark in parameerizaions of he US economy, while Mankiw and Weinzierl (26) use τ =. 25. Mendoza, Razin, and Tesar (1994) esimae effecive labor income ax raes in he percen range for he US, he UK, anada and Japan in 1988 (he las year for which hey do his exercise). Their esimaes are much higher for he oher indusrial counries included in heir sample (reaching 47 percen for France). However, considering ha in many emerging markes (such as hose which have recenly moved owards he fla ax) income ax raes are much lower, we believe ha τ =. 2 percen is an appropriae parameerizaion for our exercises, which aim a saring from a somewha sandard fiscal sance. Similar consideraions were made in he choice of τ = n secions V, V and V we repor and discuss he resuls of calibraions under his benchmark parameerizaion derived using he algorihm developed by Klein (2) and Mcallum (21). n Secion V we presen some sensiiviy analysis experimens in which we look a he implicaions of a wider range of variaion for he labor disuiliy and nominal rigidiy parameers. V. THE DOMEST AND NTERNATONAL EFFETS OF A UT N THE NOME TAX RATE n his secion we analyse he implicaions of an unexpeced reducion in he income ax rae τ from.2 (a 2 percen income ax rae) o.19 (a 19 percen income ax rae), which corresponds o a 5 percen cu in he rae. Since we wan o iniially focus on he macroeconomic and budgeary implicaions of income ax reform absracing from changes in oher axes, we leave he consumpion ax rae τ unchanged a is iniial seady sae level of.8 (an 8 percen consumpion ax rae) in his secion. 1 The resuls of his policy exercise are presened in Figure 1. The verical axes show percenage deviaions from he iniial seady sae. For variables whose iniial seady-sae value is zero, deviaions are expressed in relaion o iniial oupu. Figure 1 shows he response of domesic and foreign macroeconomic variables (including oal revenue collecion). Alhough much of he public opinion debae on he budgeary implicaions of ax reforms is cas in erms of he impac on nominal revenue collecion, wha ulimaely maers for he governmen abiliy o carry ou is funcions is he amoun of real resources available o he public secor. Since, unlike mos of he papers which have looked a similar issues, we use a moneary model wih nominal rigidiy, we can analyze he response of boh nominal revenue collecion (NR) and real revenue collecion (RR). Those are repored in Figure 1(e.g.) and can be derived from (5) as follows (using has o denoe log-deviaions): 9 The esimae of effecive ax raes on consumpion made by Mendoza, Razin, and Tesar (1994) vary from abou 5 percen for he US and Japan o abou 21 percen for France. 1 n secion V we consider he case of changes in he consumpion ax rae.

14 12 ˆ ˆ N R = u( ˆ τ + wˆ + l ) + (1 u)( ˆ τ + ˆ + Pˆ ) (21) ˆ ˆ ˆ R R = u( ˆ τ + wˆ + l P ) + (1 u)( ˆ τ + ˆ ) (22) where u denoes he share of income ax on oal axes in he iniial seady sae (in our benchmark parameerizaion his share is equal o.7, see Appendix for he derivaion). Equaions (21) and (22) refer o oal revenue collecion. n equaion (21), u ( ˆ τ ˆ ˆ + w + l ) denoes nominal income ax collecion, while u ( ˆ τ + ˆ ˆ + P ) denoes nominal consumpion ax collecion. Real income and consumpion ax collecion are derived in an analogous way in equaion (22). Figure 1(d) also includes he response of he domesic erms of rade, defined as he alvoweighed relaive price of domesic expors in erms of domesic impors. Thus he increase in Figure 1(d) implies an improvemen of he domesic erms of rade in he shor run. Figure 1. The Effecs of a Domesic ncome Tax Reducion

15 13 A. The mpac on he Domesic Economy Reducing he income ax rae increases he opporuniy cos of leisure. Domesic households herefore subsiue ou of leisure and ino consumpion. The impac of his subsiuion effec is shown in Figure 1(a) and 1(b): boh domesic oupu and consumpion increase. These resuls are consisen wih he analysis of real business cycle models. Baxer and King (1993), for example, look a he impac of an exercise symmeric o ours, finding symmeric resuls: a balanced-budge increase in public spending financed by income axes reduces boh oupu and consumpion in heir model. As Figure 1(g) shows, domesic real revenue collecion permanenly falls by abou 3 percen compared o is iniial seady-sae level as a consequence of he policy ha we are analyzing. Figures 1(e) and 1(g) show ha, alhough a lower income ax rae increases labor supply, he impac of his on income and oal revenue