Is Fiscal Policy Alone Enough for Growth? A Simulation Analysis for Bolivia

Transcription

1 Is Fiscal Policy Alone Enough for Growh? A Simulaion Analysis for Bolivia Carlos Gusavo Machicado Paul Esrada Ximena Flores Ocober, 2010 (Final Draf) Absrac In he las years here has been a renewed ineres in fiscal policy as a ool o promoe economic growh, and his has increased he use of quaniaive models o evaluae he macroeconomic effecs of changing axes, increasing governmen expendiures and/or public invesmen. This paper develops a Dynamic Sochasic General Equilibrium (DSGE) model for a small open economy wih five secors: Non-radable or services, imporable or manufacuring, hydrocarbons, mining and agriculure. The model is parameerized and solved for he Bolivian economy and several ineresing scenarios are simulaed by changing all he available fiscal policy insrumens. In paricular we analyze he susainabiliy of he Bolivian social policy based on ransfers from he governmen o households and also he growh and welfare implicaions of he fiscal policy over ime. We find ha solely he fiscal policy is unable o generae he appropriae raes of growh o reduce povery. I has o be accompanied by an efficien provision of public capial and produciviy booss in he economic secors. This shor-run and long-run analysis is relevan for he Bolivian economy, because he governmen is using fiscal policy as one of is main ools o aack povery and aims o pu public invesmen as he foremos insrumens o promoe growh and welfare. Keywords: Fiscal Policy, Infrasrucure, Mulisecor Growh Model JEL Classificaion: E62, H54, O41 We would like o hank able research assisance. This work was carried ou wih financial and scienific suppor from he PEP resource person for heir useful commens and suggesions. We also hank Tim Kehoe and Juan Pablo Nicolini for all heir help during he Sudy Visi a he Federal Reserve Bank of Minneapolis. We are also indebed o Diego Vera for he Povery and Economic Policy (PEP) Research Nework, which is financed by he Ausralian Agency for Inernaional Developmen (AusAID) and by he Governmen of Canada hrough he Canadian Inernaional Developmen Agency (CIDA), and he Inernaional Developmen Research Cenre (IDRC). Insiue for Advanced Developmen Sudies (INESAD), cmachicado@inesad.edu.bo Insiue for Advanced Developmen Sudies (INESAD), paulesrada2@gmail.com Insiue for Advanced Developmen Sudies (INESAD), xflores@inesad.edu.bo -1-

2 1. INTRODUCTION During he early nineies here was a boom of papers in he economic lieraure analyzing expansionary fiscal policies based on ax cus or spending increases. In paricular mos of hem were aimed o analyze he fiscal adjusmens in a conex of economic crisis. Giavazzi and Pagano (1990) were he firs o argue ha fiscal adjusmens (defici reducions) large, decisive and on he spending side could be expansionary. Alesina and Peroi (1997) invesigae various episodes of fiscal adjusmens arriving a he conclusion ha fiscal simuli based upon ax cus are more likely o increase growh han hose based upon spending increases. 1 Recenly here has been a revival of his lieraure, in paricular in he developed counries and mainly in he Unied Saes, due o he Financial Crisis of and he recovery plans ha have been implemened which are based mainly in fiscal policy responses. Feldsein (2009) indicaes ha alhough in he pas here was a widespread consensus among economiss ha fiscal policy was no useful as a counercyclical insrumen, now governmens in Washingon and around he world are developing massive fiscal simulus packages, suppored by a wide range of economiss in universiies, governmens, and businesses. There has been a revival of he so called Fiscal Policy Mulipliers. 2 In Lain America and in Developing Counries he recenly lieraure has concenraed mainly in verifying he idea ha fiscal policy is procyclical, which is a puzzle ha has sparked a growing heoreical lieraure ha ries o explain i. Gavin and Peroi (1997) were he firs o call aenion o he fac, Talvi and Vegh (2005) claimed ha procyclical fiscal policy seemed o be he rule in all he developing world and recenly Ilzezki and Vegh (2008) using a baery of economeric ess, find overwhelming evidence o suppor he idea ha procyclical fiscal policy in developing counries is in fac a ruh and no a ficion. The common hing in all his lieraure is he fac ha all he analysis has concenraed in ax and expendiure policies and has lef aside public invesmen policies, in paricular hose policies based on public invesmen in infrasrucure. Infrasrucure becomes an imporan source of growh as shown by Aschauer (1989a, 1989b). These works concenraed on he esimaion of he producion elasiciies of governmen expendiure, using aggregaed daa for counries, mainly he 1 There is quie a rich lieraure ha sudies he deerminans and economic oucomes of large fiscal adjusmens. A non exhausive-lis includes Ardagna (2004), Giavazzi, Jappelli and Pagano (2000), Lamberini and Tavares (2000), McDermo and Wesco (1996), Von Hagen and Srauch (2001), Von Hagen, Hughes, and Srauch (2002), among ohers. 2 See Maunford and Uhlig (2008), Alesina and Ardagna (2009), Cogan, el.al. (2009), Ramey (2009), and Romer and Romer (2010), among ohers. -2-

