Fiscal Policy in CEE Countries. Evidence from Czech Republic, Hungary, Poland and Romania 1,2 Dr. Petre Caraiani 3*

Transcription

1 Fiscal Policy in CEE Counries. Evidence from Czech Republic, Hungary, Poland and Romania, Dr. Pere Caraiani 3* I invesigae he fiscal policy in CEE counries using evidence from he four mos imporan economies, Czech Republic, Hungary, Poland and Romania. A wo-counry open economy model wih a Taylor fiscal rule is esimaed on quarerly daa for hese counries. I explore he poenial of couner-cyclical fiscal policy in he conex of he ongoing financial crisis, he reacion of he fiscal policy o negaive demand shocks or o a more relaxed moneary policy, as well as he impac of fiscal shocks. I show ha he fiscal rule can ac as an auomaic sabilizer in he conex of negaive domesic and exernal demand shocks. The resuls also show ha in hese counries he fiscal policy can posiively influence he oupu when negaive shocks from Euro Area affec he economy. There are differences wih respec o he reacions of macroeconomic variables in hese economies, wih respec o boh he magniude and persisence of responses. Based on he hisorical decomposiion, here are evidences ha fiscal shocks during he las years behaved in a pro-cyclical way and i appears ha he counercyclical poenial of fiscal policy during he financial crisis remained largely unused. Running counerfacual scenarios confirms ha a considerable governmen spending effor would have improved he dynamics of GDP during he crisis for hree of he counries, namely Czech Republic, Hungary and Romania,. Using he Bayesian comparison approach, I also found evidence in he favor of including he fiscal Taylor rule agains he alernaive of simple AR() processes for he fiscal variable. Keywords: DSGE models, small open economy, CEE counries, fiscal policy, moneary policy. JEL Classificaion: C, E, F. This research was suppored by a gran from he CERGE-EI Foundaion under a program of he Global Developmen Nework. All opinions expressed are hose of he auhor and have no been endorsed by CERGE-EI or he GDN. Many hanks o Sergey Slobodyan (CERGE-EI) for his insighful commens which very much helped in refining his paper. 3 Prof. Lucian Liviu Albu, Insiue for Economic Forecasing, Dr. Elena Pelinescu, Insiue for Economic Forecasing and Hyperion Universiy, Crisian Sănică, Naional Insiue for Saisics and Radu Lupu, Academy of Economic Sudies were also acively involved in his projec. * Researcher a he Insiue for Economic Forecasing, Romanian Academy; a: Caraiani@ipe.ro.

2 . Inroducion The accession of he CEE counries o he European Union brough in he forefron he necessiy of righ economic policies. This asks for a good evaluaion of he reacion of macroeconomic variables o he differen shocks. These economies are in a process of harmonizaion of economic policies and of ensuring he economic convergence. In his conex he analysis of he differences beween he fiscal policies from he CEE counries offers a basis for undersanding he paricular reacion mechanisms o he inernal shocks in hese economies. Moreover, he ineres in he fiscal policy poenial for sabilizaion was spurred by he ongoing crisis which showed ha simple couner-cyclical moneary policy are no enough. Unil recenly he opic of fiscal shocks was mosly addressed from a VAR perspecive, see Peroi (), Giuliodori and Beesma (), Corsei and Muller () or Mounford and Uhlig (8), mos of hem for he case of US and OECD. A more and more used framework o sudy he modeling of fiscal policy is he srucural approach, namely he dynamic sochasic general equilibrium approach (DSGE, hereafer). The firs papers focused on he real business cycles (RBC, hereafer) model where markes are compeiive and no fricions are presen. The sandard RBC model was augmened wih a fiscal side and he model calibraed usually on daa from US economy. However, he predicions of he RBC model wih respec o he impac of fiscal policy conrased wih hose implied by he sandard IS-LM model. For example, in he RBC model, like Baxer and King (993), a rise in he governmen spending leads o a reducion in final consumpion. Gali, Lopez-Salido and Valles (7) showed ha in a New Keynesian model (NK, hereafer) wih sicky prices and augmened wih rule of humb consumers, here was a posiive effec of governmen spending on consumpion. Forni, Monefore and Sessa (7) exended he conribuion of Gali, Lopez- Salido and Valles (7) o a more complex DSGE model, inspired by Smes and Wouers (3). The model was esimaed on Euro Area daa using Bayesian echniques. They showed ha, alhough fiscal shocks were persisen, he impac of governmen purchases had small and shor lived expansionary effecs on privae consumpion. They

3 also showed ha mos fiscal policy variables do no conribue oo much o he variabiliy of he main macro variables. A similar approach was ha in Sraub and Coenen () who augmened he model in Smes and Wouers (3) o he case of heerogenous consumers and sudied he impac of fiscal policy in his framework. In he open economy seing, one of he firs conribuions was ha in Furlaneo (6). He exended he model in Gali and Monacelli () in order o sudy he impac of governmen spending in small open economy NK model wih sicky prices. He found a posiive bu lower han one impac of fiscal policy on oupu. The posiive fiscal shock also leads o an increase in he nominal ineres rae and an appreciaion of he nominal exchange rae hrough he uncovered ineres pariy which lowers he overall impac of fiscal shocks on consumpion. Herz, Roeger and Vogel () discussed he issue of shor-erm sabilizaion poenial of fiscal policy in an open economy NK model wih Ricardian households. They considered wo ypes of fiscal variables, ax raes (on income and on consumpion) and public expendiure. They showed ha in heir model, he fiscal policy had boh demandside and supply-side effecs. They also showed ha sae-dependen expendiure policies ouperform he opimal ax policies for demand side shocks. Also, sae-dependen expendiure policies, according o heir model, perform similarly o moneary policy under discreion or under commimen. A he same ime, sae-dependen ax policies are efficien in sabilizing supply side disorions for moneary policy under commimen. Mos of he sudies on fiscal policy wihin he NK framework considered US or Euro Area. During he las years, some sudies were also underaken for emerging economies. One good example is ha of Garcia and Resrepo (7) who esimaed a small open economy NK model for he case of Chilean economy. They exended he sandard model by including rule of humb consumers, sicky wages and disorionary axes. They showed ha he impac of governmen spending on consumpion depends on he degree of price elasiciies of expors and impors, he share of rule of humb consumers, he share of domesic goods in governmen consumpion, he presence of disorionary axes and on he fiscal and moneary rules. A he same ime he lieraure on CEE counries is limied. Several DSGE models were recenly esimaed for some of he CEE counries, like Hradisky e al. (7) or 3

