Causality between Spending and Revenue in Case of Greece through Toda and Yamamoto Methodology

Transcription

1 Journal of Business & Economic Policy Vol. 5, No. 1, March 218 Causaliy beween Spending and Revenue in Case of Greece hrough Toda and Yamamoo Mehodology Chaido Drisaki Associae Professor Deparmen of Accouning and Finance Wesern Macedonia Universiy of Applied Sciences Greece Absrac This paper examines he relaionship beween governmen spending and revenues in Greece for he period The Augmened Dickey-Fuller (ADF) and Phillips-Perron ess are employed o examine he order of inegraion of he variables used in he model. For he long run analysis, ARDL bounds esing approach is used as i was formed by he papers of Pesaran e al. (21). Causaliy es using he Toda and Yamamoo (1995) Granger non-causaliy procedure was employed in order o examine Granger causaliies beween variables. The resuls of co-inegraion of ARDL es showed ha here is a co-inegraed relaionship beween governmen spending and revenues. Also, causaliy es showed ha here is a unidirecional causal relaionship beween spending and revenues in Greece wih direcion from governmen revenues owards spending. Keywords: Governmen Spending, Governmen Revenue, ARDL Bounds Tesing, Toda-Yamamoo Causaliy Tes. JEL Classificaions: C5, E23, J24. 1.Inroducion The relaionship beween governmen spending and revenues is one of he ordinary problems on public economics. One of he main discussions on public economics is he pursui of he relaionship beween governmen revenues and spending. For his issue, imporan heoreical and empirical research has been conduced so ha governmens can maping ou heir economic policies and forecasing heir budge s defici. During he las decades, many papers have examined he relaionship beween governmen spending and revenues. From a poliical perspecive, his relaionship is imporan especially for Greece, which suffers from fiscal deficis during las years. Budge defici in Greece is presen hroughou he examined period of his sudy. In oher words, budge defici is one of he key characerisics of Greek economy. Greek economy had impressive upward rends afer he Second World War. The coninuous convergence wih developed counries sopped in 1973 due o he firs Oil Crisis. The siuaion became worse in 1979, when he second Oil Crisis burs. In January 22, Greece and oher eleven counries of Eurozone obained a common currency, euro. The accession of Greece in Eurozone was accomplished wih he successful convergence roue of public measures and he realizaion of four ou of five Maasrich crieria on 2 (inflaion, governmen s defici, long erm borrowing rae, exchange rae mechanism, public deb). Gross domesic produc coninued o increase above he average of European Union due o invesmen infrasrucure relaed o he Olympic Games 24. However, from 21 unil 25, Greece violaed he crieria of Sabiliy Treay which referred o he defici under 3%. From he end of 29 and he beginning 21, Greek economy faces serious problems no only due o inernaional economic crisis bu also o unconrolled expenses unil he 29 elecions. I recorded he second larges annual budge defici and he second larges public deb in European Union. Public deb of 29 reached 15,4% of GDP and he increasing deb levels reached 127,1% of GDP. All hese led o a high borrowing cos hus a serious economic crisis. There was an alarm for excessive public deb on 21.In Commission i was argued ha he unpleasan siuaion of some member counries was he resul of he Keynesian borrowing policies ha some policy makers ogeher wih Cenral Bankers of European Union have followed. Many economiss proposed he imposiion of a number of srucural policies in order o conrol public deb such as he imposiion of resricive measures and higher axes. 9

2 ISSN (Prin), (Online) Cener for Promoing Ideas, USA Oher officers claimed ha he emergency measures should bring cruel punishmens o hose counries ha receive help from Communiy. Furhermore, here was a savage criicism agains speculaors for marke manipulaion: Chancellor Angela Merkel saed ha organizaions ha were suppored by public capials, exploi fiscal crisis in Greece and anywhere else. On May 21 a Memorandum signed among Inernaional Moneary Fund, European Union and European Cenral Bank covering he borrowing needs of he counry. When here exiss budge defici in a counry, hen he revenues are lower han spending. Reversely, when governmen spending is less han revenues, here is budge surplus. When governmens need o reduce unemploymen hey use he policy of budge defici. On he oher hand, when he defici exiss for a long ime period, hen an imporan problem arises in sociey and he governmen should find a soluion. One soluion is eiher o reduce governmen spending or increase revenues. Cerainly, he funds of governmen revenues should be sable having he lowes variaions. In order o achieve he aforemenioned aim, each governmen should be familiar wih governmen spending and revenues. I has been observed ha in some