3 U.S. 3 There are also cross-counry sudies ha emphasize he role of infrasrucure for a counry's growh. 4 Papers in his lieraure have ypically used regression analysis on eiher "growh accouning" or seady sae equaions. While hese papers have been useful in poining ou he imporance of infrasrucure, heir mehodology does no allow for he analysis of imporan general equilibrium feedback effecs among key macroeconomic variables and welfare. In his sense, his paper examines he impac of fiscal policy on oupu, consumpion, invesmen and foreign rade, using a Dynamic Sochasic General Equilibrium (DSGE) model for a small open economy wih five secors and wih he new feaure ha firms in each secor employ public capial or infrasrucure as a facor of producion. These secors which are he non-radable secor (services), he imporable secor (manufacuring), hydrocarbons, mining and agriculure are represenaive of he Bolivian economy. In paricular he hydrocarbons secor, which is inensive in capial, is conceived by he governmen as a sraegic secor ha will generae resources necessary o aack povery and underdevelopmen. In he paper, firs we analyze he macroeconomic and secoral impac of a change in governmen expendiures, ax srucure and public invesmen in infrasrucure on oupu, consumpion, invesmen, rade balance and welfare, i.e. he macroeconomic impacs of fiscal policy. Second, we idenify he combinaion of fiscal policy insrumens ha allow susaining a social policy based on governmen ransfers o households, and hird we show ha a fiscal policy solely is unable o generae he appropriae raes of oupu growh and welfare gains o reduce he povery levels. A combinaion of an effecive provision of public capial wih an increase in produciviy (TFP) is needed. In all he simulaions we provide quaniaive calculaions for he long-run as well as he dynamic ransiions for seleced cases. The DSGE model is based on Chumacero, Fuenes, & Schmid-Hebbel (2004) bu modified o include public invesmen in infrasrucure in a way similar o Rioja (2003) and secor division for he exporable secor as in Esrada (2006). We calibraed he model for he Bolivian economy and solved i using he second-order-approximaion echnique developed by Schmi-Grohé and Uribe (2004). The advanage in using his perurbaion mehod is ha i allows considering second-order effecs, which arise as imporan feaures in an economy wih high levels of uncerainy. An imporan aspec is ha he model allows us o exrac precise quaniaive implicaions, because we examine he effecs of a range of differen scenarios on real oupu and welfare, as well as on oher macroeconomic variables like consumpion, invesmen and on oupu of he differen five secors. Model simulaion resuls are repored firs, for seady-sae effecs and hen for he dynamic effecs on he composiion of hese variables. This is imporan if we consider ha 3 Munnell (1990) and García-Milá, McGuire and Porer (1993) use Panel Daa o esimae producion elasiciies. 4 See Easerly and Rebelo (1993), Ford and Pore (1991), Hulen (1996) and Canning (1998) among ohers. -3-

4 in he las years, he Bolivian governmen has based is policy o aack povery in ransfers o households hrough bonuses, and aims o pu public invesmen as he foremos insrumen o promoe growh and welfare. Cerainly, quaniaive measures of he impac of hese policies are needed o guide he policymakers. The paper is organized as follows. Secion 2 briefly describes fiscal policy in Bolivia. Secion 3 describes he dynamic general equilibrium model and is calibraion for he Bolivian economy. Then in secion 4 we presen he simulaion resuls for seady-sae effecs as well for he dynamic effecs on seleced macroeconomic and secoral variables. Secion 5 concludes. 2. FISCAL POLICY IN BOLIVIA Since 2006 he Bolivian economy has recorded a fiscal surplus, explained mainly by he inernaional economic boom, ranslaed ino high expor prices of he producs ha he counry expors, more revenues from he Direc Tax on Hydrocarbons and from royalies which increased public secor revenues in an imporan manner. Beween 2004 and 2008 governmen revenues increased from 28.5 percen of GDP o 48.4 percen of GDP. This increase of almos 20 percenage poins of GDP is huge in erms of GDP. 5 The following figure shows how hese evens affeced he pah of he fiscal defici as a percenage of GDP and is close relaionship wih he rae of growh of he economy. Since 2002 he rend of he fiscal defici was decreasing unil urning ino a surplus in The fiscal surplus was 4.5 percen of GDP, he highes in recen years, and he fiscal balance remained posiive unil 2009 inclusive. Figure 1: Fiscal Surplus and Economic Growh (in percenages) Source: Cenral Bank of Bolivia Defici/Surplus Fiscal Economic Growh 5 For comparison, he fiscal revenues of he U.S. federal governmen increased by 18.7 percen of GDP in he las 40 years (U.S. Congressional Budge Office, 2009). -4-

5 The growh rae of he economy has been seadily growing since 2002 unil 2008 and i is posiively correlaed wih he public secor fiscal resuls. In he period he average growh rae of he economy was 4.8 percen and in 2008 i was 6.1 percen, he highes rae of growh since In ha sense, he economy experienced an unprecedened scenario wih relaively imporan raes of growh and fiscal surpluses. This opens up a firs quesion: Can hese revenues be used o susain economic growh? On he oher hand, governmen expendiures as a share of GDP remained consan unil 2006 wih a slighly decrease in 2003, bu hen hey grew in he las 2 years. They represened 36 percen of GDP on average beween 2000 and 2006 and hen increased up o 42 percen and 45 percen of GDP in he years 2008 and 2009 respecively. In he following figure i can be seen ha in he las years governmen expendiures grew bu revenues grew in a larger magniude. Figure 2: Toal Expendiures and Revenues for he Non-Financial Public Secor (% of GDP) (p) Toal Revenues Toal Expendiures Source: Cenral Bank of Bolivia One explanaion for he increase in governmen revenues lies in he coninuous increase in ax revenues. Toal ax revenues increased by percen (average beween 2005 and 2008). The principal axes are he Value Added Tax (IVA) and he Direc Tax on Hydrocarbons (IDH), which ogeher represen 50 percen of oal ax revenues. In paricular he Direc Tax on Hydrocarbon s, which is a ax on he expor of hydrocarbons, displays he larges increase (52 percen on average), and his increase has been imporan in year 2006 which has been he year of he naionalizaion of he oil company (YPFB). -5-