4 Musil and Vasicek (6) for Czech Republic, Kolasa (8) for Poland, Jakab and Vilagi (8) for Hungary, or Caraiani (8) for Romania. Some of hese models discussed fiscal issues, bu heir focus was no on fiscal policy. This paper exends he ongoing developmen of lieraure on fiscal policy for he case of he CEE counries. I consider an open economy wo-counry DSGE model wih a fiscal side which I esimae for he cases of Czech Republic, Hungary, Poland and Romania on quarerly daa using Bayesian echniques. This paper is organized as follows. The following secion oulines he model and discusses is building blocks. I esimae he model in he hird secion using Bayesian echniques and discuss significance of he esimaions. In he fourh secion I analyze he reacion of fiscal policy given he impac of domesic and exernal shocks on he domesic economy. The fifh secion discusses how much fiscal policy conribues o he overall performance of he model as well as is role in sabilizing he economy. In he las secion I conclude and draw some possible policy implicaion.. The Model Iniially, he applied work on DSGE models focused more on opics like business cycles, moneary policy, inernaional economics, or forecasing. Less imporance was accorded o fiscal policy. However, he ineres in he fiscal policy may increase, as in he ongoing crisis, when he moneary policy ools seem no o be enough o sop he downurn in he naional economies. I use an open economy DSGE model as in Moons e al. (7). The purpose of heir aricle was an analysis of he moneary policy in he Euro Area, using a NK model wih a Taylor moneary policy rule and a Taylor fiscal rule. The model was esimaed on quarerly daa for Euro Area, and differen opimal and ad-hoc moneary rules were discussed. The model considers simple AR () rules for he foreign economy. The model is presened below in he following equaions: = y + ψ ) E y+ α + * * d ( r E Δp r) + μg + σy + ( e + p p ) u y ψ ( δ + () Δ = Δp + + * s ( ω) E Δp + γy + Δ( p + e ) u p ω τ + ()

5 * e ( e ) e = r r u E + + (3) r r ( ρ )( χ Δp + χ y + Δe ) u r = ρ r + π χ + () g r g ( λ ) y u g g y e g = λ g χ + () y ρ + u (6) * * = yf y * * = pf p df p ρ + u (7) * * = rf r rf pf r ρ + u (8) The firs equaion is an open economy IS curve. Since he IS curve resul from he opimizing decision of household who maximize heir lifeime uiliy, a forward looking elemen appears, y +. The backward looking elemen y - is he resul of exernal habi formaion. Moreover, since i characerizes an open economy, he domesic oupu is influenced by boh he real exchange rae and by he foreign oupu y *. The fiscal balance g also influences he oupu since his model includes he governmen side. Equaion () is an open economy New Keynesian Phillips Curve, where p sands for domesic prices. This curve is derived from he profi maximizaion decision of he domesic firms under he monopolisic compeiion and sicky prices. The NK Phillips curve is forward looking, as expecaions abou fuure inflaion influence he curren inflaion. I also comprises a backward looking elemen in inflaion, so ha pas inflaion maers for curren inflaion. This exension due o Gali and Gerler (999) improves he ineria of he inflaion. As Moons e al. (7) poined ou, he elemen y can be inerpreed as demand pull inflaion. Since i is an open economy Phillips curve, he inflaion of impor prices appears oo, namely p + e *. Equaion (3) specifies he dynamics of he exchange rae e. The exchange rae follows he uncovered ineres pariy o which a shock is added, he so called risk premium shock, which akes ino accoun he measuremen errors. The moneary policy rule, equaion (), is a ypical Taylor rule. Here, he sandard Taylor formulaion is modified o allow for ineres rae smoohing, as proposed by Clarida e al. (999). The moneary rule also comprises he exchange rae elemen, which is a reasonable hypohesis in he conex of CEE counries.

6 The fiscal rule is given in equaion (), following he proposal in Taylor (). Moons e al. (7) considered also he srucural fiscal balance g which is reasonable in he ligh of he Sabiliy and Growh Pac. However since our sample daa includes years for which such a rule was no considered by hese counries, and since in his paper I focus on he cyclical aspecs of fiscal policy, I ignored his elemen. Thus he fiscal rule reduces o he defici smoohing elemen and he cyclical fiscal sance. Equaions (6) (8) specify he foreign economy, in his case, he Euro Area, which is considered as a large open economy. As in Moons e al. (7), I considered simple AR() rules for foreign producion y *, foreign prices p *, and foreign ineres rae r *. 3. Daa and Esimaion of he Model 3.. Esimaion for Romania I esimae he model given in he equaions ()-(8) using Bayesian echniques. In order o obain daa which is similar in inerpreaion o he variables in he model, I apply he logarihm, and de-rend all he variables in he model, excep he nominal ineres rae in he Euro Area. The nominal ineres rae in Romania is de-rended due o he fac ha i conains a decreasing rend. The esimaion is done for he period beween Q and 9 Q, using quarerly daa on prices, domesic GDP, ineres rae, Euro Area GDP, Euro Area ineres rae and domesic governmen spending. For prices I used he quarerly GDP deflaor. The parameers of he foreign block were calibraed using resuls from running OLS on AR() processes on foreign producion, foreign prices and foreign nominal ineres rae. Thus, ρ rf was calibraed o.9, ρ yf o.9 and ρ πf o.69. The sandard deviaions for hese hree shocks were calibraed o.3,.3 and., which is in line wih wha esimaion would resul in. I run wo Meropolis Hasings chains each of draws. The average accepance raio was for he chains of abou.67% and.7% respecively. As for he convergence saisics, he mulivariae convergence and univariae saisics proposed by Brooks and Gelman (998), Annex B, indicae ha he convergence was achieved. We 6

