The Relationship between Government Revenue and Expenditure in Qatar: A Cointegration and Causality Investigation

Transcription

1 Inernaional Journal of Economics and Finance; Vol. 4, No. 9; 2012 ISSN E-ISSN Published by Canadian Cener of Science and Educaion The Relaionship beween Governmen Revenue and Expendiure in Qaar: A Coinegraion and Causaliy Invesigaion Abdulla S. Al-Khulaifi Correspondence: P.O.Box 1112 Doha, Qaar. Tel: a.alkhulaifi@qu.edu.qa Received: June 28, 2012 Acceped: July 15, 2012 Online Published: July 31, 2012 doi: /ijef.v4n9p142 URL: hp://dx.doi.org/ /ijef.v4n9p142 Absrac The main purpose of his paper is o examine he relaionship beween governmen revenue and expendiure in Qaar. Annual daa for he period from were used and a ime series analysis of uni roo ess, Engle-Granger s coinegraion and Granger causaliy were applied. The paper found ha variables were inegraed of order one when Augmened Dickey-Fuller and Phillip-Perron uni roo ess were employed. The variables were found o be coinegraed. Empirical resuls found unidirecional causaliy ha runs from governmen revenue o expendiure in suppor of revenue-spend hypohesis. Keywords: coinegraion, granger causaliy, governmen revenue, governmen expendiure, Qaar JEL: C22, H2, H5, H62 1. Inroducion The relaionship beween governmen revenue and expendiure is a major concern for economiss and policy makers alike. This conroversial issue has been he subjec of exensive heoreical and empirical research for decades. This research became more imporan and relevan since governmens have been incurring coninuous budge defici in boh developed and developing counries. Undersanding his relaionship is an imporan elemen for an effecive fiscal policy. The sudy of he causal link beween governmen revenue and expendiure has resuled in four hypoheses ha have been under empirical scruiny in he lieraure. The firs hypohesis is he ax (revenue)-spend hypohesis. I suggess ha changes in revenues induce change in expendiures, bu no he reverse. In his case governmen revenue is said o Granger cause governmen expendiure. The second is he spend-ax (revenue) hypohesis. This hypohesis akes he opposie view of he firs one. Changes in governmen expendiure are said o Granger cause governmen revenue wihou a feedback. The hird hypohesis is he fiscal synchronizaion. This hypohesis proposes ha revenue and expendiure decisions are made joinly. This implies bidirecional causaliy beween he wo variables. The fourh hypohesis is he insiuional separaion of governmen revenue and expendiure decisions of governmen. In his case here is no causaliy beween he wo variables, and hence hey are independen of one anoher. Fiscal policy occupies an imporan saus among economic policies in Qaar. The use of fiscal policy, in addiion o achieving is macroeconomic goals of economic growh, employmen and price sabiliy, is used o achieve economic and social developmen. Therefore, Qaar sae budge is used o promoe and achieve he governmen economic goals, hrough direcing governmen expendiure owards building infrasrucure, educaion, healhcare and housing o raise sandard of living. Governmen expendiure acs as a major componen of GDP in Qaar averaging abou 30% in he las decade. During he period of sudy, governmen expendiure increased from 11 billion Qaari Riyals (QR) in 1980/81 o a record QR Billion in fiscal year 2011/2012. The successive annual budges have shown increasing expendiure on economic services, social welfare, housing and infrasrucure consrucion all for he goal of improving qualiy of life of ciizens. Curren expendiure increased over he period o accommodae he increase in wages and salaries which consiue a large porion of sae budge. Governmen revenue on he oher hand, is procyclical due o is reliance on oil revenue and hence oil prices. Oil prices are deermined in he global marke by supply and demand. Therefore, governmen revenue has experienced a sharp decline when oil prices decreased. This was he case in he mid-1980s. This forced he governmen o decrease is expendiure, and experience budge defici for he years and from Deficis were financed from previous years surplus and from reserves. Recen years revenues have experienced an increase in oil and gas revenues. Anoher inflow o he budge comes from invesmen revenue. This iem has been increasing in he las decade since he governmen uses surplus in he 142

