Achieving a Soft Landing: The Role of Fiscal Policy

Transcription

1 WP/8/69 Achieving a Sof Landing: The Role of Fiscal Policy Daniel Leigh

2

3 28 Inernaional Moneary Fund WP/8/69 IMF Working Paper FAD Achieving a Sof Landing: The Role of Fiscal Policy Daniel Leigh Auhorized for disribuion by Benedic Clemens and Seven Symansky March 28 Absrac This Working Paper should no be repored as represening he views of he IMF. The views expressed in his Working Paper are hose of he auhor(s) and do no necessarily represen hose of he IMF or IMF policy. Working Papers describe research in progress by he auhor(s) and are published o elici commens and o furher debae. This paper uilizes an open-economy New Keynesian overlapping generaions model o assess he exen o which fiscal policy, along side an inflaion-forecas-based moneary policy, could enhance macroeconomic sabiliy in Colombia. The model simulaions indicae ha, in addiion o sabilizing oupu and inflaion, a sronger response of he fiscal balance o excess ax revenue would reduce he burden on he cenral bank of adjusing ineres raes, lessen he associaed degree of exchange rae volailiy, and conribue o a more sable exernal curren accoun balance. The analysis also assesses how he success of fiscal policy in enhancing macroeconomic sabiliy depends on he ype of shock, he response of moneary policy, and he lengh of fiscal policy implemenaion lags. JEL Classificaion Numbers: E3, E62, E63 Keywords: Business cycle sabilizaion; Fiscal policy; Moneary policy Auhor s Address: DLeigh@imf.org

4 2 Conens Page I. Inroducion...3 II. Mehodology...4 III. Resuls: Macroeconomic Effecs of Fiscal Policy Tighening...9 IV. Resuls: Fiscal Policy and Macroeconomic Sabiliy...4 V. Conclusions...2 References...22 Figures Figure. Domesic Risk Premium and he Deb-o-GDP Raio...9 Figure 2. Exogenous Permanen Cu in Public Consumpion of.5 Percen of GDP... Figure 3. Exogenous Permanen Cu in Public Invesmen of.5 Percen of GDP...3 Figure 4. Privae Savings Shock and Srengh of Fiscal Policy Response...5 Figure 5. Markup Shock and Srengh of Fiscal Policy Response...6 Figure 6. Efficiency Fronier: Markup Shock...7 Figure 7. Efficiency Fronier for Differen Moneary Policy Response Parameers...8 Figure 8. Temporary Fall in Risk Premium by Basis Poins...9 Figure 9. Risk Premium Shock and Srengh of Fiscal Policy Response...2

5 3 I. INTRODUCTION. Wha are he macroeconomic effecs of fiscal policy, and how can fiscal policy help achieve a sof landing for an overheaed economy? While a consensus regarding he role and effecs of moneary policy has emerged, here is less agreemen regarding fiscal policy. A number of cenral banks have developed large scale models o predic he effecs of moneary policy, bu, as discussed in several recen papers, hese models ofen canno adequaely replicae he dynamic effecs of fiscal policy found in he empirical lieraure. 2 Convenional models used for moneary policy ypically feaure agens wih infinie planning horizons, and predic ha fiscal policy is ineffecive in influencing aggregae demand and exernal curren accoun balances. 2. Assessing he macroeconomic effecs of fiscal policy is relevan for Colombia, given evidence of overheaing pressures. Real GDP growh is esimaed a 6.8 percen in 26, and Fund saff projec he same growh for 27. Average CPI inflaion is expeced o rise percenage poins o 5½ percen in 27, despie an increase in he Banco de la República s (BdR) policy ineres rae by 35 basis poins from 6 percen in April 26 o 9½ percen in November 27. Overheaing pressures are also eviden in he increase in he exernal curren accoun defici, which is projeced o double o abou 4 percen of GDP in 27. Oupu is esimaed a abou 3 percen above poenial as of he second quarer of 27. While he ighening of moneary policy is expeced o help reduce growh in 28 o abou 5 percen, a posiive oupu gap is neverheless projeced o remain in ha year. 3. This paper uilizes an open-economy srucural model o assess he exen o which fiscal policy could enhance macroeconomic sabiliy in Colombia. The model, developed a he Inernaional Moneary Fund (IMF), is called he Global Inegraed Moneary and Fiscal (GIMF) model (Kumhof and Laxon, 27a). Ricardian equivalence does no hold in he model as consumers are assumed o have finie lifeimes and lifecycle income. GIMF reflecs he laes advances in new open economy macroeconomic heory, and embodies a number of nominal and real rigidiies ha permi i o make empirically plausible predicions regarding he dynamic effecs of fiscal and moneary policy. 4. The paper provides quaniaive assessmens of he effecs of changes in fiscal policy on key macroeconomic aggregaes, such as real aciviy and he curren accoun, using GIMF. In addiion, he paper assesses how he effecs of macroeconomic shocks depend on he response of fiscal policy. In paricular, he analysis focuses on assessing (i) how a sronger fiscal policy response o he cyclical posiion can reduce he burden on moneary policy in responding o shocks; (ii) o wha exen a sronger response of fiscal Recen papers on he new consensus in moneary policy include Goodfriend (27), and Mishkin (27). 2 See, for example, Kumhof and Laxon (27a), and Gali, López-Salido and Valles (27). Examples of largescale models used for moneary policy analysis include he Banco de la República s Model of Transmission Mechanisms (MMT), he IMF's FPAS and GEM, he Federal Reserve's SIGMA, and he ECB's NAWM.