7 also observe, see Annex A, ha he marginal poserior disribuions indicae ha here are differences beween poserior means and prior means. The esimaes of he Taylor rule for Romanian economy indicae ha he weigh on inflaion is considerable confirming he fac ha he Naional Bank of Romania (NBR, hereafer) followed firs of all o sabilize he prices as i prepared for and hen i adoped he inflaion argeing regime during he sudied sample. Since χ π =.93, which is less han one, he esimae implies a passive moneary policy. A he same ime he weigh pu on he oupu gap is high compared o he usual findings in he lieraure, as χ y =.86. The esimaion also suggess ha NBR reaced o changes in he exchange rae, as he mean esimae for χ e is.8. The esimaion for he IS curse indicaes ha half of he agens are backward looking; however he esimae for ψ is equal o he prior mean.. The coefficien relaed o he influence of he fiscal variable has a considerable value, wih a mean esimae for μ of.38. The parameer σ relaed o he impac of foreign demand on domesic oupu shows a value significanly differen from zero, bu close o he prior mean. The coefficien δ relaed o he open economy elemens is raher low, wih a mean esimae a.. For he case of he Phillips curve we noice ha mos of he firms are forward looking, ω=., which is similar o he case of Euro Area, see Moons e al. (7). The coefficien associaed o he oupu gap is significanly differen from zero, γ=.9, and slighly higher han ha for he case of he Euro Area. As for he fiscal rule, we can noice ha here is a significan esimaed value for he parameer relaed o he cyclical fiscal sance as χ g =.9, while fiscal ineria is raher low, λ g = Esimaion for Czech Republic The daa for he esimaion for he model for he case of Czech Republic is similar in definiion and inerpreaion as in he previous case. The daa for Euro Area block is idenical wih ha for Romanian case. 7

8 For daa regarding he domesic block I choose quarerly daa saring wih 996 quarer. The bigger lengh of he sample was possible due o he daa availabiliy from Eurosa. I esimaed he model in equaion () o (8) using Bayesian echniques, as moivaed above, afer calibraing he values for he parameers corresponding o he foreign block in a similar was as in he previous secion. As in he case for Romania, he esimaion was based on wo Meropolis Hasings chains, each one wih draws. The average accepance raio was of abou.%, and.% respecively for he wo blocks. Annex B shows he mulivariae and univariae Brooks-Gelman saisics which indicae ha convergence was achieved. The esimaion of he IS curve shows ha he backward looking elemen is more imporan for he agens, suggesing an imporan degree of ineria, as he poserior mean for ψ was esimaed a.63. There is sronger influence of governmen expendiures as he poserior esimae for μ is.6. For he case of he Phillips curve, he esimaion shows ha mos of he firms are forward looking, ω=.. The coefficien characerizing he forward-lookingness is sronger han for he case of Romania. The mean esimae for he oupu-gap coefficien is significanly differen from zero, bu, a he same ime, close o he prior mean, γ=.6. The Taylor rule esimaion resuled in raher expeced values he case of inflaion coefficien. The inflaion argeing regime is suggesed o have been followed again, while moneary policy can be characerized as mildly acive, χ pi esimaed a.9. The coefficien relaed o he oupu gap is high bu much smaller han he one for Romania, χ y esimaed a.67. The Czech Naional Bank also reaced o changes in he exchange rae bu in a weaker manner han Romanian auhoriies, χ e =.9. The esimaion of he fiscal rule is close o ha for Romania, wih a similar coefficien relaed o he cyclical fiscal sance, a poserior mean of χ g =.9 and a higher coefficien relaed o fiscal ineria λ g =.3. 8

9 3.3. Esimaion for Poland The daa for he esimaion for he model for he case of Poland is similar in definiion and inerpreaion as in he previous wo cases. The daa for Euro Area block is idenical wih ha for he previous cases. For daa regarding he domesic block I choose quarerly daa saring wih 999 quarer since his is he sample ha provided he bes Bayesian esimaion. I esimaed he model in equaion () o (8) using Bayesian echniques, as moivaed above, afer calibraing he foreign economy block. As in he oher cases, he esimaion was based on wo Meropolis Hasings chains, each one wih draws. The average accepance raio was of abou 3.6%, and 3.7% respecively for he wo blocks. Annex B shows he mulivariae as well as he univariae Brooks-Gelman saisics which indicae ha convergence was achieved. The esimaion of he IS curve shows ha backward looking elemen is slighly more imporan han he forward looking elemen, ψ=., however he value is very close o he prior mean. Again, he coefficien relaed o he fiscal side is considerable, wih a poserior mean of.6 for γ. For he case of he Phillips curve, he esimaion shows ha mos of he firms are forward looking as for he previous cases, as he poserior mean for ω is.. The coefficien relaed o he oupu gap is moderae, as γ is esimaed a.9. The Taylor rule esimaion gives a value for he inflaion coefficien ha is close o he one for he Czech Republic, χ pi esimaed a.6. The Naional Bank is shown o have followed he inflaion argeing regime, as suggesed by he parameer relaed o inflaion. A he same ime, he esimaion suggess a raher weakly acive moneary policy. I also reaced o changes in he exchange rae, as he mean esimae for χ e is.3. The oupu gap coefficien is very srong; he poserior mean of χ y was esimaed a.98. As for esimaion of he fiscal rule, he values are similar wih respec o ineria and cyclical fiscal sance wih some sligh differences relaive o he oher wo counries, wih he poserior mean for χ g =.6 and he esimae for λ g =.3. 9

10 3.. Esimaion for Hungary The daa for he esimaion for he model for he case of Hungary is similar in definiion and inerpreaion as in he previous hree cases. The daa for Euro Area block is idenical wih ha for he previous cases. For daa regarding he domesic block I choose quarerly daa saring wih 99 quarer. In his case, he bes approach was o use he chain index quarerly GDP deflaor wih he base in for inflaion. The quarerly ineres raes daa series was again aken from Cenral Bank. I esimaed he model in equaion () o (8) using Bayesian echniques, as moivaed above, afer calibraing he foreign economy block. As in he oher cases, he esimaion was based on wo Meropolis Hasings chains, each one wih draws. The average accepance raio was of abou.7%, and.% respecively for he wo blocks. Annex B shows he mulivariae as well as he univariae Brooks-Gelman saisics which indicae ha convergence was achieved. The esimaion of he IS curve shows ha backward looking elemen is as imporan as he forward looking elemen, basically, in line wih he findings for he oher counries. Here he poserior mean for ψ was esimaed a.9, which suggess a weak idenificaion. There is also a significan coefficien for governmen expendiures, wih a mean esimae for μ of.39. For he case of he Phillips curve, he esimaion shows ha mos of he firms are forward looking as for he previous cases, ω=.9. The coefficien relaed o he oupu gap is also considerable, as he poserior mean for γ was esimaed a.6, which is however close o he prior mean. The Taylor rule esimaion gives a raher low inflaion coefficien, χ pi esimaed.7 which implies a passive moneary policy. The reacion o he oupu gap changes seems o be sronger in his case, as he poserior mean for χ y was esimaed a.88. As for esimaion of he fiscal rule, he values are similar wih respec o ineria and cyclical fiscal sance wih some sligh differences relaive o he oher hree counries, wih he poserior mean esimaes for χ g a.9 and for λ g a.. Thus, he esimaes for he four economies confirm he findings in Saehr (8) who showed ha, for he case