collecion is no big enough o compensae for he rae reducion. For he reasons discussed above, real consumpion increases following he income ax reducion. This implies ha consumpion ax revenue collecion increases in real erms even a an unchanged consumpion ax rae. However, he quaniaive impac of he increase in consumpion ax collecion is small (Figure 1(g)). The change in oal real ax collecion can be explained wih reference o equaion (22). Since he consumpion ax rae does no change in his policy exercise ( ˆ τ = ), he change in real consumpion ax collecion is given by ( 1 u) ˆ. The low iniial share of consumpion axes in our benchmark parameerizaion (( 1 u ) =.3)) explains why he impac of an increase in real consumpion ax collecion is no big enough o preven a srong decrease in overall real ax collecion. n erms of nominal ax collecion, here is no consumpion ax impac under our parameerizaion. Since we have fixed he consumpion elasiciy of money demand o uniy, and we keep nominal money supply fixed, equaion (1) implies ha changes in nominal consumpion and prices are mirror images of each oher. Nominal collecion of consumpion ax herefore does no change. n erms of equaion (21) ˆ τ =, and Ĉ and Pˆ offse each oher. Table 1. ncome ax rae cu: ax revenues collecion changes 1/ hange in nominal ax revenues hange in real ax revenues = 1 = 3 = 5 New seady sae = 1 = 3 = 5 New seady sae / Percenage changes wih respec o he iniial seady sae. Table 1 summarizes he impac on ax collecion of a change in he income ax rae a various ime horizons. As i is clear from Figure 1 and Table 1, in our model he minimum necessary condiion for dynamic Laffer effecs ha following a reducion in ax raes oday axes will be higher a some poin in he fuure in he absence of furher policy changes is no saisfied. Since no dynamic Laffer effecs arise in our model, i is clear ha ax cus do no

16 14 resul in a free lunch in our analysis. This is in sark conras wih he resuls of some of he closed-economy, endogenous growh models reviewed in he inroducion (such as reland (1994) and Novales and Ruis (22)) in which significan dynamic Laffer effecs emerge for parameerizaions of he iniial ax raes similar o ours. As we have already sressed in he inroducion, his is no surprising in he endogenous growh lieraure, in which he growh effecs of ax cus are likely o be srong. A more relevan comparison of our resuls is wih RB models in which he rae of growh is exogenous. Traband and Uhlig (26) look a he impac of reducions in income axes financed by cus in governmen spending ransfers for a parameerizaion similar o ours. Their resuls are quaniaively similar o ours, since in he new seady sae revenue collecion falls by abou 2 percen, compared o abou 3 percen in our case. 11 How can quie differen models generae such close resuls? One explanaion is ha ha while he inroducion of marke imperfecions compared o he RB framework enails ha our iniial seady-sae is subopimal and larger efficiency gains can be derived from reducing he ax burden, hose gains are offse by he open-economy dimension, which implies ha par of he growh benefis of he domesic ax reducion accrue o he foreign counry. The imporance of he open economy dimension in prevening he emergence of dynamic Laffer effecs in our framework is even more eviden if our resuls are compared wih he RB model and Leeper and Yang (28), in which a 1 percen reducion in income axes saring from an iniial ax rae of 25 percen implies a revenue loss much smaller han ours (abou.3 percen compared o 3 percen in our case). The possibiliy of analysing he ineracion of marke imperfecions and open economy channels is an advanage of using a NOEM framework, on which we will focus more explicily in secion V B. The Degree of Self Financing: if no a Free Lunch, a Leas a heap Lunch? As we have discussed above, free lunches deriving from ax cus are generally ruled ou boh in our NOEM model and in previous RB models. Recen research (Mankiw and Weinzierl 26; Traband and Uhlig 26) however, has shifed he focus in he RB lieraure from free lunches o cheap lunches, in he sense of invesigaing wheher ax cus which do no generae dynamic Laffer effecs can be a leas largely self-financing. The degree of self financing has been defined as 12 where 1*(1-x) (24) 11 See Traband and Uhlig (26), Table 7. The fall in revenue collecion is smaller if he iniial ax rae is higher (see Traband and Uhlig (26), Table 8). 12 See Mankiw and Weinzierl (26) and Traband and Uhlig (26).