6 4 policy o he cyclical posiion can induce an inflaion-oupu volailiy radeoff; and (iii) how he effeciveness of fiscal policy in enhancing macroeconomic sabiliy depends on he response of moneary policy, and on fiscal policy implemenaion lags. 5. The remainder of he paper is srucured as follows. Secion B presens he key feaures of he model. Secions C and D repor he resuls, and Secion E concludes. The model II. METHODOLOGY 6. GIMF is an open economy general equilibrium model developed a he Fund ha is equipped for boh moneary and fiscal policy analysis. The model s muliple non- Ricardian feaures, nominal and real rigidiies, moneary policy reacion funcion, and fiscal policy reacion funcion yield plausible macroeconomic responses o changes in fiscal and moneary policy. 7. Ricardian equivalence is assumed no o hold for four reasons: Firs, he model feaures overlapping generaion agens (OLG) wih finie lifeimes. These agens are myopic in he sense ha hey perceive deb-financed ax cus as an increase in heir human wealh, and aach a low probabiliy o having o pay for hem in he fuure. Second, workers have a life-cycle labor produciviy paern ha implies a declining rae of produciviy as workers age. This feaure means ha workers discoun he effecs of fuure payroll ax increases as he laer are likely o occur when hey are older and less producive. Third, he model conains liquidiy-consrained consumers (LIQ) who do no have access o financial markes o smooh consumpion, and change heir consumpion one-for-one wih changes in afer-ax income. 3 Finally, he model includes payroll and capial income axes ha are disorionary because labor effor and privae invesmen respond o relaive price movemens ha resul direcly from variaions in ax raes. 3 These consumers do solve an inraemporal opimizaion problem for choosing consumpion and leisure levels. However, wihou access o financial markes, hey canno smooh consumpion in response o emporary changes in disposable income.

7 5 8. Imporanly, GIMF relaxes he convenional assumpion ha all governmen spending is waseful and does no conribue o aggregae supply. Insead, GIMF allows for producive public infrasrucure spending ha adds o he public capial sock, and enhances he produciviy of privae facors of producion. Real rigidiies embedded in he model include consumer habis ha induce consumpion persisence, invesmen adjusmen coss ha induce invesmen persisence, and impor adjusmen coss. Nominal rigidiies include sicky inflaion Phillips curves in each secor of he economy The governmen deermines how he fiscal balance-o-gdp raio responds o excess ax revenue using a simple fiscal policy rule. The rule for he fiscal balance-o-gdp raio is of he following form: fbal gdp τ τ = φ + d gdp () where fbal is he fiscal balance-o-gdp raio. If he response parameer d =, he fiscal gdp balance is kep equal o φ a all imes. For example, if d = and he economy experiences an upswing wih acual ax revenue τ exceeding seady-sae ax revenue τ, he fiscal balance remains unchanged, and he excess ax revenue is spen. Such a response corresponds o a balanced budge rule and is here defined as procyclical. A response of d < would also qualify as procyclical. As he response parameer d increases in he posiive range, a greaer share of he excess ax revenue is saved. When d =, a percenage poin of GDP increase in excess ax revenue ranslaes ino a percenage poin increase in he fiscal balance, a response consisen wih a srucural balance rule. The rule can be implemened by adjusing axes or spending. A response of d > implies ha a percenage poin of GDP increase in excess ax revenue induces an improvemen in he fiscal balance of more han percenage poin of GDP, and is, for he purposes of his paper, defined as being counercyclical.. The cenral bank arges inflaion by manipulaing he nominal ineres rae following a sandard inflaion forecas-based (IFB) rule. The specificaion of his moneary policy rule is consisen wih he IFB rule embodied in he BdR s MMT, as described by Gómez, Uribe, and Vargas (22), and López (23), and is of he following form: i ( μ i ) μ π μ μ π i + 4 i i ( r π + 4 ) E π = (2) For furher deails regarding household preferences and firm echnology in he model s raded and non-raded secors, see Kumhof and Laxon (27a).