6 Table 1: Tax Revenues (% variaion) (p) Average Average Value-added ax (IVA) 19.27% 21.75% 21.08% 23.24% -7.31% 21.33% 15.61% Transacions ax (IT) 8.77% 6.32% 14.87% 23.02% % 13.25% 7.52% Firm s profis ax (IUE) 41.75% 38.05% 6.49% 47.18% 59.30% 33.37% 38.56% Specific consumpion ax (ICE) 18.72% 18.08% 18.87% 19.64% 5.28% 18.82% 16.12% Complemenary regime-valueadded ax (RC-IVA) 10.65% 1.23% 0.72% 18.83% 11.54% 7.86% 8.59% Hydrocarbon s special ax (IEHD) 64.44% 6.04% 19.15% 6.18% % 23.95% 13.32% Financial ransacions ax (ITF) % % % 5.15% -0.48% 12.47% 9.88% Ohers % 8.90% 52.11% 32.29% 64.10% 6.99% 18.41% Toal Domesic Taxes 20.68% 16.63% 16.10% 24.79% 6.36% 19.55% 16.91% Direc ax on Hydrocarbons (IDH) % 8.32% 11.57% -2.68% 52.00% 38.33% Impor ariff (GA) 18.38% 15.74% 21.31% 25.99% % 20.35% 13.04% Toal Domesic Taxes + IDH + GA 41.25% 34.10% 14.32% 21.61% 3.26% 27.82% 22.91% Source: SIN-ANB (p) preliminary informaion unil Sepember 2009 Regarding he composiion of curren revenues of he Non-financial public secor, i can be seen in figure 3 ha ax revenues remain he main source of income, alhough non-ax revenues increased heir paricipaion due o he naionalizaion of he oil company (YPFB) in year Noice ha he hydrocarbon s ax was an imporan source of revenues in years 2005 and In 2007 and 2008 non-ax revenues replaced he hydrocarbon s ax revenues. Therefore, i seems ha i is no clear a all ha he naionalizaion of he oil company represened an addiional source of revenues as mos people argue. Furhermore, non-ax revenues also include he sales (domesic and foreign) of oher public companies ha have been naionalized or have been creaed. 6 Figure 3: Composiion of Curren Revenues of he Non-Financial Public Secor (% of oal) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% (p) Tax revenues Hydrocarbon's ax Non-ax revenues Curren ransfers Source: UDAPE Oher 6 Here we can menion he naionalizaion of Enel-he elecom company- in 2008, he creaion of BOA an aviaion company- in 2008 and he reacivaion of Comibol he mining company- in

7 Public invesmen was around 8 percen of GDP in 2006 and I decreased o 5 percen in 2008 and i increased again up o 8.5 percen of GDP in Figure 4 shows ha public invesmen is mainly concenraed in infrasrucure; he governmen invesed on average 3.8 percen of GDP in infrasrucure in he las 5 years. Social invesmen which comprises invesmen in healh, educaion, saniaion and urbanizaion is in second place wih an average of 2 percen of GDP in he las 5 years. I seems ha public invesmen in infrasrucure increased in line wih he growh of he economy, bu social invesmen remained almos unchanged. In recen years, governmen s capial spending has focused on road infrasrucure and waer resources, however since 2005 invesmen o suppor producive aciviies has again increased bu no up o he levels repored in Cerainly, invesmen in infrasrucure plays a crucial role in any developmen sraegy for Bolivia, since ranspor coss are very high, abou 20 imes higher han in Brazil, according o Weisbro e al. (2009). 9.00% Figure 4: Public Invesmen (percenage of GDP) 8.00% 7.00% 6.00% 5.00% 4.00% 3.00% 3.32% 1.43% 2.36% 1.04% 2.55% 2.01% 0.95% 1.06% 2.28% 2.15% 1.08% 1.14% 0.98% 0.68% 2.76% 0.99% 2.00% 1.00% 2.93% 2.88% 3.43% 3.39% 4.17% 4.16% 3.06% 4.03% 0.00% Infrasrucure Producion Exracive Social Source: Cenral Bank of Bolivia and Minisry of Economy and Public Financing Bolivia s povery reducion sraegy is based solely in he condiional ransfers ha he governmen gives o households. The amoun ransferred o households as a share of GDP has increased in he las 2 years. In fac, hey increased from 0.7 percen of GDP in 2006 o 2.3 percen of GDP in In figure 5, we display also he pensions ha he governmen had o ransfer o households due o he change in he pension sysem from a pay-as-you-go sysem o a fullyfunded sysem ha occurred in In any case, i is imporan o have in mind ha hese -7-

8 ransfers represen an imporan pressure for he fiscal budge and need o be managed efficienly. 7 Figure 5: Curren Transfers of he Non-Financial Public Secor (% of GDP) (p) Pensions Source: Cenral Bank of Bolivia Transfers o households Finally, figure 6 shows he financing of he fiscal defici in he las years. As i is common in budge accouning we observe he ne exernal credi and he ne domesic credi. During he years of a fiscal defici ( ) we observe an increase in he ne exernal credi unil 2002, bu hen i decreased coninuously unil The ne domesic credi decreased also seadily from 2002 unil 2006 and here i is imporan o menion ha here has been a subsiuion beween foreign and domesic deb, wih domesic deb becoming more imporan han foreign deb since Mos of he domesic deb is wih he so called AFP s (Pension Funds Adminisraors) and wih he Cenral Bank. However, he governmen has been able o reduce is deb wih he Cenral Bank, due o he fiscal surpluses experienced in he las years. Therefore we observe negaive values for he ne domesic credi since (2.0) (4.0) (6.0) (8.0) (10.0) Figure 6: Financing of he Fiscal Defici (% of GDP) (p) Global Defici Ne Exernal Credi Ne Domesic Credi Source: Cenral Bank of Bolivia 7 There are currenly hree ypes of condiional ransfers: Rena Dignidad (for persons over 60 years of age), bono Juancio Pino (for sudens in primary school) and bono Juana Azurduy (for mohers during and afer pregnancy). -8-