17 15 dr / dτ ( gen) x = (25) dr / dτ ( par) where he numeraor of (25) is he response of oal revenue collecion o ax rae changes in a general equilibrium sense (in which all endogenous variables reac o he rae change) while he denominaor is he response of oal revenue collecion o ax rae changes in a parial equilibrium sense (in which he response of endogenous variables is shu off). The inuiion behind his definiion is ha if x=1 hen he general equilibrium effec of ax changes on revenue collecion is equal o he parial equilibrium effec. n his case here is no selffinancing (he degree of self-financing is zero) and he parial equilibrium mehodology used by he JT o evaluae he impac of proposed ax legislaion is appropriae. n mos cases, however, he degree of self-financing is likely o be posiive. Mankiw and Weinzierl (26), for example, calculae i o be 5 percen for capial income axes and 17 percen for labor income axes for he US. While Traband and Uhlig (26) find values of 47 percen (capial income) and 19 percen (labor income) for he US and 85 percen (capial income) and 54 percen (labor income) for EU-15. Leeper and Yang (28) mosly focus on revenue-neural exercises in which equaion (25) is no applicable. They do look, however, a a case in which income ax cus are financed by lower ransfers, finding a degree of self-financing of 47 percen. Since we have inroduced money and nominal rigidiies in our model, we can calculae he degrees of self-financing in erms of boh nominal and real revenues. Using (21) and (22), we can easily calculae he raio defined in (25) for our model. This raio is given a any ime horizon by ˆ u( ˆ τ ˆ ) (1 )( ˆ ˆ ˆ + w + l + u τ + + P) uˆ τ + (1 u)( ˆ τ ) (26) for he case of nominal revenue collecion and by ˆ ˆ ˆ ˆ u( τ ) (1 )( ˆ ˆ + w + l P + u τ + ) uˆ τ + (1 u)( ˆ τ ) (27) for he case of real revenue collecion. The numeraors of equaions (26) and (27) are he general equilibrium responses of ax collecion o ax raes changes, while he denominaors are he parial equilibrium responses in which endogenous variables do no reac o ax changes ( w ˆ lˆ = ˆ = Pˆ = ). =

18 16 Table 2. ncome Tax Rae u: Degree of Self Financing 1/ Nominal ax revenues Real ax revenues = 1 = 3 = 5 New seady sae T = 1 = 3 = 5 New seady sae / See equaions (25) and (26) for he definiion. Table 2 repors he degree of self-financing a various horizons following an income ax cu. One resul emerging from Table 2 is ha in our model i maers wheher we consider he degree of self-financing in erms of nominal or real ax revenues, since he resuls are significanly differen. More imporanly, Table 2 shows ha he degree of self-financing in real erms sabilizes a abou 17 percen in he new seady sae. 13 This value is of he same order of magniude of hose calculaed by Mankiw and Weinzierl (26) and Traband and Uhlig (26) for a labor income ax cu in he US. As discussed above, however, he same auhors have shown ha he degree of self-financing can be significanly larger in he case of capial income ax cus or when he parameerizaion is based on higher iniial ax raes (such as hose used by Traband and Uhlig (26) in heir parameerizaion of he EU-15). Furhermore, Leeper and Yang (28) find a much larger degree of self-financing. Given ha in our exercise he degree of self-financing is a he boom of he range derived in he exising lieraure, we can conclude ha in our model no only a cu in he income ax rae does no produce a free lunch, bu also ha he lunch is no ha cheap. We are now ready o move o he analysis of global implicaions of ax reforms. B. The nernaional Effecs As we have already sressed above, one advanage of using he NOEM framework is he possibiliy o analyze how marke imperfecions inerac wih he open economy dimension in deermining he resuls. One imporan open-economy channel obviously works hrough exchange rae movemens. Figure 1 (c) shows ha he reducion in domesic income axes implies an appreciaion of he domesic nominal exchange rae (a fall in he price of foreign currency in erms of domesic currency). This appreciaion is due o a money demand effec semming from he increase in domesic consumpion caused by he domesic ax cu. Since money demand is a posiive funcion of consumpion including axes (see equaions (1) and (11)), he increase in domesic consumpion (boh in absolue erms and relaive o foreign consumpion) increases domesic money demand compared o foreign. This implies ha he domesic currency appreciaion displayed in Figure 1(c) is required o reesablish equilibrium in he money marke. 