8 6 where he gross policy ineres rae is i, he inflaion forecasing horizon is 4 quarers, he inflaion arge, π is for oal 4-quarer gross inflaion, π + 4 = π + π + 2π + 3π + 4, and E μ, denoes denoes expecaions based on informaion available a ime. 5 Coefficien [ ) he degree of nominal ineres rae ineria. If μ i =, equaion (2) implies ha when he inflaion forecas exceeds he arge by percenage poin, he nominal ineres rae increases by + μ. Moneary policy responds o oupu, bu only o he exen ha i is relevan for π forecasing inflaion. The equilibrium real ineres rae r is endogenous, and is deermined by he global marke for loanable funds, as well as a counry-specific risk premium.. Given he imporance of risk premiums in emerging markes, and heir possible relaionship wih fiscal policy, he model includes an endogenous counry-specific risk premium. In paricular, he risk premium on he ineres paid on domesic deb is denoed ρ and eners he model via a risk-adjused uncovered ineres pariy (UIP) equaion for foreign currency bonds: 6 i i RW = i E ε + ( + ρ ) (3) where i RW is he gross nominal ineres rae in he res of he world, and ε + denoes fuure gross nominal exchange rae depreciaion. 7 The domesic risk premium ρ is assumed o have he following non-linear form: δ ρ = δ + (4) 2 δ max ( deb / gdp) ( deb / gdp )) 3 If δ = 2, hen he risk premium always equals he exogenous level δ, regardless of he level of he deb-o-gdp raio deb / gdp ). If δ, a decline in governmen deb reduces ( 2 > max he risk premium. As he deb-o-gdp raio rises owards he level ( deb / gdp), he risk premium rises a an increasing rae. The assumpion of an increasing slope is broadly consisen wih empirical sudies ha find a posiive linear relaionship beween he logarihm of he risk premium and he deb raio, such as Arora and Cerisola (2). Such 5 The gross rae equals one plus he rae. For example, an inflaion rae of 3 percen corresponds o a gross rae of.3. 6 There are wo financial asses in he model, privae bonds ha are raded inernaionally, and governmen bonds ha are subjec o complee domesic bias. 7 If he risk premium ρ =, an expeced depreciaion of he domesic currency by percen is, via arbirage, associaed wih an increase in he domesic ineres rae by abou percenage poin above he res-of-he-world ineres rae.

9 7 esimaes imply ha he level of he risk premium grows a an increasing rae as he deb raio rises. The parameer δ deermines he curvaure of he risk premium funcion.8 Calibraion 3 > 2. The model is calibraed o conain wo counries, Colombia and he res of he world. Each period corresponds o one quarer. Colombia is assumed o comprise.3 percen of world GDP, and o have a seady sae inflaion rae of 3 percen per year. The res of he world is assumed o have a seady sae inflaion rae of 2 percen per year. The seady-sae rae of echnological progress is assumed o be 2 percen per year, populaion is assumed o grow a percen per year, and he real ineres in he res of he world is assumed o be 3 percen per year in he iniial seady sae. The srucural parameers regarding household preferences and firm echnology are se following Kumhof and Laxon (27b), who calibraed GIMF for he case of Chile and he res of he world. In paricular, he parameers ha govern he degree of household myopia, a key non-ricardian feaure of he model, are calibraed as follows. Households in boh Colombia and he res of he world are assumed o have a planning horizon of 5 years, i.e., a probabiliy of deah of 6.7 percen per year, and a decline in lifecycle worker produciviy of 5 percen per year. Fify percen of Colombian households are assumed o be liquidiy consrained. This proporion is he same as ha assumed for Chile by Kumhof and Laxon (27b), and is larger han ha assumed for he Unied Saes by Kumhof and Laxon (27a), 33 percen. Given ha financial developmen is lower in Colombia han in he Unied Saes, a larger share of liquidiy consrained households in Colombia seems plausible Fiscal parameers, such as he raios o GDP of governmen ransfers, purchases of goods and services, and public invesmen are calibraed based on daa from he Colombian auhoriies. The produciviy of public capial is calibraed following Lighar and Suárez (25) who presen a mea analysis of a large number of sudies (for OECD counries including Mexico) on he elasiciy of aggregae oupu wih respec o public capial, and esimae his elasiciy a.4. Accordingly, he model is calibraed so ha a percen real increase in public invesmen is associaed wih a long-run increase in real GDP ne of depreciaion of abou.4 percen. The depreciaion of public capial is se a 4 percen per year. 4. Regarding he parameer ha governs he fiscal policy response o he business cycle, d, his paper considers a range of values. In paricular, he analysis evaluaes macroeconomic and exernal curren accoun volailiy for values of d ranging from he srongly procyclical response of d =. 5 o he srongly counercyclical response of 8 While he risk premium funcion is assumed o be coninuous, a sudden, disconinuous change in he risk premium could be simulaed by changing he exogenous componen of he premium δ. 9 A fully saisfacory calibraion of hese parameers would require he model o be esimaed.