11 of he New Member saes, here was less ineria in he fiscal rule and he fiscal policy was more couner-cyclical han for he older member saes.. The Analysis of Fiscal Policy Using he Impulse Response Funcions In his secion I analyze he impulse response funcions of he endogenous variables o a se of seleced domesic and exernal shocks. I focus on hose ypes of shocks which I considered o be providing evidence for he fiscal policy in Czech Republic, Hungary, Poland and Romania. The se of shocks o focus on are: he domesic demand shocks, he domesic ineres rae shocks, he fiscal policy shocks, he domesic moneary policy shocks, he exernal demand shocks and, respecively, he foreign ineres rae shock. I simulae he model wih he parameers se o he mean values of he poserior disribuions. Shocks are unanicipaed and appear in period one. For all he cases, hey represen % emporary shocks and hey are no auo correlaed. In each cases he y-axis in given in percenage poins. A he same ime, given he curren conex of a global crisis, and since he economiss are ineresed in he way he fiscal policy reacs o such challenges, I considered specific shocks, wih demand shocks considered as negaive, he ineres rae shocks appearing as negaive due o he use of moneary policy as a ool o simulae he economy, while he fiscal policy shocks are posiive, reflecing he use of fiscal policy o counerac he negaive effec of he economic crisis... The Domesic Demand Shocks A negaive demand shock leads o a drop in domesic oupu in all cases, see Annex C. The moneary auhoriy responds by lowering he ineres rae. The governmen also uses an expansionary fiscal policy o counerac his drop in demand. In all counries we see ha he oupu reurns o he seady sae afer -3 quarers. The reacion is almos similar in hese counries. However here is a sronger response of

12 oupu and prices in Romania. A he same ime, inflaion reacs more persisenly in Czech Republic and Hungary... The Domesic Ineres Rae Shocks In he presen conex of he financial crisis, he usual reacion of he cenral banks is o lower he ineres rae. We discuss here he impac of a negaive shock in he ineres rae, corresponding o laxer moneary policy (see Annex D). The impac of he lower ineres leads o a rise in he oupu, wih he sronges response for he case of Romania. The prices rise in he four counries, wih he sronges magniude for Romania, while he larger persisence is produced for he oher counries..3. The Fiscal Policy Shocks Given he inheren limis in he use of moneary policy o sabilize he economies during a financial crisis, a second ool available o he auhoriies is he fiscal policy. The prices and exchange raes rise in hese counries wih a realisic hump-shaped reacion, see Annex E. The maximum poin is reached in all counries afer four five quarers. The ineres rae reacs posiively o couner he rise in prices. The sronges reacion of prices is again in Romania, while for he oher hree counries he response is moderae bu more persisen. The sronges impac on oupu appears in he case of Romania, bu is persisence is very low (a few quarers)... The Foreign Moneary Policy Shocks I simulae here he impac of an unanicipaed % emporary negaive shock in he Euro Area ineres rae, see Annex F. Since in his model he Euro Area economy is he large economy, while Romanian, Czech and Polish economies are small open economies, shocks ha are produced in he Euro Area economy influence he wo economies, while he reverse is no rue.

13 The lower ineres rae in Euro Area leads o an appreciaion in he domesic exchange rae. The oupu in all four counries reacs iniially posiively, wih he maximum reached a he iniial momen, and aferwards i diminishes gradually, fading away afer - quarers. The sronges reacion is in he case of Poland for boh prices and oupu... The Foreign Demand Shock In he ongoing crisis, he CEE counries can also be affeced hrough he lower demand in he Euro Area. I simulae here he impac of an unexpeced negaive shock in he Euro Area demand, see Annex G. The negaive shock on foreign demand leads o a negaive impac on domesic demand. The impac is slighly sronger for Poland, and i is also considerable for Czech Republic and Romania. For he case of prices, we can again a hump-shaped reacion, wih a sronger and more persisen impac for he cases of Czech Republic, Poland and Romania. The fall of prices and he oupu leads o a boh relaxaion in moneary policy and an expansionary fiscal policy.. The Imporance of Fiscal Policy In his chaper we address he imporance of fiscal policy for he sudied counries. We sudy several aspecs. Firs of all we quesion wheher he inclusion of fiscal policy rule in he DSGE model improves he fi of he model. Second, we look a how much fiscal defici maers for he variaion of he main macroeconomic variables. We also look a he hisorical decomposiion of oupu for he sample considered in he esimaion. Finally, we run several counerfacuals scenarios in order o simulae he impac of differen assumpion regarding he fiscal policy on he oupu dynamics. 3

14 .. The Fi of he Model Annex H presens he resuls of he esimaed DSGE model wih and wihou he Taylor fiscal rule. For he case of he model wihou he fiscal rule, equaion () is reduced o a simple AR process as saed below: g g + u = λ g g The modified version of he model was esimaed for he four counries using he same daa, priors as well as esimaion procedure. The esimaion led o good resuls in erms of mulivariae and univariae convergence, poserior disribuions or average accepance raio. The models are compared using he bayesian facor expressed in logs, following Jeffreys (96). We find he log-bayes facors of around for Romania, Hungary and Poland, implying ha we would need a prior probabiliy of M, he model wih fiscal rule as a simple AR() process, o be 8 (=e ) imes larger han he prior probabiliy of M, he model wih Taylor-like fiscal rule, in order o prefer M based on poserior odds. This can be inerpreed as evidence in he favor of he model feauring a Taylor fiscal rule. For he case of Czech Republic, he evidence in favor of a fiscal Taylor rule is much sronger, implying a prior probabiliy of M being 8 (=e 7. ) imes larger han he prior probabiliy of M in order o prefer M. We can conclude ha he DSGE model feauring a fiscal rule is favored for all counries considered agains he simple AR processes... Explaining he Variaion of Macroeconomic Variables We discuss here how much he differen shocks (including he fiscal shocks) explain from he variaion of main macroeconomic variable, namely, producion, inflaion, or he ineres rae. The Annex I shows he variance decomposiion for he main macroeconomic variables in he hree counries. The variance decomposiion compued here is an asympoic one. The resuls show ha he fiscal shock has an imporan role in explaining he variaion of oupu for he hree economies: i explains abou 6% of oupu variaion for