13 The resul of a negaive degree of self-financing a =1 is due o he dynamics of he nominal wage, which in he shor run undershoos is new long-run seady-sae level, hus implying ha he general equilibrium revenue loss is emporarily higher han he parial equilibrium one in equaions (26) and (27).

19 17 Figure 1(a) also shows a emporary increase of foreign oupu following he domesic ax cu. This is parly due o he increase in world demand deriving from higher domesic consumpion. Since household preferences do no display home bias he increase in domesic consumpion falls on foreign as well as on domesic goods (an expendiure boosing effec). Figure 1(a) also shows ha he shor run increase in foreign oupu is faser han he one in domesic oupu. This suggess ha, in addiion o he global expendiure boosing effec discussed above, an expendiure swiching effec is also a work: par of he increase in foreign oupu is due o he fac ha foreign goods become cheaper as a consequence of he appreciaion of he domesic currency. The expendiure swiching effec of a nominal exchange rae change alers he relaive price of goods only as long as prices are sicky. Since under our parameerizaion half of he firms adjus heir prices in every period, he expendiure swiching effec peers ou fas. This explains why he increase in foreign oupu is emporary. Even hough he impac of he expendiure swiching effec is emporary, he fac ha par of he benefis in erms of oupu simulaion of he domesic ax reducion accrue o foreigners explains why dynamic Laffer effecs do no emerge in our model and he degree of self-financing is lower han he ones calculaed in RB models for similar parameerizaions. The shor-run increase in foreign oupu is mached by a shor-run reducion in foreign consumpion. n he long run, foreign consumpion slighly increases compared o is iniial seady-sae level. This dynamics of foreign consumpion is driven by changes in he erms of rade and by foreign households desire o use heir emporary income gains o smooh consumpion over ime. Since in he shor run, due o he appreciaion of he domesic currency, he foreign erms of rade worsen (an increase in Figure 1(d)), foreign households reduce heir shor-run consumpion. n his way hey save par of heir exra shor-run income (Figure 1(a)), hus running a curren accoun surplus (Figure 1(i)). n he medium and long run, however, he foreign erms of rade improve (a fall in Figure 1(d)) due o a relaive increase in he supply of domesic goods which implies a fall in he relaive price of domesic goods. This erms of rade effec, ogeher wih exernal wealh accumulaion due o he curren accoun surplus, allow foreign households o increase heir long-run consumpion even hough heir long-run income reurns o almos iniial seady-sae levels. Figure 1(h) also shows ha he domesic ax reform has a emporary posiive impac on foreign ax collecion, due o he fac ha foreign households increase heir labor supply a an unchanged income ax rae level. The responses of macroeconomic variables o ax cus presened in his secion are broadly consisen wih findings of he empirical lieraure. The resul ha domesic aciviy is simulaed in response o a ax reducion is a sandard one in sudies of he US economy (see for example Blanchard and Peroi (22)). Alhough empirical sudies of he inernaional ransmission of fiscal policy are scarce, our resul of a posiive inernaional oupu spillover is consisen wih foreign oupu mulipliers calculaed by Giuliodori, Beesma, and Klassen (26) for several European counries.