cases, revenue increase or expendiures reducion influences is corresponding variable and makes he adoped policy ineffecive. Thus, a governmen should know how he dependen variables affec governmen spending before deciding for he spending reducion or revenues increase. So, i is necessary he governmen o find a relaionship beween governmen spending and revenues o reduce defici. The aim of his paper is o examine his relaionship for Greece for he period , which is in an unfavourable siuaion in relaion o oher counries of EU during las years. The res of his paper is organized as follows: Secion 2 is a brief overview of he heoreical lieraure. Secion 3 gives an empirical lieraure review. Secion 4 describes daa and mehodology. Secion 5 presens he empirical he resuls. Finally, Secion 6 provides he concluding remarks. 2. Theoreical Lieraure Review There are four aspecs abou he relaionship of governmen spending and revenues. The firs one refers ha governmen spending mus be expanded according o revenues. Thus, spending should follow revenues. This means ha if revenues (axes) increase, in ha case governmen can increase spending. So revenues are remedy for minimizing public deficis. This view is suppored by Friedman (1972, 1978) and Blackley (1986) who show ha here is posiive causal relaionship beween revenues and spending. Specifically, Friedman claims ha ax increase will lead o more spending and also he says: A governmen could no reduce defici by increasing axes because his increase will cause more spending hus leaving he defici in a high level. Friedman suggess he ax reducion as a soluion for budge s defici, given ha axes have a posiive causal relaionship wih governmen spending. According o Friedman, a reducion on axes will resul on higher deficis which will influence he governmen o decrease governmen spending. Conrary o Friedman, Buchanan and Wagner (1977,1978) claim ha ax increases resul on governmen spending reducions. According o Buchanan and Wagner, he reducion of governmen spending resrics governmen o finance he budge defici. So, hey argue ha while ax changes drive o spending changes, he relaionship beween hese wo variables is a negaive one. The second view is suppored by Peacock and Wiseman (1961) claiming ha increases on governmen spending generaes increases on revenues. According o his view, he level of spending is firs deermined by he governmen and he revenues ax policy is defined which will accommodae he desired level of spending. Anoher view on his issue is ha of Robers (1978) as far as he deep recessions are concerned. He claims ha he increases on spending and axes are necessary. Anoher version on his view can be found on he works of Barro (1974, 1979, 1986). On his papers, Barro refers o he ax smoohing hypohesis, where governmen spending is considered as an exogenous variable o which axes adjus. Moreover, he ineremporal budges show ha an increase in curren expendiures is mached by higher fuure axes. The hird view is ha governmen can change spending and revenues (axes) a he same ime. This view is suppored by Musgrave (1966) and is referred as fiscal synchronizaion hypohesis which enails ha here is a bilaeral causaliy beween spending and revenues. Furhermore, Barro (1979) suggesed a ax smoohing model for he hypohesis of ax synchronizaion. This model is based on he Ricardian equivalence and suppors ha defici financed by governmen s expendiure oday resuls in fuure ax increase. Finally, he fourh view is suppored by Baghesani and McNown (1994) refers ha governmen spending and revenues is deermined by long run economic growh so a causal relaionship of revenues and spending is no expeced. 1

3 Journal of Business & Economic Policy Vol. 5, No. 1, March Empirical Lieraure Review Even if during he las decades many papers have been published in various counries, he direcion of causal relaionship beween governmen spending and revenues has no ye been found. Many papers refer on he four aspecs menioned in he previous secion. The use of differen economeric mehods and differen periods ended up on differen conradicory resuls. The resuls also differ as far as he direcion of causaliy is concerned having an effec on he economic policymaking of each governmen boh in long and shor run level. For developing counries here have been many sudies which examined he relaionship beween governmen spending and revenues. Shah and Baffes (1994) on heir paper for hree Lain American counries (Argenina, Mexico and Brazil) found a bidirecional causal relaionship beween governmen spending and revenues for Argenina and Mexico whereas for Brazil his relaionship was unidirecional wih direcion from revenues o spending. Owoye (1995) invesigaed he causal relaionship beween revenues and spending for G7 counries. He found a bidirecional causaliy for five ou of seven counries and for Japan and Ialy he found a unidirecional causal relaionship wih direcion from revenues o spending. Hasan and Lincoln (1997) using quarerly