15 he cases of Romania, Hungary and Poland, and i has a slighly smaller explanaive power for Czech Republic, wih oupu variaion explained in a % proporion by governmen spending. I also has a significan conribuion o he explanaion of he variance of inflaion (for he cases of Romania and Poland) and ineres rae (for he cases of Romania, Czech Republic and Poland). The resuls for inflaion are no so homogenous. Thus, he variaion in domesic inflaion is explained mosly by is own shocks in Romania, Hungary and Czech Republic. However, for he case of Poland, mos of inflaion variaion is explained by foreign demand and supply shocks. Also, for Czech Republic, a considerable par of variaion in inflaion is explained by foreign demand and supply shocks..3. Hisorical Decomposiion of Variables In Annnex J, I presen a hisorical decomposiion for he main domesic variable of ineres, namely for he oupu, wih conribuions of each shock along he sudied sample. Quie ineresingly, he fiscal shocks appears o have had a posiive effec during he years of economic growh, especially for he las wo years of high growh (7 and 8), and i did no help a all hese economies during he curren economic crisis. A he same ime, for pas periods for some counries, like Hungary beween and 3 or Czech Republic beween and, here are periods when fiscal shocks behaved in a couner-cyclical way. The recen lieraure argues ha he fiscal policy is pro-cyclical in he emerging economies see for example Ilzezki and Vegh (8) for one of he laes researches on his opic. However, he findings here presen mixed evidences. While based on he esimaion one could argue ha fiscal policy is couner-cyclical, he fiscal shocks, a leas for he las years, had a pro-cyclical behavior. Some more research should be underaken on his opic.

16 .. Some counerfacuals I compare here he dynamics of he model-economy under differen scenarios regarding he pah of governmen spending as well as he parameers characerizing he fiscal rule, see Annex K. I consider he pah of GDP beween 7Q and 9Q (basically, wo years before he crisis, and he year of he crisis in hese counries) under wo alernaive scenarios: a) he pah of GDP if here is no Taylor fiscal rule, namely χ g is equal o.. In his case he fiscal rule collapses o an AR() process; b) he pah of GDP if he governmen would have kep a consan posiive considerable governmen spending effor. This is assumed o be a 3% posiive gap for governmen spending and i is based on he average figures for counries like Romania and Poland for which he average governmen spending gap during 8 was beween % and 3%. The resuls for he four counries are prey consisen. For he cases of he Czech Republic, Hungary and Romania, he pah of GDP, during he crisis year 9, under no fiscal rule is worse han ha of acual GDP, wih he bigges impac for Romania by abou.%. Also, a consan posiive budgeary effor would have increased he GDP, wih he bigges impac for he case of Hungary by 3.%. For he only counry where he GDP did no decrease alhough here was a slowdown in he economy (implying an acual negaive oupu gap), namely Poland, he no-fiscal-rule case would have also led o a negaive impac on GDP during he crisis year 9. However, a consan governmen spending effor would have no improved he pah of GDP during 9, which may also be explained hrough he fac ha he counercyclical scenarios used daa calibraed on Poland case. 6. Conclusion The curren economic crises was no only unprediced bu i also surprised hrough he scale and complexiy of he issues i raised. In he face of a recession ha 6

17 hreaens o become a depression, policy makers are forced o find quick soluions o complex problems. The CEE economies were, unil recenly, he fases growing economies in Europe and had brigh prospecs. However, he severiy of he ongoing crisis pu in evidence he weaknesses in hese economies and how fragile was heir sabiliy. In his paper I explored he effecs of fiscal policy in CEE counries in he conex of he ongoing financial crisis by using evidence from an esimaed DSGE model for Czech Republic, Poland and Romania. The esimaion and he simulaions showed ha he fiscal policy can counerac no only he negaive domesic shocks, bu also adverse shocks from Euro Area (like negaive demand shocks). Running counerfacual scenarios confirms ha a considerable governmen spending effor would have improved he dynamics of GDP during he crisis, excep for he case of Poland which coninued o enjoy, a a smaller scale, economic growh. Using he Bayesian comparison approach, I also found evidence in he favor of including he fiscal Taylor rule agains he alernaive of simple AR() processes for he fiscal variable. References [] Baxer, M. and R.G. King "Fiscal Policy in General Equilibrium," American Economic Review 83(3): [] Brooks, S.P., and A. Gelman General Mehods for Monioring Convergence of Ieraive Simulaions, Journal of Compuaional and Graphical Saisics 7(): 3. [3] Buncic, D. and M. Melecky, 8. "An Esimaed New Keynesian Policy Model for Ausralia," The Economic Record 8 (6): -6. [] Caraiani, P. 8. An Analysis of Domesic and Exernal Shocks on Romanian Economy Using a DSGE Model, Romanian Journal of Economic Forecasing, 9 (3): -. [] Clarida, R., J. Gali and M. Gerler The Science of Moneary Policy: A New Keynesian Perspecive, Journal of Economic Lieraure 7(): [6] Corsei, G. and G. Muller.. Twin Deficis: Squaring Theory, Evidence and Common Sense, Economic Policy (8):