20 18 V. ONSUMPTON TAX UTS n his secion we look a a policy alernaive o he one considered in he previous secion, ha is one in which he governmen decides o unexpecedly reduce he consumpion ax rae raher han he labor income ax rae. n he policy experimen ha we presen here τ is herefore kep consan a is iniial seady-sae level.2 (a 2 percen income ax rae) while he consumpion ax rae τ is reduced from.8 o.7 (from 8 o 7 percen, which amouns o a 12½ percen decrease in he rae). n addiion o allowing us o invesigae he issue of wheher a consumpion ax cu can deliver a free or a cheap lunch, he exercise presened in his secion also makes possible a comparison wih he one discussed in secion V, hus highlighing he differen impac on domesic oupu and consumpion, as well as on he foreign counry, of alernaive fiscal simulus packages. The resuls of his exercise are summarized in Figure 2 and in Tables 4 and 5. Figure 2. The Effecs of a Domesic onsumpion Tax Reducion Table 4. onsumpion Tax Rae u: Domesic Tax Revenues ollecion hanges 1/ hange in nominal ax revenues hange in real ax revenues = 1 = 3 = 5 New seady sae = 1 = 3 = 5 New seady sae / Percenage changes wih respec o he iniial seady sae.

21 19 Table 5. onsumpion Tax Rae u: Degree of Self Financing 1/ Nominal ax revenues Real ax revenues = 1 = 3 = 5 Seady sae = 1 = 3 = 5 Seady sae / See equaions (25) and (26) for he definiion. One imporan difference beween income and consumpion ax reducions is ha hey have opposie effecs on he nominal exchange rae. While in he case of an income ax cu he domesic exchange rae appreciaes (Figure1(c)), a consumpion ax cu resuls in a domesic depreciaion (Figure 2(c)). The inuiion for his resul is he money demand effec already discussed in secion V. Since money demand is a posiive funcion of consumpion including axes, he reducion in he consumpion ax rae has a negaive effec on domesic money demand, which for our parameerizaion is sronger han he effec of he increase in real consumpion excluding axes. The negaive impac on money demand implies a ransmission mechanism symmeric o he one described in secion V, resuling here in a domesic depreciaion (raher han in an appreciaion as in secion V). A resul ha can be clearly seen in Figure 2 and Table 4 is ha, even in he case of consumpion ax cus, no dynamic Laffer effecs emerge in our model. Domesic labor income ax collecion increases in his case, because domesic households increase heir labor supply a an unchanged income ax rae. The increase in labor income ax collecion, however, is no large enough o compensae for he reducion in consumpion ax collecion. As a resul, oal revenue collecion in real erms drops by 3.4 percen in he seady-sae, an order of magniude comparable o he one derived in he case of income ax reducions (compare Tables 1 and 4). Furhermore, he degree of self-financing in real erms in he new seady sae is 11.5 percen (Table 5), which is lower han he 17.3 percen derived for he case of income ax reducions (Table 2). Table 5 also shows ha he degree of self-financing is larger in he shor-run (27.8 percen) han in he long-run. This resul is due o he fac ha he expendiure swiching effec, which shifs demand owards domesic goods hus increasing domesic income ax collecion due o higher labor, is sronger in he shor run. n summary, consumpion ax cus do no generae a free lunch for he budge. n fac, he lunch hey deliver is, in our model, less cheap han he one delivered by labor income ax cus. While previous lieraure did no delve ino he sudy of he self-financing feaures of consumpion ax cus, our resuls are broadly consisen wih he findings of he endogenous growh model of Novales and Ruiz (22), in which no dynamic Laffer effecs emerge for consumpion ax reducions, as well as wih hose of Traband and Uhlig (26), who find mixed resuls (depending on he specific uiliy funcional form) on he exisence of seadysae Laffer curves for consumpion axes. One advanage of inroducing boh income and consumpion axes in he framework we use is ha we can compare alernaive opions for fiscal simulus packages. omparing Figures 2(a) and 1(a) we can see ha a fiscal expansion policy based on a one percenage poin reducion in consumpion axes has a sronger and faser impac on shor-run domesic oupu compared o a fiscal package based on a one percenage poin reducion in income axes. The