daa for Unied Kingdom from and Johansen coinegraion echnique found causal relaionship beween governmen spending and ax revenues. Park (1998) examined causal relaionship beween governmen revenues and spending for Korea for he period The resuls showed a unidirecional causal relaionship from revenues o spending. Abdul e al. (2) examined he causal relaionship beween governmen spending and ax revenues for Malaysia for he period using he Toda-Yamamoo echnique. The resuls of heir paper showed bidirecional causal relaionship beween he examined variables. Kollias and Makrydakis (2) examined he relaionship beween axes and governmen spending for four counries of EU. The resuls of heir paper showed ha he long run relaionship beween he variables is valid only for Greece and Ireland bu no for Spain and Porugal. Furhermore, causaliy resuls showed ha here is a bidirecional causal relaionship for Greece and Ireland, whereas for he oher wo counries his causal relaionship does no exis. Chang e al (22) on heir paper examined he relaionship beween revenues and axes for en indusrial counries including hree Asian counries like Taiwan, Korea and Thailand. On heir paper excep for governmen spending and axaion hey include GDP as a conrol variable. The Johansen coinegraion es showed a coinegraing vecor for seven ou of en indusrial counries (UK, USA, Souh Korea, Japan, Taiwan, Souh Africa and Ausralia). Causaliy esing showed a unidirecional causal relaionship from governmen revenues o spending for UK, USA, Souh Korea, Japan, Taiwan and one way causal relaionship from governmen spending o revenues for Souh Africa and Ausralia. Finally, for New Zealand and Thailand here was no causal relaionship. Al-Qudair (25) examined he long run relaionship beween public spending and revenues for he Kingdom of Saudi Arabia using Johansen coinegraion echnique and error correcion model for causaliy esing. Coinegraion resuls showed he exisence of long run relaionship beween public spending and revenues. Causaliy esing demonsraes he exisence of bidirecional causal relaionship beween governmen spending and revenues in long and shor run basis. Narayan and Narayan (26) on heir paper hey used he Toda and Yamamoo echnique for he counries of Mauriius, Souh Africa, Peru, Guyana, Haii, Chile, Uruguay, Venezuela, Ecuador, El Salvador, Guaemala and Paraguay in order o examined he causal relaionship beween governmen spending and revenues. The resuls showed ha here is a unidirecional causal relaionship from revenues o governmen spending for Mauriius, Chile and El Salvador while for Haii here is evidence for supporing he fiscal synchronizaion hypohesis. Finally, for Souh Africa, Peru, Guyana, Uruguay, Guaemala and Ecuador here is no causaliy beween he examined variables. Emelogu and Uche (21) sudied he relaionship beween governmen spending and revenues in Nigeria using daa from 197 ill 27. Using coinegraion echniques such as Engel-Granger wosep mehod and Johansen procedure, hey found a long run relaionship among variables. Aferwards, causaliy es using error correcion model showed a unidirecional causal relaionship wih direcion from revenues o spending. The empirical paper of Ali and Shah (212) in he case of Pakisan for he period showed ha here is no causal relaionship beween revenues and spending boh in long and shor run level. Apergis e al. (212) examine budgeary disequilibria for Greece using annual daa from and TAR and MTAR models. 11

4 ISSN (Prin), (Online) Cener for Promoing Ideas, USA The resuls of heir paper showed a unilaeral causal relaionship beween revenues and expendiure wih direcion from revenues o expendiure while long run resuls showed asymmeric responses boh in revenues and expendiure. Saysombah and Kyophilavong (213) invesigaed he relaionship beween spending and revenues for Lao People s Democraic Republic during he period 198 unil 21. Applying ARDL coinegraion procedure in combinaion wih Granger causaliy hey found a long run causal relaionship beween spending and revenues wih direcion from spending o revenues. Kollias e al. (214) in heir paper examine he relaionship beween governmen revenues and expendiure in Unied Kingdom beween wo poliical paries. Using quarerly daa from and asymmeric error correcion models, hey conclude ha here are differences in he speed a which fiscal imbalances are correced from Labor pary and Conservaive pary. Labors operae under a sof budge while Conservaives suppor a hard one. Finally, Nwosu and Oka for (214) examined he relaionship beween revenues and spending and divide each one in wo groups. Revenues are divided in revenues on oil and non-oil, whereas spending is divided in curren and capial. This paper employs daa for he period and Johansen coinegraion echnique and error correcion mechanism. The resuls of his paper showed ha oal spending (curren and capial) have a long run and a unidirecional causaliy relaionship wih oal revenues (oil and non-oil) wih direcion from oal spending o oal revenues. 