18 [7] Forni, L., L. Monefore, and L. Sessa. 7. The general equilibrium effecs of fiscal policy: Esimaes for he euro area, Banca d Ialia Discussion Paper 6. [8] Furlaneo, F. 6. Fiscal Shocks in a Small Open Economy, Universiy of Lausanne, DEEP Working Paper. [9] Gali, J. and M. Gerler Inflaion Dynamics: A Srucural Economeric Approach, Journal of Moneary Economics (): 9-. [] Gali, J. and T. Monacelli.. Opimal Fiscal Policy in a Moneary Union, working paper. [] Galí, J., J.D. López-Salido and J. Vallés. 7. Undersanding he effecs of governmen spending on consumpion, Journal of he European Economic Associaion (): 7-7. [] García, C.J. and J.E. Resrepo. 7. How Effecive is Governmen Spending in a Small Open Economy wih Disorionary Taxes manuscrip. [3] Giuliodori, M. and R. Beesma.. Wha Are he Spillovers from Fiscal Shocks in Europe? An Empirical Analysis, De Economis 3(): [] Herz, B., W. Roeger and L. Vogel.. Opimal Simple Rules for Fiscal Policy in a Moneary Union BGPE Discussion Paper No.. [] Hradisky, M., R. Girardi and M. Rao. 7. Think Twice, he EMU has is Prices: The Czech Republic Case, [Working Paper.] Ispra (Ialy): Euro-area Economy Modeling Cenre, Join Research Cenre, European Commission. [6] Jakab, Z.M. and B. Vilagi. 8. An Esimaed DSGE Model of he Hungarian Economy, MNB Working Paper 8/9. [7] Jeffreys, H. 96. Theory of Probabiliy. Oxford Universiy Press. [8] Gali, J. and T. Monacelli,. "Opimal fiscal policy in a moneary union," Federal Reserve Bank of San Francisco Proceedings. [9] Ilzezki, E. and C.A. Vegh. 8. Procyclical Fiscal Policy in Developing Counries: Truh or Ficions? NBER Working Paper 9. [] Kolasa, M. 8. Srucural Heerogeneiy or Asymmeric Shocks? Poland and he Euro Area hrough he Lens of a Two-Counry DSGE Model, Naional Bank of Poland Working Paper No. 9. 8

19 [] Monacelli, T. 3. Moneary Policy in a Low Pass-Through Environmen, European Cenral Bank Working Paper 7. [] Moons, C., H. Garresen, B. van Aarle and J. Fornero. 7. "Moneary policy in he New-Keynesian model: An applicaion o he Euro Area," Journal of Policy Modeling 9(6): [3] Mounford, A. and H. Uhlig. 8. Wha are he Effecs of Fiscal Policy Shocks? NBER Working Papers no.. [] Peroi, R.. Esimaing he Effecs of Fiscal Policy in OECD Counries, CEPR Discussion paper 8. [] Rabanal, P. and J.F. Rubio-Ramirez. 3. Comparing New Keynesians Models in he Euro Area: A Bayesian Approach, Federal Reserve Bank of Alana Working Paper 3-3. [6] Smes, F. and R. Wouers. 3. An Esimaed Sochasic Dynamic General Equilibrium Model for he Euro Area, Journal of he European Economic Associaion (): 3-7. [7] Saehr, K. 8. Fiscal Policies and Business Cycles in an Enlarged Euro Area, Economic Sysems 3 (): [8] Sraub, R. and G. Coenen, G.. Non-Ricardian Households and Fiscal Policy in an Esimaed DSGE Model of he Euro Area, Compuing in Economics and Finance no.. [9] Taylor, J.. Reassessing discreionary fiscal Policy, Journal of Economic Perspecives (3): -36. [3] Vasicek, O. and K. Musil. 6. Behavior of he Czech Economy. New Open Macroeconomics DSGE Model, Prague: Research Cener for Compeiiveness of Czech Economy Working Paper, No. 3. 9

20 ANNEX A. Resuls of he Bayesian Esimaion Annex A.. Tables Bayesian Esimaion Resuls for Romania Table A. PARAMETERS PRIOR MEAN POSTERIOR MEAN CONFIDENCE INTERVAL CONFIDENCE INTERVAL PRIOR DISTRIBUTION STANDARD DEVIATION ψ Normal. α....7 Normal. μ Normal. σ..6.. Normal. δ....9 Normal. ω....3 Bea. τ Bea. γ Bea. λg Normal. χg Normal. χpi Normal. χy Normal. χe..8.. Normal. σd Inv. Gamma Infinie σs Inv. Gamma Infinie σg Inv. Gamma Infinie σe Inv. Gamma Infinie σr Inv. Gamma Infinie Source: Own Compuaion

21 Bayesian Esimaion Resuls for Czech Republic Table A. PARAMETERS PRIOR MEAN POSTERIOR MEAN CONFIDENCE INTERVAL CONFIDENCE INTERVAL PRIOR DISTRIBUTION STANDARD DEVIATION ψ Normal. α Normal. μ Normal. σ Normal. δ....3 Normal. ω Bea. τ.... Bea. γ Bea. λg Normal. χg Normal. χpi Normal. χy Normal. χe Normal. σd Inv. Gamma Infinie σs Inv. Gamma Infinie σg Inv. Gamma Infinie σe Inv. Gamma Infinie σr Inv. Gamma Infinie

22 Bayesian Esimaion Resuls for Poland Table A.3 PARAMETERS PRIOR MEAN POSTERIOR MEAN CONFIDENCE INTERVAL CONFIDENCE INTERVAL PRIOR DISTRIBUTION STANDARD DEVIATION ψ Normal. α Normal. μ Normal. σ Normal. δ.... Normal. ω....8 Bea. τ....3 Bea. γ Bea. λg Normal. χg Normal. χpi Normal. χy Normal. χe Normal. σd Inv. Gamma Infinie σs..8.. Inv. Gamma Infinie σg....8 Inv. Gamma Infinie σe Inv. Gamma Infinie σr Inv. Gamma Infinie Source: Own Compuaion

23 Bayesian Esimaion Resuls for Hungary Table A. PARAMETERS PRIOR MEAN POSTERIOR MEAN CONFIDENCE INTERVAL CONFIDENCE INTERVAL PRIOR DISTRIBUTION STANDARD DEVIATION ψ Normal. α Normal. μ Normal. σ Normal. δ.... Normal. ω Bea. τ.... Bea. γ Bea. λg Normal. χg Normal. χpi..7.. Normal. χy Normal. χe Normal. σd..8.. Inv. Gamma Infinie σs Inv. Gamma Infinie σg Inv. Gamma Infinie σe Inv. Gamma Infinie σr Inv. Gamma Infinie Source: Own Compuaion 3

24 Annex A.. Prior and poserior disribuions Esimaion for Romania SE_u_d SE_u_s SE_u_g.. SE_u_e SE_u_i psi...6 alfa.. miu. sig dela omega au.. gam lam_g...3 xi_g...6 xi_p xi_y -... xi_e

25 Esimaion for Czech Republic SE_u_d SE_u_s SE_u_g SE_u_e SE_u_i psi.. alfa miu. sig dela omega au.. gam...6 lam_g...3 xi_g.. xi_p xi_y.. xi_e....