9 In sum, here is a new scenario for he fiscal policy in Bolivia, wih increasing revenues bu also wih new responsibiliies for he public secor no only in erms of he Sae Owned Enerprises (SOE s), bu also in erms of he social policy ha he governmen is commied o susain o reduce he povery levels in Bolivia. Of course his fiscal policy should be compaible wih a susainable growh and improving levels of welfare for people. These issues will be analyzed in he following secions. 3. MODELLING AND CALIBRATION 3.1 THE BASIC MODEL The model is based on Chumacero, Fuenes, & Schmid-Hebbel (2004) and modified in order o address he issue of public invesmen in infrasrucure as in Rioja (2003) and expanded for a mulisecor economy as in Esrada (2006) The Households The economy is inhabied by infiniely-lived individuals who derive uiliy from consumpion of imporable goods (c m, ), consumpion of non-radable goods (c n, ) and governmen consumpion (g ) which is basically a public good ha does no suffer from congesion. Therefore, a represenaive agen maximizes he expeced value of lifeime uiliy as given by =0 E β u( c (1) m,, cn,, g ) The oher goods ha are produced in he economy are exporable goods which we denoe xh as he hydrocarbon good (naural gas), xm as he mineral good (zinc, gold, silver or in) and xa as he agriculural good (soya, Brazilian nus or quinoa). Each household receives ineres income rk, lump-sum ransfers from he governmen Γ, profis from he imporable, non-radable, hydrocarbons, mineral and agriculural firms π m, π n, π xh, π xm and π xa respecively 8 and can also conrac foreign deb abroad, b. The household's budge consrain is (1 + τ )(1 + τ ) c m (1 τ ) rk k c + π m, xh, + (1 + τ ) p + π xm, c + π n, xa, c n, + π + (1 + τ )(1 + τ ) i + (1 + r~ ) b m, + π m n, + b +1 c + Γ (2) where τ m is an impor ariff, τ k is he ax rae on capial income, τ c represens he ax rae on consumpion of imporables and non-radables, p n is he relaive price on he non-radable good in erms of he imporable good (used as numeraire) and r ~ is he (ne) ineres rae paid on foreign 8 Profis can be inerpreed also as he reribuion o he labor facor, since we are assuming ha labor is secor specific. -9-

10 deb. Privae invesmen, which we denoe by i, follows he sandard law of moion for privae capial: k = i + (1 ) k + 1 δ (3) where δ is he depreciaion rae of privae capial sock and k is he capial sock. Household make choices of c m,, c n,, b +1 and k +1, i.e he problem of he represenaive consumer can be summarized by he following Bellman equaion: v( k, b ) = βv( k + 1, b + 1 ) + u( c The firs-order condiions are: m,, c n, (1 + τm )(1 + τc ) cm, + (1 + τc ) pn, cn, + (1 + τm )(1 + τc )( k+ 1 (1 δ ) k ) + (1 + r~ ) b, g ) λ (1 τ k ) rk πxh, πxm, πxa, πm, πn, ) b + 1 Γ (4) p u cn, = (1 + τ ) (5) u n, m cm, u cm, + 1 β E + ~ (1 r ) (6) u cm, 1 = + 1 u cm, + 1 (1 τ k ) βe ( r + 1 δ ) (7) u cm, (1 + τm )(1 + τc ) 1 = + 1 Equaion (5) saes ha he relaive price beween imporables and non-radables mus equae he raio of marginal uiliies beween boh goods. The nex wo ineremporal condiions are he sandard Euler equaions ha indicae ha he marginal rae of subsiuion beween consumpion oday and omorrow, mus equae heir relaive price, evaluaed a he cos of foreign borrowing and he rae of reurn of capial invesmen, respecively The Firms This economy is represened by five secors which are: imporables (manufacuring), nonradables (services), hydrocarbons, mining and agriculure. In each secor here is an equal number of represenaive firms ha require privae capial k and public capial k g o produce heir goods. Firms do no direcly own capial, hey simply ren k each period from households a he domesic marke renal rae of capial r. Public capial is provided freely by he governmen. We -10-

11 assume ha labor is secor specific, which means ha labor canno move across secors. The imporable secor is in fac a domesic secor ha produces impor subsiues. 9 The firm s problem is saic in his framework and secoral profis are given by he following equaions: * m, = ( 1+ τm ) f ( zm, km,, k ) rkm, π (8) * n, = p n, f ( zn, kn,, k ) rkn, π (9) * xh, = ( 1 τxh ) q xh, f ( zxh, kxh,, k ) rkxh, π (10) * xm, = q xm, f ( zxm, kxm,, k ) rkxm, π (11) * xa, = q xa, f ( zxa, kxa,, k ) rkxa, π (12) where π i are he profis of each secor i, z i is a producive shock in secor i, k i is he amoun of privae capial demanded in secor i and q i is he relaive price of good i in erms of he imporable good. Public capial is he same for all secors. The only difference is he inensiy of usage of public capial in each secor. There is also a ax on he producion of hydrocarbons which is denoed by τ xh. By maximizing he above profis direcly wih respec o he relevan capial sock we obain he following firs-order condiions: (1 + τ ) f ( z, k, k ) = r (13) m km m, m, * p f ( z, k, k ) = r * n, kn n, n, (14) (1 τ ) q f ( z, k, k ) = r (15) xh xh, xh xh, xh, * q f ( z, k, k ) = r * xm, xm xm, xm, (16) q f ( z, k, k ) = r * xa, xa xa, xa, (17) These equaions describe he demand for capial services by firms of each producion secor of he economy. 9 Impors and impor subsiues are perfec subsiues, his means ha hey should be sold a he same price, herefore he domesic price of y m, is equal o (1+τ m ). -11-

12 3.1.3 The Governmen The governmen invess in infrasrucure I, has curren expendiure consumpion g and provides lump-sum ransfers o households Γ. The Governmen has he following budge consrain: g + Γ + I = τ ( c τ y + τ rk + τ q y (18) c m, + pn, cn, + i ) + τm (1 + τc )( cm, + i ) m m, k xh xh, xh, Public capial evolves according o k g, 1 = I + ( 1 δg ) kg, + (19) where 0 k g 1 is a consan depreciaion rae of public capial. As in Rioja (2003) we assume ha only he effecive measure of he sock of public capial k g is useful for privae producion. Tha is, * k θk g, = (20) where 0<θ<1 is an infrasrucure effeciveness index. The closer θ is o 1, he more effecive he public capial sock, and he larger he benefi ha firms ge. As usual, he governmen does no opimize any explici objecive funcion bu insead is curren expendiure follows he rule: g 2 ρ vg, + 1 ~ N(0, σg ) (21) + 1 = ( 1 g ) g + ρgg + vg, The Foreign Secor An open economy has differen properies han a closed economy. The main difference is ha, if i is possible o impor goods and capial, he economy having a low iniial capial sock would prefer o run a curren accoun defici in he firs periods, susain a high level of consumpion and pay laer he res of he world wih a curren accoun surplus. This is somehing ha cerainly does no happen in realiy and makes he soluion of he model difficul, because i induces muliple equilibrium for each differen pah of deb, and non-saionariy of he variables. 10 Schmi-Grohé and Uribe (2003) sugges five modificaions o he sandard small open economy model wih incomplee asse markes o induce saionariy. We use he modificaion ha employs a deb-elasic ineres-rae premium, which has been used also by Bhandari e. al. (1990), Turnovsky (1997), and Osang and Turnovsky (2000) See Fernandez de Cordoba and Kehoe (2000) for an applicaion for Spain. 11 Oher modificaions imply a model wih an endogenous discoun facor, wih convex porfolio adjusmen coss, wih complee asse markes, and wihou saionariy-inducing feaures. -12-