22 2 medium and long-erm effecs on domesic oupu of he wo policies are, however, basically equivalen. The fac ha he domesic exchange rae depreciaes insead of appreciaing implies ha he main spillover channels hrough which he domesic ax reducion affecs he foreign counry he expendiure swiching effec, he erms of rade dynamics, and he accumulaion of exernal surplus/defici are now reversed compared o hose presened in secion V. This resul in a dynamics of foreign oupu and consumpion following domesic consumpion ax cus which is almos a mirror image of he one semming from domesic income ax cus (compare Figure 1(a,b) wih figure 2(a,b)). n paricular, foreign oupu falls and foreign consumpion increases in he shor run when he domesic counry chooses o simulae he economy hrough a consumpion-ax based fiscal package. The emporary reducion in foreign oupu implies ha overall foreign revenue collecion falls, because he drop in foreign income ax collecion is no compensaed by higher foreign consumpion ax collecion (Figure 2(f,h)). Since a domesic income-ax based fiscal package has a posiive impac on shor-erm foreign revenues (Figure 1(f,h)), he negaive shor-run budgeary spillover derived here is anoher imporan difference beween he wo fiscal packages. Alernaive Opions o Achieve a Given Reducion in Transfers n Figure 2 and Tables 4 and 5 we have presened he resuls relaed o a 1 percenage poin reducion in he consumpion ax rae in order o compare such a fiscal simulus package wih he one (discussed in secion V) based on a 1 percenage poin reducion in he income ax rae. A differen exercise which can also give an ineresing perspecive on he comparison of income-ax based versus consumpion-ax based fiscal simulus packages is one in which he governmen reduces τ by he amoun needed o sabilize seady-sae public ransfers a he same level implied by he income-based fiscal simulus package sudied in secion V. This comparison can give some insighs on he implicaions of following differen fiscal sraegies for governmens who have decided o reduce public spending by a given amoun. Since in our model he governmen budge is balanced in every period and, in he absence of seignorage, oal ax collecion is always equal o ransfers, his exercise boils down o analysing he reducion in consumpion ax rae which implies he same seady-sae level of oal revenue collecion derived in secion V. The income ax rae reducion analysed in Secion V reduces oal ax collecion and herefore ransfers by 2.9 percen in he new seady-sae. The same reducion in ransfers can be achieved wih a.85 percenage poins cu in he consumpion ax rae. Figure 3 provides a comparison of hese wo differen ways of achieving he same reducion in ransfers.

23 21 Figure 3. omparison of Alernaive Tax Policy Opions o Achieve a Given Reducion in Transfers Figure 3(a) shows ha, once he domesic governmen has decided o reduce ransfers and oal ax collecion by a cerain amoun (in our example o 2.9 percen less han in he iniial seady sae) implemening his plan by reducing he consumpion ax rae implies a larger shor-run impac on domesic oupu compared o an income-ax based sraegy. n he medium and long run, however, he income-ax based sraegy simulaes he domesic economy more han he consumpion-ax based one. V. A REVENUE NEUTRAL TAX REFORM n all he exercises considered so far, ax raes were exogenously reduced and revenue collecion adjused endogenously. n his secion, we consider a policy in which he income ax rae is reduced as in secion V and he consumpion ax rae is increased by he amoun needed o compensae he long-run revenue loss semming from lower income axes. Following an income ax rae reducion from 2 o 19 percen (he same as in secion V), he consumpion ax rae needs o be increased from o 8 o 8.85 percen in order o keep oal revenue collecion consan in he new seady sae. The macroeconomic impac on he domesic and foreign counry of such revenue-neural ax reform is presened in Figure 4.