10 8 d = 4.5. For he purposes of his paper, he fiscal rule is implemened by adjusing payroll, consumpion, and capial income ax raes by he same proporion. 5. In he absence of business cycle shocks, he fiscal balance is assumed o equal he value ha sabilizes he deb-o-gdp raio a he projeced end-27 level of 38 percen. In paricular, in he seady sae, here is a one-o-one correspondence beween he fiscal defici-o-gdp raio and he governmen deb-o-gdp raio ha depends on he rae of nominal GDP growh, i.e.: fdef gdp NG = + NG deb gdp (5) where NG denoes he seady sae nominal growh rae, and fdef denoes he fiscal defici. For example, if he seady sae nominal growh rae is 6 percen, and he seady sae deb-o- GDP raio is 38 percen, hen he seady sae deb-sabilizing fiscal defici equals 2.2 percen of GDP ( φ = 2. 2 ). A decline in he fiscal defici by.5 percenage poins of GDP would be associaed wih a long-run decline in he deb-o-gdp raio of 8.8 percenage poins. 6. The moneary policy response funcion is calibraed in line wih previous work on moneary policy in Colombia and oher counries. In paricular, he baseline calibraion of he inflaion response parameer is μ =. 5. An inflaion response of.5 is in he range of π coefficiens found o be efficien for Colombia,.-2., (in erms of minimizing a weighed funcion of inflaion, oupu, and ineres rae volailiy) by Lopez (23) using sochasic simulaions of he BdR s MMT model. For sensiiviy analysis, he paper also considers a less aggressive inflaion response of μ π =.25, and a more aggressive response of μ = 4. The nominal ineres rae ineria parameer is se a μ =. 5, a value consisen π wih empirical evidence for a number of counries. 7. Regarding he elasiciy of he domesic risk premium o changes in governmen deb, calibraion is complicaed by he scarciy of empirical evidence for emerging marke counries. Based on a sample of daa for 6 emerging marke including Colombia, Rowland and Torres (24) find ha a percenage poin increase in he deb-o-gdp raio is associaed wih an increase in he risk premium, proxied by he EMBI spread, of basis poins afer conrolling for a number of oher explanaory variables. 2 Guided by his esimae, he baseline calibraion of his paper implies ha, saring from he end-27 i The long-run decline in he deb-o-gdp raio of 8.8 is found as follows: 8.8 =.5 (.6/.6). See, for example, Clarida, Gali, and Gerler (998) who esimae an ineres rae ineria parameer of abou.8 using monhly daa for Germany, Japan, and he Unied Saes. This value corresponds o abou.5 a a quarerly frequency (.8 3 =.5). 2 See Rowland and Torres (24), Tables 5.3a and 5.3b.

11 9 posiion, a percenage poin increase in he deb-o-gdp raio raises he risk premium by 8 basis poins. For example, raising he deb raio from he end-27 expeced level of 38 percen of GDP by 5 percenage poins o 43 percen of GDP he end-26 level increases he risk premium by abou 4 basis poins. This calibraion is broadly consisen wih he observed change in he Colombia EMBI spread from 26 o The assumed degree of curvaure of he risk premium funcion is illusraed in Figure. Figure. Domesic Risk Premium and he Deb-o-GDP Raio Risk Premium in Percen p.a Governmen Deb in Percen of GDP III. RESULTS: MACROECONOMIC EFFECTS OF FISCAL POLICY TIGHTENING 8. This secion repors he resuls of a fiscal ighening experimen in which he fiscal balance is improved permanenly by.5 percenage poins of GDP. In he base case, he governmen implemens he consolidaion by cuing public consumpion by.5 percenage poins of GDP, and he sensiiviy analysis explores how he resuls depend on he composiion of he adjusmen. In each case, he resuls are repored in erms of he deviaion from he baseline, i.e., he seady sae ha would occur if he fiscal consolidaion were no implemened. 3 The Colombia EMBI spread declined from an average level of 96 basis poins during 26 o an average of 59 basis poins during January Sepember 27, a change of 4 basis poins, alhough a number of facors besides Colombia s deb-o-gdp raio probably conribued o his change.

12 Base case: a cu in governmen consumpion 9. A fiscal consolidaion based on a cu in governmen consumpion is associaed wih a shor-run decline in aggregae demand, bu a medium-erm increase in GDP. The baseline experimen implemens a permanen cu in real governmen consumpion by 3 percen (he equivalen of.5 percen of GDP), which improves he overall fiscal balance-o- GDP raio by.5 percenage poins on impac (Figure 2). As deb and he associaed cos of ineres obligaions decline, axes are reduced o keep he overall fiscal balance unchanged a.5 percenage poins of GDP above he iniial posiion. The improvemen in he fiscal balance, as well as he reduced ineres cos, reduces he governmen deb-o-gdp raio by evenually 8.8 percenage poins. 2. The cu in governmen consumpion reards aggregae demand, alhough he effec is shor-lived. In paricular, real GDP declines by.58 percen on impac, a response ha implies a shor-run muliplier of.5 4. This response is broadly consisen wih he effec of governmen spending shocks esimaed for Colombia using a Bayesian VAR approach (Abrego and Öserholm, 27). 5 Moneary policy responds o he associaed decline in expeced inflaion by reducing ineres raes, which simulaes privae spending, and conribues o a real depreciaion. This moneary policy simulus reduces he disinflaionary effec of he fiscal conracion, while he resuling improvemen in ne expors increases ne foreign asses as overall Colombian savings rise. Consequenly, he curren accoun improves by.36 percenage poins of GDP wihin he firs year, illusraing a srong reversed win defici phenomenon. Noe ha in convenional infinie horizon models, he fiscal ighening would have a negligible effec on he curren accoun as he increase in public savings would be offse by lower privae savings. 4 The shor-run muliplier of.5 is obained by dividing he impac on GDP (.58 percen) by he cu in governmen consumpion in percen of GDP (.5 percen). 5 The response is also consisen wih Blanchard and Peroi (22) who esimae a governmen spending muliplier beween.9 and.29 using a mixed srucural VAR/even sudy approach for he Unied Saes ha idenifies fiscal shocks based on insiuional informaion (see Table 4 of heir paper). While he predicions of GIMF as calibraed for he purposes of his paper are in line wih empirical sudies such as Blanchard and Peroi (22), hey differ from he esimaes of some oher sudies, such as Giavazzi and Pagano (99) ha find negaive mulipliers for some European counries.