13 Therefore we assume ha he counry faces an upward-sloping supply schedule for deb, reflecing he degree of risk associaed wih lending o he economy. This is expressed by assuming ~ + ha he borrowing rae r 1, charged on foreign deb is of he form: ~ r b * + 1 = (1 ρr ) r + (1 ρr ) ϕ + ρrr + vr, + 1 y ω ~ ω ' > 0, ω' ' > 0 (22) where r * is he exogenously given world ineres rae and ϕ(b /y ) ω is he counry-specific risk premium ha increases wih he sock of deb as a share of oupu. Two elemens are fundamenal in his specificaion. Firs, he convexiy of he funcion is a convenien way of incorporaing he ceiling on borrowing suggesed by Eaon and Gersoviz (1981). Second, he ype of AR(1) specificaion ha incorporaes uncerainy explains why we need o employ a sochasic model. Oherwise, by using a non-sochasic specificaion he model sill presens he failure menioned by Fernandez de Cordoba and Kehoe (2000). The relaive prices of he exporable goods in erms of he imporables, i.e. he erms of rade are assumed o have he following law of moion: q 2 ρ vqxi, + 1 ~ N(0, σqxi ) (23) xi, + 1 = ( 1 qxi ) qxi + ρqxiqxi, + vqxi, + 1 where i=xh, xm and xa Marke-Clearing Condiions We define he producion funcion of any of he secors by: * y = f ( z, k, k ) (24) i, i, i, where i again represen each of he five secors. Remark ha he public capial k * is he same for all he secors, which means ha infrasrucure will benefi all he secors in he same manner. Public capial is a non-rival good. Equaions (24) and (25) represen he marke clearing condiions. The firs equaion describes he equilibrium in he imporable good marke, which shows ha he curren accoun (CA) balance mus be compensaed by he capial accoun balance. The second equaion is he ypical equilibrium condiion in he non-radable good marke. CA r~ + b (24) ( b 1 b ) = qxh, yxh, + qxm, yxm, + qxa, yxa, + ym, cm, g i I and p nyn, pn, cn,, = (25) -13-

14 3.1.6 Compeiive Equilibrium A compeiive equilibrium is a se of allocaion rules c m =C m (s), c n =C n (s), k+₁=k(s), b+₁=b(s), k+₁ =K (s), k xh+1 =K xh (s), k xm+1 =K xm (s), k xa,+1 =K xa (s), k m,+1 =K m (s), k n,+1 =K n (s); a se of pricing funcions r=r(s), and p n =P n (s); and he laws of moion of he exogenous sae variables s+₁=s(s), such ha: Households solve he problem (4) aking as given s and he form of he funcions R(s), P n (s), and S(s), wih he equilibrium soluion o his problem saisfying c m =C m (s), c n =C n (s), k+₁=k(s), and b+₁=b(s). Firms of he hydrocarbons, mining, agriculure, imporable and non-radable secors maximize profis (13)-(17) aking as given s and he form of he funcions R(s), P n (s), and S(s), wih he equilibrium soluions o hese problems saisfying k xh,+1 =K xh (s), k xm,+1 =K xm (s), k xa,+1 =K xa (s), k m,+1 =K m (s), k n,+1 =K n (s) and k+₁ =K (s). The economy-wide resource consrains (24) and (25) hold each period, and he facor marke clears: 3.2 FUNCTIONAL FORMS AND CALIBRATION K xh (s)+ K xm (s)+k xa (s)+k m (s)+k n (s)=k(s) The model which is clearly non-linear, is difficul o solve analyically. The alernaive is o use numerical mehods. Therefore, we adop funcional forms for he uiliy and producions funcions and give values o he parameers of he model o mach he Bolivian macroeconomic conex in year Funcional Forms The generic model presened above suggess he following funcional form for preferences: u ( cm, cn,, g ) = θ m ln( cm, ) + θn ln( cn, + µ g, wih θ m, θ n >0 and θ m +θ n =1. The parameer µ measures how a ypical individual values public consumpion relaively o privae consumpion. The specificaion for he relaionship beween privae consumpion of non-radable goods and public consumpion follows Aschauer (1985), Barro (1981) and Chrisiano and Eichenbaum (1992). We are assuming ha consumpion of public goods can be subsiued (depending on he value of µ) wih consumpion of non-radable goods and vice versa. 12 For he producion funcions we employ he following specificaion: ) 12 I is reasonable o hink ha when individuals wan o increase heir consumpion of healh for example, hey will sacrifice consumpion of non-radables like haircus. -14-