24 22 Figure 4. mpac of a Revenue Neural Tax Reform Saring wih he domesic economy, i was already clear from secions V and V ha he impac of he revenue neural ax reform would be equal o he sum of he expansionary impac of an income ax reducion (as in Figure 1(a)) and he conracionary impac of a consumpion ax increase (as suggesed by Figure 2(a), in which a consumpion ax cu simulaes he economy). Figure 4(a) and 4(b) shows ha he conracionary impac of he increase in he consumpion ax rae dominaes quaniaively in he shor run, so ha he final effec of he revenue neural policy on he domesic economy is a reducion in oupu, which is only parially reversed in he medium and long run. Figure 4(b) shows ha domesic consumpion excluding slighly increases. Such an increase and he consumpion ax rae hike implies ha domesic money demand which is a posiive funcion of consumpion including axes increases. The consequen appreciaion of he domesic exchange is quaniaively sronger han he one observed in he non revenueneural policy of secion V (compare Figures 4(c) and 1(c)). This appreciaion obviously conribues o shif demand owards foreign goods, hus resuling in a emporary increase of foreign oupu compared o he iniial seady sae. Foreign agens, however, use heir higher income o accumulae exernal asses (Figure 4(i)) raher han o increase consumpion. While in his exercise domesic revenue collecion is consan in he seady-sae due o he policy followed by he domesic governmen, he impac on foreign revenue. is posiive and especially pronounced in he shor run (Figure 4(f,h), due o higher foreign labor supply a an unchanged foreign ax rae. The impac on foreign ax collecion is quaniaively sronger han in he case of a non-revenue neural reducion in domesic income axes (compare Figures 1(f,h) and 4(f,h)) due o he fac ha foreign oupu increases

25 23 more because of a sronger expendiure swiching effec in he revenue-neural policy considered here. Our exercise is similar in spiri o he one carried ou by Mendoza and Tesar (1998) in an open economy RB model wih capial. The resuls are no direcly comparable, since hey look a he implicaions of a much more radical policy han ours, in which he US federal income ax is compleely eliminaed and replaced wih consumpion axes. is ineresing, however, o noice ha in heir case unlike in our model his would imply a posiive effec on domesic oupu and consumpion boh in he long run and along he ransiion pah, while he impac on foreign oupu is posiive in he shor run bu negaive in he long run. While i is difficul o draw srong conclusions from he comparison, one possible explanaion of he differen resuls ha we ge is again ha in our model, due o he presence of nominal rigidiies, he expendiure swiching effec implies ha a sronger par of he benefi of income ax reducions accrues o he foreign counry. This can, a leas parially, explain why he domesic effec of a revenue neural policy is expansionary in heir model bu conracionary in ours. V. SENSTVTY ANALYSS n his secion we presen he resuls of some sensiiviy analysis. n paricular, we look a how he degree of self-financing changes when we change he degree of nominal rigidiy γ and he labor disuiliy parameer ν. Overall, our sensiiviy analysis confirms he robusness of he resuls ha we have discussed in secions V and V. ncome Tax u Table 6 shows ha he resuls on he degree of self-financing in real erms are robus o changes in he degree of nominal rigidiy in he economy. n paricular, he degree of self financing of real revenue collecion sabilizes around 17 percen in he new seady sae regardless of he value of γ. n he case of γ = he nominal degree of self-financing is zero, since in a perfecly compeiive labor marke wih no rigidiies and a uniary disuiliy parameer he nominal wage response is a mirror image of he increase in he labor supply semming from he ax cu ( wˆ = lˆ ). 14 n Table 7 we show how our benchmark resuls change when we vary he labor disuiliy parameer along he range esimaed for i by Roemberg and Woodford (1997). No surprisingly, Table 7 shows ha a decrease in he disuiliy parameer of supplying labor increases he degree of self-financing of labor income ax cus (and vice versa). Even when we se ν =. 47 (a he boom of he range esimaed by Roemberg and Woodford (1997)), however, our esimaed degree of self-financing remains close o 2 percen. 14 This implies ha he general equilibrium and he parial equilibrium effecs are he same in eq. (26) because ŵ and lˆ compensae each oher in he numeraor.