26 Esimaion for Poland SE_u_d SE_u_s SE_u_g SE_u_e SE_u_i psi.. alfa miu. sig dela omega au.. gam...6 lam_g...3 xi_g.. xi_p xi_y.. xi_e.. 6

27 Esimaion for Hungary SE_u_d SE_u_s SE_u_g SE_u_e SE_u_i psi.. alfa miu. sig

28 dela omega au.. gam...6 lam_g...3 xi_g.. xi_p xi_y.. xi_e...3 8

29 Annex B. Brooks Gelman saisics for convergence Annex B.. Mulivariae saisics for convergence Esimaion for Romania Figure B.. Inerval x m x m x Esimaion for Czech Republic Figure B.. 9 Inerval x m x m x 9

30 Esimaion for Poland Figure B.3. Inerval x m x m x Esimaion for Hungary Figure B Inerval x m x m x 3

31 Annex B.. Univariae saisics for convergence Annex B... Resuls for Romania.3 SE_u_d (Inerval). x - SE_u_d (m) x -6 SE_u_d (m3)... x SE_u_s (Inerval). x x - SE_u_s (m) x x -7 SE_u_s (m3) x SE_u_g (Inerval) x. x - SE_u_g (m) x x -6 SE_u_g (m3).. x. x x. SE_u_e (Inerval). SE_u_e (m) x -3 SE_u_e (m3)...6 x SE_u_i (Inerval) x x - SE_u_i (m) x x - SE_u_i (m3)....3 x psi (Inerval). x psi (m) x 3 x -3 psi (m3).3. x. x x 3

32 . alfa (Inerval). alfa (m) x -3 alfa (m3) x x miu (Inerval) x -3 miu (m) x x - miu (m3)..3 x sig (Inerval). x sig (m) x x -3 sig (m3).... x x x.8 dela (Inerval) x -3 dela (m) x - dela (m3).6.. x omega (Inerval) x x -3 omega (m) x x -3 omega (m3).... x au (Inerval) 3 x x - au (m) x. x - au (m3). 3.3 x x. x 3

33 . gam (Inerval) x -3 gam (m) x - gam (m3). 3.. x lam_g (Inerval). x lam_g (m) x x -3 lam_g (m3).3... x xi_g (Inerval).3 x xi_g (m). x xi_g (m3)... x. x x.7 xi_p (Inerval).8 xi_p (m). xi_p (m3) x xi_y (Inerval) x xi_e (Inerval) x x xi_y (m) xi_y (m3)... x x xi_e (m) x -3 xi_e (m3).. x x x 33

34 Annex B... Resuls for Czech Republic 8 x SE_u_d -3 (Inerval) 6 x 6 x SE_u_s -3 (Inerval) x 6 x SE_u_g -3 (Inerval). x - SE_u_d (m) x x -6 SE_u_s (m) 3 x x - SE_u_g (m). x -8 SE_u_d (m3) x x -8 SE_u_s (m3) x x -8 SE_u_g (m3) x x x.8 SE_u_e (Inerval) x -3 SE_u_e (m) x - SE_u_e (m3).6. x x SE_u_i -3 (Inerval) 9 8 x psi (Inerval).3.3 x.6 x - SE_u_i (m).. x psi (m).. x x SE_u_i -8 (m3) x x -3 psi (m3). x. x x 3

35 . alfa (Inerval). alfa (m). x -3 alfa (m3) x x miu (Inerval) x -3 miu (m) 3. x x - miu (m3).. x sig (Inerval). x sig (m) x x -3 sig (m3).. x x x. dela (Inerval) 3 x - dela (m) 3 x -7 dela (m3).... x omega (Inerval) x 3 x -3 omega (m) x x - omega (m3)..6 x au (Inerval) x 3. x - au (m) x 3 x -7 au (m3). 3.. x. x x 3

36 . gam (Inerval) 3 x -3 gam (m) x - gam (m3).. x lam_g (Inerval) x 8 x -3 lam_g (m) x. x -3 lam_g (m3). 6. x xi_g (Inerval). x xi_g (m). x x -3 xi_g (m3).3.. x. x x.6 xi_p (Inerval). xi_p (m). xi_p (m3)..3.. x xi_y (Inerval)... x xi_y (m). x xi_y (m3)..3.. x xi_e (Inerval). x 3 x -3 xi_e (m) x x - xi_e (m3). x x x 36

37 Annex B..3. Resuls for Poland x SE_u_d -3 (Inerval) 8. x - SE_u_d (m). x -7 SE_u_d (m3) 6. x SE_u_s (Inerval). x 8 x -6 SE_u_s (m) x x -8 SE_u_s (m3). 6 x 8 x SE_u_g -3 (Inerval) 7 x 8 x -6 SE_u_g (m) 7 x x -8 SE_u_g (m3) 3 6 x 6 x x.8 SE_u_e (Inerval) x -3 SE_u_e (m) x - SE_u_e (m3).6... x SE_u_i (Inerval) x x - SE_u_i (m) x. x -7 SE_u_i (m3) x psi (Inerval). x psi (m). x 3 x -3 psi (m3) x x x 37

38 . alfa (Inerval) x -3 alfa (m). x -3 alfa (m3).. x miu (Inerval) 8 6 x 3 x -3 miu (m). x 3 x - miu (m3).3... x sig (Inerval). x sig (m) x x -3 sig (m3)... x x x.3 dela (Inerval) 6 x - dela (m) 8 x -7 dela (m3). 6.. x omega (Inerval) x 3 x -3 omega (m) x x - omega (m3).. x au (Inerval) x 6 x - au (m) x 6 x -7 au (m3).. x x x 38

39 . gam (Inerval) 6 x -3 gam (m) x -3 gam (m3).... x lam_g (Inerval) x 8 x -3 lam_g (m) x. x -3 lam_g (m3). 6. x xi_g (Inerval).3 x xi_g (m). x x -3 xi_g (m3).3. x. x 3 x.6 xi_p (Inerval).6 xi_p (m). xi_p (m3) x. x. x xi_y (Inerval) xi_y (m).8 8 x -3 xi_y (m3) x x x xi_e (Inerval) x -3 xi_e (m). x - xi_e (m3)... x x. x 39