15 f ( z * i,, ki,, k ) = z i, k αi i, ( k ) * φi where α i is he compensaion for capial as a share of oupu of secor i=xh, xm, xa, m and n and φ i is he coefficien of public capial in he producion funcion ha reflecs he imporance of infrasrucure in each of he differen five secors of he economy. The produciviy shocks z i follow sandard AR(1) processes of he form: z 2 ρ v ~ N(0, σ ) i, + 1 = ( 1 i ) z + ρizi, + vi, + 1 i, + 1 i Calibraion Once he laws of moion are specified, we accuraely calibrae he model so ha i can display he main characerisics of he Bolivian economy. We are considering 2006 as our base year and he daa used is quarerly. In able 2, we display he parameers of he model, which we assume, for now, ha are invarian o changes in economic policies. The firs column of able 2 shows he deep parameers of preferences. The subjecive discoun facor β was se o make i consisen wih a percen annual rae a which Bolivians can borrow (r ~ in our model). The parameers θ m and θ n are calibraed so as o reproduce he share of oal consumpion over GDP in seady sae, where we define oal consumpion as consumpion in imporables plus consumpion in non-radables imes is relaive price. We se µ=0.5 as a benchmark, implying here is imperfec subsiuion beween privae and public consumpion. Table 2: Parameers Preferences Prod. Funcions Technology Shocks Fiscal Variables Exogenous Prices β=0.975 δ=0.028 (yearly) ρ xh =0.72 g=0.18 q xh =0.174 θ m = α xh =0.66 ρ xm =0.53 ρ g = q xm =0.14 θ n = α xm =0.25 ρ xa =0.45 σ g =0.01 q xa =0.19 µ=0.5 α xa =0.19 ρ m =0.4 δ g =2*δ ρ qxh =0.87 α m =0.58 ρ n =0.51 θ=0.613 ρ qxm =0.91 α n =0.38 σ xh =0.014 τ m =0.1 ρ qxa =0.86 Source: Auhor s calculaions φ xh =0.25 σ xm =0.009 τ c =0.13 σ qxh =0.017 φ xm =0.14 σ xa =0.011 τ k =0.13 σ qxm =0.01 φ xa =0.12 σ m =0.017 τ xh =0.32 σ qxa =0.11 φ m =0.07 σ n =0.03 ϕ=0.248 φ n =0.25 z xh =0.53 ρ r = z xm =0.72 σ r = z xa =0.57 r * =0.048 z m =0.15 ω=1.2 z n =

16 The second column describes he deep parameers of he producion funcions. The depreciaion rae of privae capial δ has been se o 2.8 percen per year, which was obained by calibraing he privae invesmen. The oupu-facor elasiciies α in each secor were obained in he following manner: We have reduced he 35 secors ha represen he Bolivian economy in he 2006 inpu-oupu marix, o 6 secors ha represen agriculure, hydrocarbons, mining, imporables, non-radables and infrasrucure. In paricular, we are considering as infrasrucure secors: i) energy, gas and waer, ii) ranspor and sorage and iii) communicaions. Then, we have used he value-added decomposiion in facor paymens for 1996 (he only year available) and impued hese shares for our secoral value-added of The corresponding calculaions are shown in able A.1 of he appendix. Key parameers are he infrasrucure shares in each secor. We have used also our inpu-oupu marix disaggregaion, bu here we employed he inermediae consumpion of infrasrucure in each of he five secors of he model. In oher words, we have calculaed he φ s as he share of inermediae consumpion of infrasrucure in agriculure, mining, hydrocarbons, imporables and non-radables. Recall ha public capial is conceived assumed o be a free good in he model, herefore i seems srange o calibrae he share parameers of each secor using inermediae consumpion which is expendiure. We have solved his concern by assuming ha he governmen is giving subsidies o he privae secor in order o produce he public goods. Alhough he governmen is providing he public capial, i is being produced by he privae secor and in some manner is usage is being also paid by he firms. The corresponding calculaions are shown in able A.2 of he appendix. 13 The hird column conains he TFP parameers. These parameers have been calibraed o mach he closer as possible he share of oupu of each secor over GDP. The auorregresive coefficiens and volailiies of he shocks were se so as o mach he auocorrelaion of oupu and sandard deviaions of residuals of AR(1) regressions for he secoral oupus. 14 The forh column shows he governmen parameers and fiscal variables. The parameers of he governmen expendiure AR(1) process have been obained by performing a simple OLS regression, bu he parameer g has been calibraed in order o mach he governmen expendiure over GDP. The rae of depreciaion of public capial δ g has been esimaed by he World Bank o be abou wice as high as he rae of depreciaion of privae capial. The benchmark effeciveness parameer θ is esimaed here based on daa of he so called "Loss Indicaors." In paricular we have employed loss indicaors of power, elecom, roads and waer. Then, he Bolivian loss index across infrasrucure ypes is calculaed by aking a weighed loss and comparing i wih he weighed average of indusrialized counries. The calculaions are shown in able A.3 of he appendix. According o hese calculaions, Bolivia has a level of effeciveness of 13 In fac a more correc specificaion of he producion funcion should be: * αi * φi f ( zi,, ki,, k ) = zi, ki, ( xk ) where x represens privae inermediae consumpion 14 This parameer values are imporan o adjus he speed of convergence o he seady sae. -16-

17 61.3 which means ha infrasrucure in Bolivia is 39 percen less effecive han in he developing counries. 15 The ax raes describe he Bolivian ax sysem. The consumpion ax τ c is approximaed by he Value Added ax (IVA) which is 13%. The ax on capial income τ k corresponds o he Complemenary Regime Value Added ax (RC-IVA) which has also a rae of 13%. The ax on hydrocarbons τ xh has a rae of 32%, which corresponds o he Direc Tax on Hydrocarbons (IDH for is iniials in Spanish). Finally, he impor ariff τ m represens he average ariff for all he impored producs; i has a value of 10%. Finally in column 5 of able 3 we display he so called exogenous prices. All of hem follow sandard laws of moion and mos of heir parameers were esimaed using OLS regressions. We calibraed he consan erms of he AR(1) specificaions of hese relaive prices using he respecive index prices calculaed by he Bolivian Cenral Bank. Finally, we calibraed ϕ equal o o mach a raio of exernal deb over GDP equal o , which is consisen wih he capial accoun balance in seady sae. This value for ϕ combined wih a value of ω equal o 1.2 gives a counry risk value equal o RESULTS In his secion we perform differen simulaions wih key parameers of he model o quanify he effecs of fiscal policy on several macroeconomic variables like oupu, consumpion, invesmen, among ohers. We disinguish he long run effecs from he shor run dynamics. The long run effecs are obained by comparing he seady saes of he model under he baseline scenario and under he simulaions scenario. The shor run dynamic effecs require ha we impose iniial condiions, solve he model (find he policy funcions of he conrol variables and he laws of moion of he endogenous sae variables), and characerize he ransiion o he new seady sae STEADY STATE COMPARISONS In his sub-secion we presen long run effecs resuls or seady sae changes in: consumpion of each good (c m, c n ), physical producion in each secor (Y xh, Y xa, Y xm, Y m, Y n ), he reciprocal of he exchange rae (p n ), governmen lump sum ransfers as a fracion of oupu (Γ, as a proxy for 15 We use he same weighs as in Rioja (2003), his means 0.40, 0.10, 0.25, 0.25 for power, elecom, paved roads and waer sysems respecively. In developing counries he effeciveness index θ is normalized o 1, which means ha infrasrucure is highly effecive. 16 According o our specificaion, he policy funcions of he conrol variables canno be obained analyically and we have o resor o numerical mehods. We used he Schmi-Grohé and Uribe (2004) second-order approximaion echnique. This perurbaion mehod has been proven superior o he radiional linear-quadraic approximaions. -17-