2 Inernaional Journal of Economics and Finance Vol. 4, No. 9; 2012 sae budge in global invesmens, which led o an increase in governmen revenue. Table (1) shows governmen revenue, expendiure and surplus (defici) for Qaar for seleced years of he period under invesigaion. Table 1. Governmen Revenues and Expendiures for Seleced Years (Mill Q.R.) Year Gov. Revenue Gov. Expendiure Surplus (Defici) (183) (2106) Qaar Cenral Bank, Annual Repors. Auhor s Calculaion. I is herefore imporan o empirically sudy he causal relaionship beween governmen revenue and expendiure in Qaar s economy. The main purpose of his paper is o invesigae he relaionship beween governmen revenue and expendiure in Qaar for he period from The paper is divided ino five secions. Following his inroducion, lieraure review of relevan sudies will be presened. Secion hree will discuss daa and mehodology ha is used in his research. Empirical resuls are repored in secion four. Secion five will conclude his exercise. 2. Lieraure Review The lieraure on researching he relaionship beween governmen revenue and expendiure are exensive for many counries in boh developed and developing economies. Tsen and Kian-Ping (2005) examined his relaionship in Malaysia for he period from Augmened Dickey-Fuller and Phillips-Perron Uni roo ess, Johansen coinegraion and error correcion models were applied o daa. The resuls suppored ax-spend hypohesis. Governmen revenue was found o Granger cause expendiure in Malaysia. Nyamongo e al. (2007) in a sudy of he governmen revenue and expendiure nexus in Souh Africa found differen resuls. A monhly daa was used, and modified uni roo es and Vecor Error Correcion Model (VECM) were applied on daa. I was found ha governmen revenue and expendiure are coinegraed, and a long-run relaionship exiss beween hem. Applying Granger causaliy hrough VECM model, i was found bidirecional Granger causaliy which suppors fiscal synchronizaion hypohesis. In he shor-run no Granger causaliy was found beween variable, suggesing fiscal neuraliy hypohesis in Souh Africa for he period of sudy. Amoah and Loloh (2008) sudied he causal linkages beween governmen revenue and spending in Ghana. Researchers used annual daa for he period from which was convered ino quarerly daa. Real and nominal values of variables were used in models. Time series variables were found o be inegraed of order one for boh nominal and real values. Engle-Granger bivariae mehodology was applied o es for exisence of coinegraion. I was found ha he revenue and expendiure are coinegraed. An error correcion model was specified, long-run and shor-run causaliy were obained. Causaliy was found o run from expendiure o revenue in he long-run supporing spend-ax hypohesis, while revenue Granger cause expendiure in he shor-run in suppor of ax-spend hypohesis. Eia and Mbazima (2008) found unidirecional causaliy beween governmen revenue and expendiure in he case of Namibia. Granger causaliy hrough vecor auoregression (VAR) mehod and annual daa for he period from were used. The paper provided evidence o suppor ax-spend hypohesis in Namibia. In Pakisan, The verdic is sill ou for he causaliy direcion beween governmen revenue and expendiure. Aisha and Khaoon (2009) in a sudy of Pakisan economy for he period from , used a ime series mehodology of uni roo es, Engle-Granger approach of coinegraion and error correcion model were applied o daa. Governmen revenue and expendiure were found o be coinegraed. Where Granger causaliy es was performed, i was found ha governmen expendiure Granger cause revenue verifying spend-ax hypohesis. Subhani e al. (2012) found he opposie causaliy direcion confirming he ax-spend hypohesis. They sudied 143