13 Figure 2. Exogenous Permanen Cu in Public Consumpion of.5 Percen of GDP (In Deviaion from Seady Sae Baseline) 2. Over he medium run, he savings generaed by lower governmen consumpion and lower ineres paymens permi reducions in payroll and capial income axes. These ax reducions, and he decline in risk premiums due o he lower deb-o-gdp raio have posiive effecs on labor supply, invesmen, GDP, and consumpion. 6 Using he baseline calibraion, a permanen improvemen of he fiscal balance by.5 percen of GDP 6 Using he addiional fiscal space o increase producive public invesmen raher han o reduce axes also resuls in long-run oupu gains. Resuls are no shown here, bu are available on reques.

14 2 ha reduces he deb-o-gdp raio by, evenually, 8.8 percenage poins raises real GDP in he long-run by.56 percen (Figure 2). Sensiiviy analysis: composiion of fiscal ighening 22. Variaions on he baseline experimen reveal ha he long-run posiive effecs of fiscal consolidaion depend srongly on he composiion of he adjusmen. In paricular, implemening he consolidaion by cuing producive public invesmen by.5 percenage poins of GDP can jeopardize he long-run gains. As Figure 3 repors, a permanen cu in public invesmen leads o an aggregae demand conracion of.62 percen on impac, bu GDP now also declines in he long run by.86 percen due o a conracion in he economy s supply capaciy. If public invesmen is no producive, however, fiscal adjusmen based on cus in capial expendiure has a broadly similar effec as a reducion in public consumpion. 23. Similarly, he shor-run conracionary effecs of he consolidaion depend on is composiion. When he fiscal consolidaion is implemened by raising axes or cuing ransfers, he shor-run conracion in GDP is smaller and more gradual han when he burden of ighening falls on governmen purchases. The difference in magniude sems from he fac ha afer-ax disposable income and ransfers can be spen by households on boh domesic and foreign oupu, while governmen purchases have a srong domesic bias. The more gradual effec resuls from he assumpion ha households in he model have consumpion habis, and respond gradually o a change in afer-ax disposable income. In paricular, a payroll ax increase of.5 percen of GDP is found o lower GDP by.39 percen afer hree quarers, compared o a.58 percen decline in GDP on impac when public consumpion is cu. 7 An increase in consumpion axes has a smaller negaive effec, reflecing he broader ax base, wih GDP falling by.24 percen in he firs year following a.5 percen-of-gdp increase in consumpion axes. A.5 percen of GDP cu in lump-sum ransfers leads o a decline in GDP by.7 percen wihin he firs year. 7 The size of he effec of changes in axes, which implies a muliplier of.78 afer hree quarers, is consisen wih he empirical esimaes of Blanchard and Peroi (22) for he Unied Saes, who esimae a ax muliplier beween.78 and.3 (see Table 3 of heir paper), bu is smaller han ha of Romer and Romer (27).

15 3 Figure 3. Exogenous Permanen Cu in Public Invesmen of.5 Percen of GDP (In Deviaion from Seady Sae Baseline)