4. Mehodology The link beween governmen spending and revenue is specified as follows: GS GR e (1) and GR 1 1GS (2) where he GS is of he overall governmen spending o GDP raio, and he revenue o GDP raio. The e and are error erms. We expec ha 1 and 1 >. Logarihmic ransformaion of he above equaions would leave he basic equaions as follows: LGS LGR u (3) and LGR 1 1 LGS v (4) L=Naural Logarihms. GR is of he overall governmen 4.1 Order of Inegraion On his secion we es he order of inegraion of ime series. For his es, we use augmened Dickey-Fuller (ADF) es (1979,1981) and Phillips-Perron (PP) (1988). The resuls on he es give he opporuniy o deermine he mos suiable es of series coinegraion or in oher words, he long run relaionship beween hem. 4.2 CoinegraionTess On his paper, we adop he Auoregressive Disribued Lag es (ARDL) as i was formed by he papers of Pesaran and Shin (1999) and Pesaran, Shin and Smih (21). This es in relaion o oher coinegraion es has some advanages such as: I can be used also in series ha are no inegraed same order. I has more power when he sample size is small. I allows he series o have differen lags. I deermines a dynamic model of unresriced error wihin a linear ransformaion. The equaions for he ARDL approach are he following: LGS b p q b1 ilgs i b2 jlgr j 1LGS 1 2LGR 1 (5) i1 j 12

5 Journal of Business & Economic Policy Vol. 5, No. 1, March 218 LGR h p q h1 ilgr i h2 jlgs j 1LGR 1 2LGS 1 (6) i1 j LGS i and LGR j respecively. Where p and q is he lag order of variables We coninue wih he Bounds es on equaions (5) and (6). This es uses F disribuion and he null hypohesis of no coinegraion of series is he following: H and H (No coinegraion of series) : 1 2 : 1 2 agains he alernaive hypohesis of series coinegraion H1 : 1 2 and H 1 : 1 2 (series coinegraion) If he bounds es will lead o series coinegraion we can coninue wih he esimaion of he long run relaionship of series from equaions (7) and (8), as well as he resriced error correcion model from equaion (9) and (1). LGS LGS... LGS LGR LGR 1... LGR LGR LGR 1... plgr p LGS 1LGS 1... qlgs u 1 1 p p 1 q q (7) v 1 q (8) p q LGS c cilgs i d jlgr j 1 z 1 1 i1 j (9) p q LGR g filgr i k jlgs j i1 j (1) Where p and q is he lag order of variables LGS i and LGR j of equaion (9) and LGR i and LGS j of equaion (1) respecively. The erms z and are he errors erms which are creaed by he coinegraing regressions of equaions (7) and (8). 4.3 Τesing Sabiliy in ECM The exisence of dynamic resriced error correcion model which comes from equaions 9 and 1 does no necessarily imply ha he esimaed coefficiens are sable. For his reason, Pesaran e al. (1995, 21) suggesed he esimaed coefficien sabiliy es on he esimaed models using he Brown e al. ess (1975) known as cumulaive sum of residuals (CUSUM) and cumulaive sum of squares residuals (CUSUMSQ) sabiliy ess. 4.4 Causaliy Analysis On his secion we examine he causal relaionship beween governmen spending and revenues using a seemingly unrelaed regression model. Toda and Yamamoo (1995) in order o invesigae causaliy, hey developed a mehod based on he esimaion of an adjused VAR model (k+d max ), where k is he opimal ime lag on he firs VAR model and d max is he larges inegraion order on he variables of he VAR model. On his paper, we adop he Toda and Yamamoo causaliy es (1995) conrary o he radiional Granger es (1969) for he following reasons: Granger es can provide spurious regressions o he funcions wih ime lags of inegraed variables. F-es can be used only when he variables are coinegraed. Error correcion model developed by Engle and Granger (1987) and VAR correcion model developed by Johansen and Juselius (199) as alernaive causaliy ess are cumbersome and sensiive o he values of he parameers in finie samples and herefore heir resuls are unreliable (see Toda and Yamamoo 1995, and Zapaa and Rambaldi 1997). The augmened causal approach inroduced by Toda Yamamoo (1995) applies he asympoic Chi-square disribuion. The mos imporan advanage on his procedure is ha i is no essenial o pre-es he inegraion or co inegraion order beween he variables (see Toda Yamamoo, 1995 and Dolado & Lükepohl, 1996). Toda and Phillips (1993) on heir paper claim ha Granger causaliy wih error correcion es can lead o misaken conclusions from he dependence of some asympoic parameers in some cases. Finally, according o Rambaldi and Doran (1996), Toda and Yamamoo es for non-granger causaliy is done wih he modified Wald es (MWald) and on Seemingly Unrelaed Regression models (SUR models). 13