40 Annex B... Resuls for Hungary. SE_u_d (Inerval) x - SE_u_d (m) x -8 SE_u_d (m3).. x 8 x SE_u_s -3 (Inerval) 6 x x SE_u_g -3 (Inerval) 8 x x - SE_u_s (m). x. x - SE_u_g (m) x x -8 SE_u_s (m3) x x -7 SE_u_g (m3). 6 x. x x. SE_u_e (Inerval) x -3 SE_u_e (m) x - SE_u_e (m3).. x SE_u_i (Inerval) x x - SE_u_i (m) x. x -7 SE_u_i (m3)...8. x psi (Inerval). x psi (m). x 3 x -3 psi (m3).3... x x x

41 . alfa (Inerval) x -3 alfa (m). x -3 alfa (m3).. x miu (Inerval) 8 6 x 3 x -3 miu (m). x x - miu (m3).... x sig (Inerval). x sig (m) x x -3 sig (m3)... x x x.3 dela (Inerval) 8 x - dela (m) x -6 dela (m3). 6.. x omega (Inerval) x. x -3 omega (m) x x - omega (m3)...3 x au (Inerval). x x - au (m) x x -6 au (m3).... x x x

42 . gam (Inerval) 3 x -3 gam (m) x - gam (m3).. x lam_g (Inerval) x 8 x -3 lam_g (m) x 8 x - lam_g (m3) x xi_g (Inerval).3 x xi_g (m) x 6 x -3 xi_g (m3).3. x. x x xi_p (Inerval) xi_p (m) xi_p (m3)...3. x. x. x xi_y (Inerval) xi_y (m).3 8 x -3 xi_y (m3). 6. x x x xi_e (Inerval) x -3 xi_e (m) x - xi_e (m3)... x x x

43 Annex C. The Impac of Domesic Demand Shocks Esimaion for Romania Figure C.. p y - - r - e - g - - Esimaion for Czech Republic Figure C.. p y r - e -. g

44 Esimaion for Poland Figure C.3. p y r e - x -3 g - - Esimaion for Hungary Figure C.. p y r - e -. g - -.

45 Annex D. The Impac of Domesic Ineres Rae Shocks Case of Romania Figure D.. p y. r - e - g - - Case of Czech Republic Figure D... p y. r - e -. g - -.

46 Esimaion for Poland Figure D.3.. p y r - e - x -3 g - - Esimaion for Hungary Figure D... p y. r - e - g - - 6

47 Annex E. The Impac of Domesic Fiscal Shocks Case of Romania Figure E.. p y r - e - g - Case of Czech Republic Figure E.. p y. r - e - g - 7

48 Esimaion for Poland Figure E.3. p y. r - e - g - Esimaion for Hungary Figure E... p y r - e - g - 8

49 Annex F. The Impac of Foreign Ineres Rae Shocks Case of Romania Figure F... p. y r -. e g - -. Case of Czech Republic Figure F.. p. y. r -. e g

50 Esimaion for Poland Figure F.3. p. y.. r -. e -. x - g - rf Esimaion for Hungary Figure F... p. y -. r -. e g - rf

51 Annex G. The Impac of Foreign Demand Shocks Case of Romania Figure G.. p. y r -. e -.. g - -. Case of Czech Republic Figure G.. p. y r -. e -.. g - -.

52 Esimaion for Poland Figure G.3. p. y r e -. x - g - yf Esimaion for Hungary Figure G.. p. y r -.. e -.. g -. yf

53 ANNEX H. Bayesian Comparison of he Models wih and wihou Fiscal Rule Table H.. Bayesian comparison of he models Log-likelihood M: DSGE wih M: DSGE wihou Log-Bayes Facor of Fiscal rule Fiscal rule M agains M Romania Czech Republic Poland Hungary ANNEX I. Explaining he Variaion of he Main Economic Variables The Decomposiion of variance for he case of Romania Domesic Source of Shocks Variables Domesic Demand Domesic Supply Fiscal Defici Domesic Ineres rae Exchange Rae Foreign Demand Table I.. Foreign Ineres Rae Foreign Supply Inflaion Oupu Ineres Rae

54 The Decomposiion of variance for he case of Czech Republic Domesic Source of Shocks Variables Domesic Demand Domesic Supply Fiscal Defici Domesic Ineres rae Exchange Rae Foreign Demand Table I.. Foreign Ineres Rae Foreign Supply Inflaion Oupu Ineres Rae The Decomposiion of variance for he case of Poland Domesic Source of Shocks Variables Domesic Demand Domesic Supply Fiscal Defici Domesic Ineres rae Exchange Rae Foreign Demand Table I.3. Foreign Ineres Rae Foreign Supply Inflaion Oupu Ineres Rae The Decomposiion of variance for he case of Hungary Domesic Source of Shocks Variables Domesic Demand Domesic Supply Fiscal Defici Domesic Ineres rae Exchange Rae Foreign Demand Table I.. Foreign Ineres Rae Foreign Supply Inflaion Oupu Ineres Rae

55 ANNEX J. Hisorical Decomposiion of Oupu Hisorical decomposiion of oupu for Romania Figure J... Iniial values. epf. erf eyf u_e u_i u_g -. u_s u_d Figure J.. Hisorical decomposiion of oupu for Czech Republic.6 Iniial values. epf. erf eyf u_e -. u_i u_g -. u_s u_d

56 Hisorical decomposiion of oupu for Poland Figure J Iniial values epf erf eyf u_e -. u_i -. u_g -.6 u_s u_d Hisorical decomposiion of oupu for Hungary Figure J Iniial values epf. erf eyf u_e u_i u_g u_s u_d 6

57 K. Counerfacuals scenarios Case of Romania. Figure K Q 7Q3 8Q 8Q3 9Q 9Q3 Y_CONSTANT_G Y_ACTUAL Y_NO_RULE Case of Czech Republic Figure K Q 7Q3 8Q 8Q3 9Q 9Q3 Y_CONSTANT_G Y_ACTUAL Y_NO_RULE 7

58 Case of Poland. Figure K Q 7Q3 8Q 8Q3 9Q 9Q3 Y_CONSTANT_G Y_ACTUAL Y_NO_RULE Case of Hungary Figure K Q 7Q3 8Q 8Q3 9Q 9Q3 Y_CONSTANT_G Y_ACTUAL Y_NO_RULE 8