18 addiional pressures on he governmen budge), privae invesmen (i), public invesmen (I), oal real consumpion (C), oal real oupu (Y) and welfare compensaion (TU). 17 Recall ha he aim of his paper is o find ou he opimal fiscal policy in erms of growh, and social ransfers, herefore we have ordered he long run analysis of he resuls in he following manner: Tax Policy We analyze a fiscal policy based solely in changes of he ax raes for he four axes considered in our model. By reducing and increasing he impor ariff, we analyze a more or less opened economy. For insance 0 percen ariff can be inerpreed as a fully opened economy, while a 20 percen ariff can be ranslaed ino a less world linked economy. In he Value-Added ax simulaion we simulae an increase in he ax rae up o he Lain America s average. According o Oalora (2009) he Lain America s average is percen. In he capial and hydrocarbon s ax simulaions we perform a 10 percen increase and decrease of hese ax raes. Table 3 displays he resuls of he ax policy simulaions. Table 3: Change in Seady Sae Values from a Tax Policy (in percenages) Variables τ m τ m τ c τ k τ k τ xh τ xh τ m, τ c τ k, τ c τ xh, τ c Simulaed Value , , , c m c n Y xh Y xa Y xm Y m Y n p n Γ/GDP i I C Y TU Source: Auhor s calculaion. 17 To absrac from changes in relaive prices, oal consumpion (C) and oal oupu (Y) are measured a he iniial baseline prices. TU is defined as he subsidy (ax if negaive) in erms of consumpion of imporables, non radables and public services ha would be needed o compensae (ake from) he consumer in order for him o be indifferen beween he siuaion before and afer he change is made. -18-

19 Noice ha a fully opened economy (column 1) allows he economy o grow by 3.3 percen, wih welfare increasing by 3.86 percen. The reducion in he price of he capial good (imporable) raises he value of he marginal produciviy in he exporable and non-radable secors, while keeping i consan in he imporable secor. Therefore Y xh, Y xa, Y xm and Y n increase and Y m decreases by percen. Consisen wih he ariff reducion, he real exchange rae depreciaes by -5 percen. The opposie effecs occur when here is an increase in he impor ariff (second column). Indeed, mos of he resuls are in he same magniude bu wih negaive sign, excep in ransfers as a share of GDP. By opening he economy, ransfers are reduced by -8.1 percen, as a resul of he decrease in governmen revenues, bu by closing he economy ransfers increase only by 1.54 percen. An increase in he value-added ax (hird column), alhough very beneficial for ransfers as a share of GDP, which increase by 17.6 percen, is no so good in erms of oupu. Toal real oupu increases only by 0.13 percen and all he secors oupu increase by less han 1 percen wih he excepion of hydrocarbons which grows by 1 percen and imporables which decreases by percen. Despie he endogenous increase in he lump sum ransfer, he ne welfare effec is almos null (-0.02 percen) which is explained mainly by he decrease in consumpion of imporables (-0.51 percen). Privae invesed is adversely affeced also, i decreases by percen, bu cerainly public invesmen is largely benefied, i can be increased by 3.89 percen. The lieraure on opimal fiscal policy saes ha capial axes should be zero. Here, a reducion in he capial ax promoes aggregae capial accumulaion and his increases he amoun of capial used in each secor, and hus oupu in each secor. The secor ha mos benefis from a reducion in capial ax is he hydrocarbon secor, is oupu increases by 2.47 percen. Cerainly, his happens because he hydrocarbon secor is inensive in capial (α larger han 0.5). The oher secor ha is inensive in capial which is manufacuring (imporables) grows by 2 percen. Noice ha he effecs of an increase or a decrease in capial ax are almos linear. When capial ax increases by 10 percen, aggregae oupu decreases by -1.6 percen while i increases by 1.56 percen when capial ax decreases by 10 percen. 18 Producion in he hydrocarbon secor is highly sensiive o variaions in he ax on he producion of his secor. Noice ha a 10 percen decrease in his ax, increases oupu in his secor by percen, while in he opposie simulaion, i decreases by percen. Wha is ineresing from his simulaion is precisely ha governmen revenues are increased, because oupu in he hydrocarbon secor increases a lo. Even hough he ax rae is reduced (column 6), ransfers over GDP are increased by percen, which is a magniude similar o he increase in oupu of he hydrocarbons secor. Consumpion of radable and non-radable goods is also posiively affeced, which allows he governmen o collec more revenues hrough he consumpion ax. 18 See Chamley (1986) and Chari, Chrisiano and Kehoe (1994) for references on opimal fiscal policy. -19-