6 ISSN (Prin), (Online) Cener for Promoing Ideas, USA VAR model of Toda and Yamamoo causaliy is formed as follows: k d k d LGS max max 1 LGS i 2LGS i 1 LGR i 2LGR i 1 (11) i1 ik1 i1 ik1 k d k d LGR max max 1 LGR i 2LGR i 1 LGS i 2LGS i 2 (12) i1 ik1 i1 ik1 Where he opimal ime lag of he firs VAR model and d max is he larges inegraion order on he variables of he VAR model. The null hypohesis of no causaliy is defined for every equaion on VAR model. For example LGR variable cause LGS variable ( LGR => LGS ) when, i. Toda and Yamamoo es for no Granger causaliy can be done for every inegraion order of variables, eiher hey are coinegraed or no, given ha he reverse roos of auoregressive polynomial should be inside of he uni circle. Thus, he Toda and Yamamoo causaliy es will be valid. 5. Empirical Resuls 5.1 Daa On diagram 1he overall governmen spending o GDP raio and governmen revenue o GDP raio for Greece is presened for he period On his diagram we have o poin ou ha governmen spending all hrough he examined period is larger han revenues. Diagram 1: The Governmen Spending and Governmen Revenues as Percen of GDP beween 198 and GS The sudy uses annual ime series daa and covers he period 198 o 215. The daa were obained from he Inernaional Financial Saisics (IFS).All daa are in real erms. The logarihm of he overall governmen expendiures o GDP raio and he logarihm of overall governmen revenues o GDP raio are used in he empirical analysis. This daa ransformaion is occurred in order o reduce he heeroscedasiciy problem (see Gujarai 24). 5.2 Order of Inegraion Table 1.UniRooTess Variable ADF P-P C C,T C C,T LGS ()*** -3.48()*** []*** -3.49[]*** ΔLGS ()* ()* [1]* [3]* LGR -1.47() -1.93() -1.54[1] [1] ΔLGR -5.68()* ()* [1]* [1]* Noes: 1. *, ** and *** show significan a 1%, 5% and 1% levels respecively. 2. The numbers wihin parenheses followed by ADF saisics represen he lag lengh of he dependen variable used o obain whie noise residuals. 3. The lag lenghs for ADF equaion were seleced using Schwarz Informaion Crierion (SIC). 4. Mackinnon (1996) criical value for rejecion of hypohesis of uni roo applied. 14 GR

7 Journal of Business & Economic Policy Vol. 5, No. 1, March The numbers wihin brackes followed by PP saisics represen he bandwidh seleced based on Newey Wes (1994) mehod using Barle Kernel. 6. C=Consan, T=Trend, Δ=Firs Differences, L=Naural Logarihms. The resuls on able 1 show ha series exhibi differen inegraion order. The governmen spending series is null order I() in 1% level of significance whereas he governmen revenues series is inegraed firs order I(1). Thus, for he long run relaionship of he series he mos suiable is ha of Pesaran e al (21), he auoregressive disribued lag (ARDL) mehodology. 5.3 ARDL Bounds Tesing Approach From equaions (5) and (6) of unresriced error model we can find he maximum values of p and q lags using he Final Predicion Error (FPE), Akaike Informaion Crierion (AIC), Schwarz Informaion Crierion (SIC), Hannan- Quinn Crierion (HQC), and Likelihood Raio (LR) crieria. The resuls of he secrearial are presened on able 2. Table 2. VAR lag order selecion crieria Lag LogL LR FPE AIC SBC HQC equaion (5) NA *.22* * * * equaion(6) NA *.21* * * * Noes: *denoes he opimal lag selecion The resuls on able 2 show ha in all crieria, he maximum number of lags for he series on boh equaions is 1. The order of opimal lag lengh on equaions (5) and (6) is chosen from he minimum value of AIC, SBC and HQC crieria. On able 3 we presen here sals of he secrearial. Table 3.Order of opimal lags ARDL (p,q) ARDL(p,q) AIC SBC HQC Equaion (5) (p=1, q=) (p=1. q 1 =1) Equaion (6) (p=1, q=) (p=1. q 1 =1) Noes: *denoes he opimal lag selecion, Saisics in bold denoe he value of he minimized AIC, SBC and HQC. There Sulson able 3 show ha ARDL (p,q) model wih p=1 q= lags is he bes for boh equaions. Coninuing on able 4, we employ he error in dependence es (LM es) unil firs order (maximum number of lags). Table 4.Error Independence Tes (LM Tes) Equaion (5) F-sa =1.384 Prob. F(1,28)=.249 N*R 2 =1.62 Prob. X 2 (1)=.256 Equaion (6) F-sa =2.453 Prob. F(1,28)=.142 N*R 2 =2.672 Prob. X 2 (1)=.121 Noes: Ν=observaions. 15

8 ISSN (Prin), (Online) Cener for Promoing Ideas, USA The resuls on he above able presen ha errors are no auo correlaed. We coninue wih dynamic sabiliy es of ARDL(1,) model for boh equaions. This es is employed wih uni cycle. If reverse roos of equaions (5) and (6) are inside he uni cycle hen he models are dynamically sable. 1.5 Diagram 2: Dynamic Sabiliy of Models equaion (5) equaion (6) Inverse Roos of AR Characerisic Polynomial 1.5 Inverse Roos of AR Characerisic Polynomial The resuls of diagram 2 show ha here is a dynamic sabiliy of models on boh equaions. I is advisable before coninue wih bounds es, o presen he acual and fied residuals from boh equaions ARDL(1,) auoregressive unresriced error correcion model. Diagram 3: Acual and Fied Residuals of Models equaion (5) equaion (6) Residual Acual Fied Residual Acual Fied We coninue by conducing coinegraion es of bounds auoregressive disribued lag. In oher words, we es if φ 1 and φ 2 as well as π 1 and π 2 coefficiens are null on our esimaed models. Table 5.Bounds Tes (Wald Tes) Tes Saisic Value df Probabiliy equaion (5) F-saisic 4.86* (2,29).86 Chi-square (2).69 equaion (6) F-saisic (2,29).137 Chi-square (2).1155 Noes: Table CI (iii) page 3 of Pesaran e al. 21 gives lower and upper bounds for 1%, 5% and 1% level of significance [4.4, 4.78], [4.94, 5.73] and [6.84, 7.84] respecively. *, ** and *** showsignificancea1%, 5% and 1% levels respecively. 16