3 Inernaional Journal of Economics and Finance Vol. 4, No. 9; 2012 he causaliy direcion beween governmen expendiure and revenue for Pakisan. Annual daa for he period from were used, and Granger causaliy was applied o variables in quesion. The paper found ha governmen revenue Granger cause governmen expendiure in Pakisan for he period under invesigaion. Sadiq (2010) on he oher hand, found no causaliy beween he wo variables in Pakisan. Granger causaliy was used o sudy he relaionship beween federal and provincial axes and expendiure in Pakisan for he period from o No Granger causaliy was found beween revenue and expendiure in he federal and provincial level for he period of sudy. Ravinhirakumaran (2011) examined he relaionship beween governmen revenue and expendiure in Sri Lanka for he period from A ime series mehodology of Engle-Granger s approach of coinegraion and error correcion model framework is invesigaed. The sudy concluded ha bidirecional causaliy exiss beween governmen revenue and expendiure and here is long-run equilibrium beween he wo variables in Sri Lanka economy. Elyasi and Rahimi (2012) found bidirecional causaliy beween governmen revenue and expendiure in Iran. Annual daa for he period from were used, and variables were esed for saionariy. The paper included a comprehensive lis of sudies on causaliy beween governmen revenue and expendiure for counry specific and for muli-counries sudies. The evidence cied on he direcion of causaliy is mixed in hose sudies. Differen daa ses, economeric mehodologies and differen counry characerisics are some of reasons cied for he differen resuls on he direcion of causaliy. The evidence on he relaionship beween governmen revenue and expendiure for Middle Easern counries is scarce. I is more so for he Gulf counries and especially for Qaar. These sudies are imporan o our paper since mos of he counries are developing counries, oil producing and oil dependen in heir sources of revenue. Peanla and Sadeghi (2012) invesigaed he relaionship beween governmen revenue and expendiure in fifeen oil-exporing counries. Panel VAR framework was applied on annual daa for he period from Oil revenue was used as a proxy for governmen revenue, since i accouns for mos of oal governmen revenue. The sudy found ha here is a unidirecional long-run causaliy ha runs from oil revenue o governmen expendiure. Fasano and Wang (2002) is a leading sudy ha has esed he relaionship beween governmen spending and revenue for he GCC counries. Those counries include Bahrain, Kuwai, Oman, Qaar, Saudi Arabia and The Unied Arab Emiraes. To he bes of he auhor's knowledge, his is he only sudy ha included Qaar in his subjec. Annual daa for he period were used for all counries under invesigaion. The empirical resuls suppored revenue-spend hypohesis for Bahrain, Unied Arab Emiraes and Oman. Governmen revenue was found o Granger cause expendiure in hese hree counries. The evidence of bidirecional causaliy was found for Kuwai, Qaar and Saudi Arabia. Al-Qudair (2005) sudied he relaionship beween governmen expendiure and revenue in Saudi Arabia. A counry ha depends heavily on is oil expors as he main source of governmen revenue. A ime series analysis of uni roo, Engle-Granger coinegraion echnique and an error correcion model were used. The sudy found bidirecional causaliy exiss beween governmen revenue and expendiure in suppor of fiscal synchronizaion hypohesis. The resul confirms earlier findings by Fasano and Wang (2002) on Saudi Arabia. Baainah (undaed) in a sudy of Jordanian economy applied Johansen coinegraion approach and error correcion model. The resul suppored he spend-revenue hypohesis. Governmen expendiure was found o Granger cause revenue for he period from Daa and Mehodology Daa on governmen revenue and governmen expendiure were obained from Qaar Cenral Bank annual repors. The variables were in nominal values. GDP deflaor was used o conver nominal ino real values of variables. This variable was available from Qaar Saisical Auhoriy web sie. Annual daa for he period from are used in his sudy. Since mos ime series variables are non-saionary, uni roo ess are useful o deermine he order of inegraion of he variables. The firs sep in he empirical esimaion is o es perform uni roo es as o see he univariae characerisics of variables, which show wheher he variables are saionary or non-saionary. If variable is saionary, i is said o be inegraed of order zero or I (0). If a variable is no saionary in is level form bu saionary in is firs differenced form, i is said o be inegraed of order one or I (1). One of he mos popular ess of uni roo is he Augmened Dickey Fuller (ADF) es. The esing procedure for he ADF is applied on he following model: 144