16 4 IV. RESULTS: FISCAL POLICY AND MACROECONOMIC STABILITY 24. Having discussed he dynamic effecs of fiscal policy in he model, his secion invesigaes how he impac of macroeconomic shocks depends on he srengh of he fiscal policy response. In paricular, he secion quanifies how he effeciveness of a fiscal policy response in conribuing o macroeconomic sabiliy depends on he ype of shock, he srengh of he moneary policy response, and he lengh of fiscal policy implemenaion lags. Regarding he ype of shocks considered, he analysis focuses on hree ypes ha have differen implicaions for oupu and inflaion: (i) a demand shock due o a change in privae savings; (ii) a supply shock due o a change in firms markups; and (iii) a risk premium shock. Demand shock: a reducion in privae savings 25. Following a demand shock due o a reducion in privae savings, a sronger response of fiscal policy helps o sabilize boh oupu and inflaion, reducing he burden on moneary policy for sabilizing inflaion. Figure 4 repors he dynamics of GDP, inflaion, ineres raes, exchange raes, and he curren accoun following a reducion in privae savings ha is due o an exogenous increase in he consumer rae of ime preference. 8 The increase in privae spending raises inflaion, and resuls in a deerioraion in he exernal curren accoun balance. Moneary policy ighens in response o he higher inflaion, inducing a real appreciaion of he currency, and furher worsening he curren accoun balance. For each variable, Figure 4 repors how he effecs of he shock depend on how cyclical fiscal policy is. A sronger response of he fiscal balance o excess ax revenue (a move from a small o a larger d) implies, due o he effec of fiscal policy on aggregae demand, ha smaller ineres rae increases are required o sabilize inflaion. Consequenly, he exchange rae appreciaes less, and he exernal curren accoun deerioraes less and hen converges more smoohly owards he seady sae. 8 The shock involves a 4 percenage poin increase in he consumer rae of ime preference, and is assumed o follow an AR() process wih a persisence coefficien of.9, implying a half-life of abou 7 quarers.

17 5 Figure 4. Privae Savings Shock and Srengh of Fiscal Policy Response (Deviaion from Seady Sae Baseline).5 GDP.25 Inflaion.5 Nominal Ineres Rae d Percenage poins Real Exchange Rae CA Balance/GDP Fiscal Balance/GDP Quarers Noe: darker lines in he figure correspond o a sronger response of he fiscal balance o excess ax revenue (larger values of parameer d). Supply shock: an increase in firms markups 26. Following a supply shock due o an increase in markups ha shifs inflaion and oupu in opposie direcions, a fiscal policy response o he excess ax revenue induces an inflaion-oupu volailiy radeoff. Figure 5 repors he dynamics of GDP, inflaion, ineres raes, exchange raes, and he curren accoun following a emporary increase in firms markups. The source of he markup shock is assumed o be an increase in he degree of marke power of firms in he disribuion secor, which resuls in a resricion of aciviy, and an increase in firms prices. 9 The decline in real demand for goods and services in response o he markup shock implies an increase in privae savings, and an improvemen in he exernal curren accoun balance. A he same ime, moneary policy ighens in response o he inflaionary effec of he markup shock. A sronger response of he fiscal balance o he cyclical posiion (a move from a smaller o a larger d) implies a larger fiscal loosening, 9 The supply shock involves a 5 percenage poin increase in he markup of prices over marginal cos in he disribuion secor, and is assumed o follow an AR() process wih a persisence parameer of.9, implying a half-life of abou 7 quarers.

18 6 which moderaes he decline in oupu, bu slighly raises inflaion. However, he oupuinflaion volailiy radeoff associaed wih a fiscal policy response is modes under he baseline calibraion of he model which assumes a srong moneary policy response o inflaion expecaions. In addiion, a sronger easing in fiscal policy during he downurn moderaes he improvemen in he curren accoun balance. Figure 5. Markup Shock and Srengh of Fiscal Policy Response (Deviaion from Seady Sae Baseline) GDP.2 Inflaion 2 Nominal Ineres Rae d Percenage poins Real Exchange Rae CA Balance/GDP Fiscal Balance/GDP Quarers Noe: darker lines in he figure correspond o a sronger response of he fiscal balance o excess ax revenue (larger values of parameer d). 27. To quanify he inflaion-oupu volailiy radeoff, an efficiency fronier is consruced. The efficiency fronier idenifies he policies ha minimize a weighed average of inflaion and oupu volailiy given he model and shock srucure. The weighed average, denoed l, akes he following form: l = σ(inflaion) + λσ(gdp) (6) where σ denoes he roo mean squared deviaion abou he seady sae, and λ is he weigh on oupu. Figure 6 shows he efficiency fronier in inflaion-oupu volailiy space, where

19 7 moving souhwes implies a decline in boh inflaion and oupu volailiy. 2 The slope of he efficiency fronier indicaes ha, for he range of fiscal response parameers considered ( d [.5,4.5] ), a percen decline in oupu volailiy is associaed wih an increase in inflaion volailiy of abou.2 percen. Noe ha in he case of he demand shock, a sronger fiscal policy response (a move from a low value of d o a large value of d) would, in Figure 6, imply moving souhwes reducing boh oupu volailiy and inflaion volailiy. Figure 6. Efficiency Fronier: Markup Shock.6 Roo Mean Sandard Deviaion of Inflaion d = d = d = Roo Mean Sandard Deviaion of Oupu 28. However, he inflaion-oupu volailiy radeoff is sensiive o he srengh of he moneary policy response o inflaion expecaions. In paricular, a weaker moneary policy response o inflaion expecaions subsanially increases he inflaionary consequences of an accommodaive fiscal response o he supply shock. To illusrae his poin, he supply shock experimen is repeaed wih wo alernaive moneary policy rules: (i) a weak moneary policy response corresponding o a value of μ π =. 25 in he policy rule; and (ii) a more aggressive moneary policy response corresponding o μ π = 4. As Figure 7 repors, he efficiency fronier is wice as seep wih he weak moneary policy response, a resul ha underlines he imporance of a srong commimen by he cenral bank o sabilizing inflaion expecaions. In addiion, for any given fiscal policy response, a weaker moneary policy response implies a higher level of inflaion volailiy. 2 The efficiency fronier is consruced as follows: (i) he markup shock is simulaed muliple imes, each ime wih an alernaive fiscal rule parameer d; (ii) for a given value of λ, he policy rule (and associaed inflaionoupu volailiy pair) ha minimizes he funcion l is idenified; and (iii) sep (ii) is repeaed for a range of λ values (from o 3 in seps of.).