9 Journal of Business & Economic Policy Vol. 5, No. 1, March 218 The resuls on he above ables how ha F-saisic value is larger only on equaion (5) from he upper bound on Pesaran e al ables (21) for 1% level of significance and (k+1)=2 variables. Thus, we say ha here is a coinegraing relaionship beween examined series only on equaion (5) for 1% level of significance. On he following able he resuls from he esimaion of unresriced error correcion model are presened (equaion 5). Table 6. Esimaion of Unresriced Error correcion Model Dependen variable = ΔLGS Shor run analysis Variables Coefficien -saisic Consan ΔLGS ΔLGR LGS LGR R F-sa D-W Diagnosic Tes X 2 Probabiliy Normalily (2).256 Serial Corr. 1.62(1).25 ARCH.775(1).378 Noes: ***, ** and * show significan a 1%, 5% and 1% levels respecively. Δ denoes he firs difference operaor, X 2 Normal is for normaliy es, X 2 Serial for LM serial correlaion es, X 2 ARCH for auoregressive condiional heeroskedasiciy, ( ) is he order of diagnosic ess. The resuls on able 6 show ha boh saisic and diagnosic ess are quie saisfying. Before coninuing on he nex sep, we ge he long run resuls from he unresriced error correcion model equaion (5). LGR LGS.237 So, we can sress ha an increase of governmen revenues by 1% will cause an increase on governmen spending by.48% approximaely. We proceed o esimae he long and shor run relaionship of he series on equaions (7) and (9). Table7.Esimaion of he Long and Shor Run Relaionship Dependen variable = LGS Long run analysis Variables Coefficien -saisic Consan 1.181*** LGR.722*** 12.9 R 2.83 F-sa D-W.56 Diagnosic Tes X 2 Probabiliy Normalily.88 (2).667 Serial Corr (1).231 ARCH.3(1).583 Dependen variable = ΔLGS Shor run analysis Variables Coefficien -saisic Consan.2421* ΔLGS * ΔLGR ** z *** R F-sa D-W Diagnosic Tes X 2 Probabiliy Normalily 2.534(2).452 Serial Corr..7(1).978 ARCH.154(1)

10 ISSN (Prin), (Online) Cener for Promoing Ideas, USA Noes: ***, ** and * show significan a 1%, 5% and 1% levels respecively. Δ denoes he firs difference operaor, X 2 Normal is for normaliy es, X 2 Serial for LM serial correlaion es, X 2 ARCH for auoregressive condiional heeroskedasiciy and X 2 Whie for whie heeroskedasiciy. ( ) is he order of diagnosic ess. The resuls of unresriced error correcion model ECM (able 6) differ from hose of resriced error correcion model ECM (able 7). Bu coefficiens signs on boh models are in accordance wih heory, bu he significance is more srong on he resriced ECM. Furhermore, he error correcion erm coefficien is negaive and saisically significan in 1% level implying ha he speed in adjusmen in long run equilibrium is.1% approximaely. In oher words, each disequilibrium caused by governmen revenues can be reformed by.1% in one year. The resuls on able 7 show ha boh saisic and diagnosics sare quie saisfying. There sriced dynamic error correcion model, derived by ARDL bounds es hrough a simple linear ransformaion, incorporaes he shor run dynamic wih long run equilibrium. The negaive and saisical significan esimaion of coefficiens on error correcion erms z -1 on equaion (9) show a long run relaionship beween he examined variables. On he following diagrams (3) and (4) we examine he dynamic sabiliy of resriced error correcion model wih Brown e al. (1975) ess. Figure 3: Plo of Cumulaive Sum of Recursive Residuals CUSUM 5% Significance Figure 4: Plo of Cumulaive Sum of Squares of Recursive Residuals CUSUM of Squares 5% Significance From he above diagrams we can see ha here is a dynamic sabiliy on model s coefficiens ha we examine. 5.4 Toda Yamamoo Causaliy Tes Table 8 presen he resuls on causaliy es of Toda and Yamamoo according o equaions 11 and 12. Table 8. Toda and Yamamoo no-causaliy es. Excluded Lag(k) Lag(k+dmax) Chi-sq Prob. Direcion of Causaliy Dependen variable: LGS LGR LGS=>LGR* Dependen variable: LGR LGS LGR # LGS Noes: The (k+dmax ) denoes VAR order. The lag lengh selecion was based on LR:sequenial modified LR es saisic (each es a 5% level), FPE: Final predicionerror, AIC: Akaike informaion crierion, SC: Schwarz 18