4 Inernaional Journal of Economics and Finance Vol. 4, No. 9; a a a c (1) 0 1 Where is he series being esed, is a linear ime rend, ρ is he number of lagged differences included o capure any auocorrelaion, and Δ is he firs difference operaor. The null hypohesis for uni roo is H0: a 2 =0. This es will be applied o each of he variables o deermine he saionariy propery in heir levels and in heir firs-difference. Anoher uni roo es is he Phillip-Perron (1988) es. This es proposes an alernaive nonparameric mehod o ake care of serial correlaion in error erms wihou adding lagged difference erms when esing for a uni roo (Gujarai and Porer 2009). This sudy will apply boh ess of uni roo. In economics wo variables are coinegraed if hey have a long-run or equilibrium relaionship beween hem. In he case of wo variables and Y, ha are inegraed of order d, I (d). If here exiss a linear combinaion such ha he disurbance erm ε from regression is of a lower order of inegraion, I(d-b), and b>0, hen Engle and Granger (1987) define and Y as coinegraed. Thus, if and Y were boh I(1), and residuals ε are I (0), he wo series would be coinegraed (Harris 1995:52). The Engle and Granger (1987) wo sep procedures for modeling he relaionship beween coinegraed variables is as follow: Firs, The order of inegraion of he variables and Y is deermined by employing he uni roo es. We run he OLS regression on wha is called he coinegraing regression: 2 a0 a1y Second, we es wheher he residuals ε, from he regression are saionary. If and Y no coinegraed, any linear combinaion of hem will be non-saionary, and hence he residuals will be non-saionary (pindyck and Rubinfeld 1998). The concep of causaliy is due o Granger (1969), and according o him a variable is caused by Y if is beer prediced from pas values of Y and ogeher raher han from pas values of alone. Four paerns of causaliy can be disinguished: a) unidirecional causaliy from o Y, denoed Y, b) unidirecional causaliy from Y o, denoed Y, c) feedback or bidirecional causaliy, and d) no causaliy. The simples Granger causaliy es is: i1 i i i (2) Y a 0 a1 i i a2iy i b1i Y i b2i i b0 (4) (3) The Granger causaliy can be esed hrough he null hypohesis in equaion (3) is ha Y does no cause, ha is a2i=o for all i, and for equaion (4) he null hypohesis is does no cause Y, or b2i=0 for all i. If a2i=0 and b2i=0 for all i, we conclude ha here is no causaliy beween and Y. If boh of he null hypoheses are rejeced, ha is a2i 0 and b2i 0, hen here is bidirecional relaionship exis beween he wo variables and Y. This simple Granger causaliy is valid only for a saionary series I (0). If ime-series included in he analysis are I (1) and coinegraed, he simple Granger causaliy es should no be used. In his case, Engle and Granger (1987) provide a more comprehensive procedure known as error-correcion model (ECM) for causaliy es. In he ECM model we incorporae informaion from he coinegraing regression in he form of residuals erm lagged once, and include he variables afer made saionary by differencing hem. The ECM is formulaed o es causaliy as follow: Y a 0 a1 i i a2iy j a3ect 1 b1i j b2iy j b3ect 1 b0 (6) Where is he error-correcion erm lagged one period. The Δ and Δ are differenced ime-series of and Y respecively. Ο and τ are whie noise error erms. The independen variable is said o cause he dependen variable if he error erm ) is significan, or he coefficiens of he lagged independen variable are joinly significan. Negaive and saisically significan values of he coefficiens of he error correcion erms indicae he exisence of long run causaliy. Causaliy es should be based on equaions (5) and (6) when series are found o be coinegraed. (5) 145