20 8 Figure 7. Efficiency Fronier for Differen Moneary Policy Response Parameers Roo Mean Sandard Deviaion of Inflaion Moneary policy response μ π = 4 Moneary policy response μ π = Roo Mean Sandard Deviaion of Oupu Risk premium shock 29. An exogenous fall in he risk premium induces a real appreciaion, an increase in invesmen and consumpion, and a deerioraion in he exernal curren accoun balance. The experimen involves an exogenous basis-poin fall in he exogenous risk premium componen δ, and is assumed o follow an AR() process wih a persisence parameer of.9. Figure 8 repors he dynamic responses of key macroeconomic variables wih no change in he fiscal balance (d = ). In paricular, he decline in he risk premium implies an increase in he risk-adjused reurn on Colombian bonds, which, via ( + ρ ) arbirage Equaion (3), raises demand for Colombian asses and induces a domesic currency appreciaion. The lower risk premium also lowers he ineres rae on borrowing from he res of he world, which simulaes consumpion, boh of domesic and of foreign oupu. In addiion, he lower risk premium reduces he cos of capial, and raises he profiabiliy of privae capial, which simulaes invesmen. Due o quadraic invesmen adjusmen coss, he response of invesmen is hump-shaped. Overall, he rise in privae spending demand worsens he exernal curren accoun balance. Over he medium run, sabilizing ne foreign liabiliies requires an improvemen in he rade balance, an adjusmen ha implies an exchange rae depreciaion. The iniial appreciaion also has a disinflaionary effec, which promps an easing in moneary policy. The disinflaionary pressure is in par off-se by inflaionary pressures associaed wih he increase in aggregae demand, and he overall response of inflaion is herefore modes. i ρ

21 9 Figure 8. Temporary Fall in Risk Premium by Basis Poins (In Deviaion from Seady Sae Baseline) 3. Faced wih he risk premium shock, and he associaed expansion in aciviy, a sronger fiscal policy response dampens he deerioraion in he exernal curren accoun balance, and sabilizes oupu. As Figure 9 repors, a sronger response of he fiscal balance o he increase in fiscal revenue (a move from a smaller o a larger d) moderaes he increase in oupu, and slighly lowers inflaion. In addiion, he igher fiscal

22 2 posiion moderaes he exchange rae appreciaion, and reduces he deerioraion in he exernal curren accoun balance. Figure 9. Risk Premium Shock and Srengh of Fiscal Policy Response (Deviaion from Seady Sae Baseline.8 GDP.8 Inflaion Nominal Ineres Rae d Percenage poins Real Exchange Rae CA Balance/GDP Fiscal Balance/GDP Quarers Noe: darker lines in he figure correspond o a sronger response of he fiscal balance o excess ax revenue (larger values of parameer d). Sensiiviy analysis: lengh of fiscal policy implemenaion lags 3. The capaciy of fiscal policy o conribue o macroeconomic sabiliy depends on he speed wih which i can respond. To illusrae his poin, he fiscal response funcion in Equaion () is alered o incorporae adjusmen lags as follows: fbal gdp L L τ τ = φ + d gdp L (7) where L denoes he implemenaion lag in quarers. The fiscal surplus now responds wih a lag of L quarers o changes in excess ax revenue. The resuls (no shown, bu available on reques) sugges ha, for implemenaion lags of or 2 quarers, a counercyclical fiscal policy coninues o enhance macroeconomic sabiliy, alhough by less han in he absence of implemenaion lags. However, when he fiscal response is delayed by 3 quarers or more, an aggressive response o excess ax revenue (d > ) can be couner-producive, and desabilize oupu.