11 Journal of Business & Economic Policy Vol. 5, No. 1, March 218 informaion crierion, HQ:Hannan-Quinn informaion crierion. ***, ** and * denoes 1% and 5%, 1%significance level, respecively. =>denoes one - way causaliy, #denoes no causaliy. EViews 9. was used for all compuaions. The resuls on he above es show ha here is a unidirecional causal relaionship beween spending and revenues for Greece wih direcion from governmen revenues o spending. 6. Summary and Conclusions In his paper, an effor was made in order o find he causal relaionship beween governmen spending and revenues in Greece. For his analysis, we used annual daa for he period We examined he relaionship beween governmen spending and revenues in Greece, using Pesaran e al. (21) coinegraion given ha daa had differen inegraion order. Aferwards, we es he direcion of causaliy among he examined variables using he Toda and Yamamoo mehodology. The resuls of our paper suppor he Friedman, Buchanan and Wagner hypohesis ha governmen revenues cause spending. The deerminaion of causal relaionship beween hese wo macroeconomic measures is of vial imporance in order o ensure and adop he mos suiable ax policy. The resuls of his paper also show ha here is a long run relaionship beween governmen revenues and spending. These resuls also show ha an increase of governmen revenues by 1%, increase he governmen expendiure only by.48% approximaely. Even if GDP enlargemen by average was higher from he average of he counries of European Union in 199, beween 21 and 28, counry s GDP increased on average annual rae 4,3% compared o ha of Eurozone which was 3,1%.A he same ime, governmen expendiure increased by 87% while revenues increased only by 31%. This resuled on deficis which were beyond he rules of Sabiliy and Developmen Pac. The boom of hese deficis can be explained from an insufficien public adminisraion, a cosly reiremen sysem and a huge ax evasion. However, Greek economy faced and coninues o face serious problems including high unemploymen, bureaucracy and corrupion. Due o inernaional economic crisis, he enlargemen rae of economy had a negaive sign for he firs ime on 29 afer nineeen years. In 29, defici was above 13% of GDP. A he same ime, he counry financed he deficis of inernaional capial markes. The reurns of 1-year Greek bonds were 1-4 unis above he German ones. During crisis, he Greek bonds were 4 unis below he German ones. This had as a resul he disloyaly of inernaional invesors o Greek economy. In 29, Greece had he second lowes grading in EU (afer Poland) according o he Index of Economic Freedom. Greece suffers from high levels of poliical and economic corrupion and low compeiiveness compared o European parners. Furhermore, he inernaional economic crisis of 28 led o a dramaic liquidiy problem in Greece in such a way ha i could no face is deb obligaions. The resul was he inroducion of auseriy measures by he governmen including dramaic reducions on expendiure and increases on ax revenues. Sill, hese measures didn soohe invesors in inernaional capial markes. As a resul, European Union and Inernaional Moneary Fund provided help o he Greek governmen in order o avoid he defaul of deb and also o avoid he domino effec o he oher counries of Eurozone. (Apergis e al. 212). Today Greece faces a serious crisis due o is deb. This paper examined he long and shor run beween revenues and expendiures, heir influence on public deb as well as he causal relaionship beween revenues and expendiures. Causaliy resuls show a unidirecional causal relaionship wih direcion from governmen revenues o expendiures. This oucome is in accordance wih he paper of Apergis e al (212) as far as he causaliy beween he wo variables is concerned. This resul also highlighs ha he increase of expendiures can be accomplished only wih he increase of revenues so ha he economy can recover. In anoher case, he defici will expand resuling in a larger deb. Therefore, o sop his policy governmen should: Reduce he size of large consecuive spending and urns o invesmens spending. Should reduce funcion s cos. Should differeniae is economic policy and ry o find ou oher revenue sources (apar from axes) in a way ha will repair he difference beween revenues and spending reducing hus reducing budge s defici. Finally, axes play an imporan role in he economy. Taxes on various secors should be reformed in such a way ha economy will sar wih new invesmen which will bring more revenues. 19

12 ISSN (Prin), (Online) Cener for Promoing Ideas, USA References Abdul Aziz, M, Muzafar, S.H, W.N.W Azman-Saini, and M. Azali (2). Tesing For Causaliy beween Taxaion and Governmen Spending: An Applicaion of Toda-Yamamoo Approach. Peranika Journal of Social Science &Humaniies, 8(1), Al-Qudair, K.H.A., (25). The relaionship beween governmen expendiure and revenues in he Kingdom of Saudi Arabia: Tesing for co-inegraion and causaliy. Journal of King Abdul Aziz Universiy: Islamic Economics, 19(1), Ali, R. and M. Shah, (212). The causal relaionship beween governmen expendiure and revenue in Pakisan. Inerdisciplinary Journal of Conemporary Research in Business, 3(12), Apergis, N., Payne, J., and J. Saunoris., (212). Tax-spend nexus in Greece: Are here asymmeric?,journal of Economic Sudies, 39(3), Baghesani, H. and R. McNown (1994). Do revenues or expendiures respond o budgeary disequilibria?, Souhern Economic Journal, 63, Barro, R. J. (1974). Are Governmen Bonds No Wealh?, Journal of Poliical Economy, Nov.