20 Finally, in he las hree columns of able 4, we simulaed hree combined scenarios, where ransfers over GDP remain consan in comparison o he baseline scenario. This is o analyze he needed increase in consumpion or value-added ax o compensae he negaive effecs of hose ax policies ha showed negaive effecs on ransfers over GDP. Among he hree combined scenarios, he bes is he scenario ha combines a fully opened economy (τ m =0) wih a 3.72 percen increase in he value-added ax. The scenario ha combines a decrease in he capial ax wih an increase in he value-added ax is also good in he sense ha he impac on all variables is posiive alhough low, whereas he scenario ha combines an increase in he hydrocarbons ax wih an increase in he value added ax is bad, because he impac on all variables is negaive. From hese simulaions, we can conclude ha he bes Bolivia can do in erms of a ax policy is o liberalize he economy by reducing he ariffs. This policy allows he economy o grow by 3.3 percen and o experience welfare gains of 3.86 percen. If in addiion, he governmen wans o mainain is social policy based on ransfers o households, i should increase he value-added ax. This combines policy allows he economy o grow by 3.41 percen and o experience welfare gains of 3.91 percen Expendiure and Invesmen Policy Nex, we analyze a fiscal policy based solely in public expendiures and public invesmen. Recall ha in our seing, public consumpion includes everyhing ha is no invesmen; his means ha healh and educaion expendiure is considered in his variable among oher expendiures like wages and benefis for public workers. Firs, we simulae a 10 percen increase in public expendiure. Second, we simulae an increase in public invesmen as a share of oal governmen revenues. We simulae an increase of his share equal o he average of years 2007and Finally, we combine a 10 percen increase in public expendiure wih a 10 percen increase in public invesmen. -20-

21 Table 4: Change in Seady Sae Values from an Expendiure and Invesmen Policy (in percenages) Variables g I/Y g g, I/Y g Simulaed Value , c m c n Y xh Y xa Y xm Y m Y n p n Γ/GDP i I C Y TU Source: Auhor s calculaion. People always expec o find a posiive effec on oupu from an increase in governmen expendiure or public invesmen. Cross-counry empirical sudies have generally found ha public infrasrucure has posiive effecs on a counry's producive performance (e.g, Easerly and Rebelo (1993), Canning and Fay (1993), Canning (1999)) and his is indeed he case of Bolivia. An increase in public invesmen as a share of governmen revenues by 3.26 percen (average of he las wo years) increases oupu and public invesmen by 1.58 percen and 5.19 percen respecively. However, he opposie happens when he governmen increases is curren expendiures. An increase by 10 percen of governmen expendiures decreases oupu by percen and consumpion by percen. Cerainly, an increase in governmen expendiure represens an imporan pressure for fiscal balance, since ransfers as a share of GDP have o be reduced by percen. The combined scenario allows us o idenify ha ransfers o households are very sensiive o an increase in governmen expendiures; hey remain high and negaive, whereas aggregae oupu, consumpion and invesmen show posiive signs. In paricular public invesmen shows an imporan increase (11.69 percen). Noice also ha in he firs and hird columns he real exchange rae depreciaes, however only in he combined scenario his depreciaion has a posiive effec on oupus of he exporable secors. In he case where only governmen expendiure increases, he negaive sign on all secoral oupus is explained by he reducion in public invesmen, which affecs public capial and herefore oupu in all secors. In conclusion, beween a public expendiure policy and a public invesmen policy, he governmen should choose he laer. This resul opens anoher quesion: If we are considering -21-

22 ha educaion and healh is included in he variable g, does his mean ha i is bad for he governmen o increase hese expendiures? This quesion will be answered in he nex exercises Tax and Expendiure Policy Le s sar answering he quesion saed above by combining he 10 percen increase in governmen expendiures wih an increase in he wo main axes ha are available for he Bolivian governmen, he value-added ax and he hydrocarbon s ax. In paricular we aim o calculae he needed variaion in ax raes in order o compensae for he negaive effec of governmen expendiures on ransfers o households. Table 5: Change in Seady Sae Values from a Tax and Expendiure Policy (in percenages) Variables g, τ c g, τ c and τ xh Simulaed Value 0.198, , and c m c n Y xh Y xa Y xm Y m Y n p n Γ/GDP i I C Y TU Source: Auhor s calculaion. The resuls in able 5 indicae ha o compensae he negaive effec on ransfers as a share of GDP from a 10 percen increase in governmen expendiures an increase of 25 percen in he consumpion ax is needed. This is cerainly a huge increase and i is refleced in he negaive effecs on growh, consumpion, privae invesmen and welfare. In fac, only public invesmen displays a posiive reacion by increasing by 6.95 percen. Therefore, in he second column we combine he 10 percen increase in governmen expendiures, wih an increase in he value added ax up o he Lain America s average and wih a decrease in he hydrocarbon s ax. We find ou ha his is indeed a beer policy, since all he variables display a posiive reacion. In paricular noice ha oupu in he long run grows by 5.48 percen, which is he bes rae of growh observed in all he simulaions performed unil now. Cerainly his imporan impac is driven by he percen increase in hydrocarbon s oupu, which is he main secor of he Bolivian economy. -22-

23 4.1.4 Fiscal Policy complemened wih Efficiency and Produciviy In his subsecion we analyze fiscal policy combined wih wo variables ha are no direcly relaed o fiscal policy, bu he governmen can in some manner provide he incenives o promoe hem. They are firs efficiency, which is relaed o an improvemen in he way public invesmen (in infrasrucure) is provided, and second produciviy, which is relaed o an improvemen in Toal Facor Produciviy in all he secors. According o he Naional Developmen Plan (PND for is iniials in Spanish) in he period , Bolivia should have reached an average annual growh in oupu of 6.3 percen. To aain his overall rae of growh, he secors should have grown (annually) by 3.1 percen (agriculure), 6.8 percen (imporables), 18.8 percen (non-radables), 13.2 percen (hydrocarbons) and 10.4 percen (mining). We have re-calibraed he TFP parameers in each secor in order o aain hese raes of growh in each of he secors. As we are using quarerly daa, he TFP parameers have been calibraed for quarerly raes of growh. 19 Rioja (2003), shows ha raising effeciveness has sizable posiive effecs on privae invesmen, consumpion and welfare. We simulaed a 5 percen increase in he level of effeciveness of he exising infrasrucure nework, i.e. he value of θ increased up o The firs wo columns in able 6 show he effecs of increasing TFP and θ, while he nex columns show five combined exercises where we found ou which should be he bes fiscal policy he governmen should follow in Bolivia. 19 The PND expresses he economic and planning sraegy ha he governmen will follow in he nex years o consolidae he process of ransformaion of he economy. -23-