23 2 V. CONCLUSIONS 32. This paper quanifies he dynamic effecs of fiscal policy using a srucural model, GIMF, and finds ha while fiscal consolidaion reards aggregae demand in he shor run, i can also yield long-run oupu gains. The shor-run slowdown is smaller when he consolidaion is based on ransfer cus han when i involves cus in governmen purchases, and when he cenral bank responds srongly o he associae easing of inflaion pressures. In paricular, he conracion of oupu wihin one year following a.5 percen-of- GDP increase in he fiscal surplus is esimaed o reduce GDP by.7.62 percen wihin one year, depending on he composiion of he fiscal ighening. The long-run gains accrue due o lower risk premiums ha crowd in privae aciviy, and are paricularly srong if he savings from lower deb-ineres paymens are used o lower disorionary axes or o increase producive public invesmen. The long-run oupu gains from a permanen improvemen in he fiscal balance by.5 percenage poins of GDP is esimaed a.56 percen of GDP when he consolidaion is based on cus in governmen consumpion. However, a fiscal consolidaion based on producive public invesmen cus alone can jeopardize such long-run gains, and reduce long-run oupu. 33. The paper also finds ha fiscal policy can subsanially conribue o a smooh landing for an overheaed economy. In addiion o sabilizing oupu and inflaion, a sronger response of he fiscal balance o excess ax revenue reduces he burden on he cenral bank of raising ineres raes and lessens he associaed degree of exchange rae appreciaion. The sronger response of he fiscal balance also conribues o a more moderae deerioraion in he exernal curren accoun balance during expansions in domesic demand. A sronger fiscal response also moderaes he deerioraion in he curren accoun balance in response o a fall in risk premiums, highlighing he sabilizing role fiscal policy can play following exernal shocks. 34. Moreover, he analysis finds ha he success of fiscal policy in enhancing macroeconomic sabiliy depends on he ype of shock, he response of moneary policy, and he lengh of fiscal policy implemenaion lags. In paricular, while a sronger fiscal policy response can lower boh oupu and inflaion volailiy during aggregae demand shocks, a supply shock inroduces an inflaion-oupu volailiy radeoff. This radeoff is modes if moneary policy is srongly commied o sabilizing inflaion expecaions. Regarding fiscal policy implemenaion lags, he analysis suggess ha wih lags of up o 2 quarers, less procyclical policy sill enhances macroeconomic sabiliy given he persisence of business-cycle shocks considered here. However, an aggressive counercyclical response ha is delayed by 3 quarers or more can be couner-producive, and desabilize oupu.

24 22 REFERENCES Abrego, Lisandro, and Pär Öserholm, 27, Exernal Linkages and Economic Growh in Colombia: Insighs from a Bayesian VAR Model, forhcoming IMF Working Paper (Washingon: Inernaional Moneary Fund). Arora, Vivek, and Marin Cerisola, 2, How Does U.S. Moneary Policy Influence Sovereign Spreads in Emerging Markes? Saff Papers, Inernaional Moneary Fund, Vol. 48, No. 3. Bayoumi, Tamim, and Silvia Sgherri, 26, Mr. Ricardo s Grea Advenure: Esimaing Fiscal Mulipliers in a Truly Ineremporal Model, IMF Working Paper 6/68 (Washingon: Inernaional Moneary Fund). Blanchard, Oliver, and Robero Peroi, 22, An Empirical Characerizaion of he Dynamic Effecs of Changes in Governmen Spending and Taxes Oupu, Quarerly Journal of Economics, Vol. 7, pp Clarida, Richard, Jordi Gali, and Mark Gerler, 998, Moneary Policy Rules in Pracice: Some Inernaional Evidence, European Economic Review, Vol. 42, pp Gomez, Javier, José Darío Uribe, and Hernando Vargas, 22, The implemenaion of Inflaion Targeing in Colombia Borradores Semanales de Economía, No. 22. Goodfriend, Marvin, 27, How he World Achieved Consensus on Moneary Policy, NBER Working Paper No. 358 (Cambridge, Massachuses: Naional Bureau of Economic Research). Kumhof, Michael, and Douglas Laxon, 27a, A Pary Wihou a Hangover? On he Effecs of U.S. Governmen Deficis, IMF Working Paper No. 7/22 (Washingon: Inernaional Moneary Fund)., 27b, Chile s Srucural Fiscal Balance Rule: A Model-Based Evaluaion (unpublished manuscrip; Washingon: Inernaional Moneary Fund). Lighar, Jenny E., and Rosa M. Marin Suárez, 25, The Produciviy of Public Capial: A Mea Analysis, manuscrip, Universiy of Groningen, and Tilburg Universiy. López, Marha, 23, Efficien Policy Rules for Inflaion Targeing in Colombia Borradores Semanales de Economía, No. 24. Mishkin, Frederic S., 27, Will Moneary Policy Become More of a Science? NBER Working Paper No (Cambridge, Massachuses: Naional Bureau of Economic Research). Romer, David, and Chrisina D. Romer, 27, The Macroeconomic Effecs of Tax Changes: Esimaes Based on a New Measure of Fiscal Shocks, NBER Working Paper No (Cambridge, Massachuses: Naional Bureau of Economic Research).

25 23 Rowland, Peer, and Jose L. Torres, 24, Deerminans of Spread and Crediworhiness for Emerging Marke Sovereign Deb: A Panel Daa Sudy, Borradores de Economía, Banco de la República, Bogoá. Giavazzi, Francesco, and Marco Pagano, 99, Can Severe Fiscal Conracions be Expansionary? Tales of Two Small European Counries, NBER Macroeconomic Annuals 5, pp. 75 (Cambridge, Massachuses: Naional Bureau of Economic Research).