-Dec., pp Barro, R. J. (1979). On he Deerminaion o he Public Deb, Journal of Poliical Economy, 81, Barro, R. J. (1989). The Ricardian Approach o Budge Deficis, Journal of Economic Perspecives 3, Blackley, R. (1986). Causaliy beween revenues and expendiures and he size of federal budge, Public Finance Quarerly, 14, Buchanan, J, M. and R. W. Wagner. (1977). Democracy in Defici:The Poliical Legacy of Lord Keynes, Academic Press, New York. Buchanan, J, M. and R. W. Wagner. (1978). Dialogues Concerning Fiscal Religion.Journal of Moneary Economics 4, Brown, R.L., Durbin, J., Ewans, J.M., (1975). Techniques for esing he consance of regression relaions overime. Journal of he Royal Saisical Sociey. 37, Chang.Τ,, W. R. Liu, ands. B. Caudill (22).Tax-and-spend, spend-and-ax, or fiscal synchronizaion: new evidence for en counries, Applied Economics, 34, Dickey, D., and Fuller, W.A. (1979). Disribuion of he esimaes for auoregressive ime series wih a uni roo. Journal of he American Saisical Associaion 74(366), Dickey, D., and Fuller, W.A. (1981). Likelihood raio saisics for auoregressive ime series wih a uni roo. Economerica, 49(4), Dolado, J.J. and Lükepohl, H. (1996). Making Wald ess work for coinegraed VARsysem. Economeric Reviews, 15(4), Emelogu, C.O. and M.O. Uche. (21). An examinaion of he relaionship beween governmen revenue and governmen expendiure in Nigeria: Co-inegraion and causaliy approach. Cenral Bank of Nigeria Economic and Financial Review, 48(2), Engle, R.F., Granger, C.W.J., (1987). Coinegraion and error correcion: represenaion, esimaion and esing. Economerica, 55, Friedman, M. (1972). An economis's proes, Thomas Horon and Company, New Jersey. Friedman, M. (1978). The limiaions of ax limiaion, Policy Review, Gujarai, D.N. (24). (Fourh Ediion). Basic economeric, New York: The McGraw-Hill companies Inc. Granger, C. W. J. (1969). Invesigaing causal relaions by economeric models and cross-specral mehods. Economerica. 37(3), Hasan, M. and Lincoln, I., Tax hen spend or spend hen ax? Experience in he UK, Applied Economics Leers, vol. 4 (4), pp Johansen, S., Juselius, K., (199). Maximum likeliwood esimaion and inference on coinegraion-wih applicaion o he demand for money. Oxford Bullein of Economics and Saisics, 52, Kollias, C., and S. Makrydakis (2). Tax and spend or spend and ax? Empirical evidence from Greece, Spain, Porugal, and Ireland.Applied Economics, 32(5), Kollias, C., Papadamou, S., and I, Psarianos., (214). Fiscal imbalances and asymmeric adjusmen under Labour and Conservaive governmens in he UK.Research in Economics, 68(3),

13 Journal of Business & Economic Policy Vol. 5, No. 1, March 218 MacKinnon, J.G. (1996). Numerical disribuion funcions for uni roo and coinegraion ess. Journal of Applied Economerics, 11(6), Musgrave, R. (1966). Principles of budge deerminaion. In Public Finance: Seleced Readings eds. H. Cameron and W. Henderson. New York: Random House. Narayan, P.K. and Narayan, S., 26. Governmen Revenue and Governmen Spending Nexus: Evidence from Developing Counries. Applied Economics, 38(3), Newey, W.K. and K.D.Wes (1994). Auomaic lag selecion in covariance marix esimaion, Review of Economic Sudies, 61, Nwosu, D.C and Okafor, H.O. (214). Governmen revenue and expendiure in Nigeria: A disaggregaed analysis, Asian Economic and Financial Review, 4(7), Owoye, O. (1995). The causal relaionship beween axes and expendiures in he G7 counries: Coinegraion and error correcion models. Applied Economics Leers, 2(1), Park, W. K. (1998). Granger causaliy beween governmen revenues and expendiures in Korea. Journal of Economic Developmen, 23(1), Peacock, A.T. and J. Wiseman, (1961). The growh of public expendiure in he Unied Kingdom, Princeon Universiy Press, Princeon: NBER. Pesaran, M.H., Shin, Y. (1995). An auoregressive disribued lag modelling approach o coinegraion analysis. Cambridge Universiy, Deparmen of Applied Economics, DP No Pesaran, M. H., Shin, Y., and Smih, R. J. (21). Bounds esing approaches o he analysis of level relaionships. Journal of Applied Economerics, 16, Phillips, P.C., and Perron, P. (1988). Tesing for a uni roo in ime series regression. Biomerika, 75(2), Rambaldi, A. N. and Doran, H. E. (1996). Tesing for Granger non-causaliy in coinegraed sysems made easy, Working Papers in Economerics and Applied Saisics, Deparmen of Economerics, The Universiy of New England, No.88. Robers, P.C. (1978). Idealism in Public Choice Theory. Journal of Moneary Economics, (Augus), Saysombah, P and Kyophilavong, P (213). The causal link beween spending and revenue: The Lao PDR, Inernaional Journal of Economics and Finance, 5(1), Shah, A. and Baffes, J. (1994). Causaliy and co-movemen beween axes and expendiure: Hisorical evidence from Argenina, Brazil and Mexico. Journal of Developmen Economics, 44(2), Toda, H. Y. and Phillips, P. C. B. (1993). Vecor auoregressions and causaliy, Economerica, 61(6), Toda, H.Y. and Yamamoo (1995). Saisical inference in vecor auoregressions wih possibly inegraed processes. Journal of Economerics, 66, Zapaa, H.O. and A.N. Rambaldi(1997). MoneCarlo evidence on coinegraion and causaion.oxford Bullein of Economics and Saisics59,