5 Inernaional Journal of Economics and Finance Vol. 4, No. 9; Empirical Resuls The univariae characerisics of he variables governmen revenue and governmen expendiure are esed. Boh Augmened Dickey-Fuller (ADF) and Phillip-Perron (PP) uni roo ess were applied, and resuls are shown in ables (2) and (3). The ADF and PP ess show ha boh variables are non-saionary in level. Only when he variables were differenced once, hey became saionary. Therefore, governmen revenue (LRGREV) and governmen expendiure (LRGEP) are inegraed of order one I(1). Table 2. Augmened Dickey-Fuller Uni Roo Tes Variable Consan Consan &Trend None LRGREV (1) (1) (1) ΔLRGREV *** (0) *** (0) *** (1) LRGEP (0) (0) (0) ΔLRGEP *** (0) ** (3) ***(0) LRGREV is log of real governmen revenue, LRGEP is log of real governmen expendiure, Δ is firs difference operaor. **, *** Significan a 5% and 1% respecively. Numbers beween parenhesis are lag lengh using Schwarz Info crierion auomaic maximum lag=7. Table 3. Phillips-Perron Uni Roo Tes Variable Consan Consan &Trend None LRGREV (0) (4) (1) ΔLRGREV *** (2) *** (9) *** (3) LRGEP (14) (4) (23) ΔLRGEP *** (12) ***(20) ***(3) LRGREV is log of real governmen revenue, LRGEP is log of real governmen expendiure, Δ is firs difference operaor. *** Significan a 1%. Numbers beween parenhesis are Band widh (Newey-Wes Auomaic) using Barle Karnel. Engle-Granger s coinegraion mehod is hen applied o boh variables since hey are inegraed of he same order. The resuls of ADF of residuals are shown in able (4). Table 4. Engle-Granger Coinegraion Tes and Long-run Relaionship Coinegraion Regression Adj. R 2 ADF LRGREV = LRGEP ** LRGEP = LRGREV ** LRGREV is log of real governmen revenue; LRGEP is log of real governmen expendiure. ** Significan a 5% MacKinnon (1996) p-values. Equaion (2) was esimaed wih governmen expendiure as he dependen variable, and governmen revenue as he independen variable. The residuals are esed for saionariy via ADF uni roo es. The ADF value was -4.66, which is less han MacKinnon criical value a 5% significance level. Also equaion (2) was esimaed wih governmen revenue as he dependen variable, and governmen expendiure as he independen variable. The residuals were esed for saionariy. The ADF value was -3.93, which is also less han MacKinnon criical value a 5% significance level. We conclude ha residuals are saionary and variables are coinegraed, and a long-run relaionship beween governmen revenue and expendiure exis. Since here is coinegraion beween he variables, we move o es he direcion of causaliy using error correcion model. Table (5) shows Granger causaliy resuls. The resuls show ha here is unidirecional causaliy runs from governmen revenue o governmen expendiure. This resul suppor revenue-spend 146

6 Inernaional Journal of Economics and Finance Vol. 4, No. 9; 2012 hypohesis. Alhough no shown here, he coefficien of he lagged error correcion erm for he governmen expendiure model is negaive and significan, which implies ha here is a long-run causal relaionship from governmen revenue o governmen expendiure in Qaar s economy. This resul is consisen wih oil-exporing counries. Table 5. Granger Causaliy Resuls Null Hypohesis (H 0 ) Wald Tes/χ 2 Prob. Conclusion LRGEP does no Granger can' rejec H 0 Cause LRGREV LRGREV does no Granger Cause LRGEP rejec H 0 LRGREV is log of real governmen revenue; LRGEP is log of real governmen expendiure. 5. Conclusion This research has aemped a providing evidence on he relaionship beween governmen revenue and expendiure in Qaar for he period from Boh uni roo ess ADF and Phillip-Perron found he variables o be inegraed of order one. Engle-Granger s approach of coinegraion found governmen revenue and expendiure o be coingraed, and hence, a long-run relaionship beween hem exiss. Granger causaliy es found unidirecional causaliy running from governmen revenue o governmen expendiure. This suppors revenue-spend hypohesis. The resul is differen han ha of Fasano and Wang (2002) findings on Qaar, bu similar o he causaliy resuls for Bahrain, The Unied Arab Emiraes and Oman in he same sudy. References Aisha, Z., & Khaoon, S. (2009). Governmen Expendiure and Tax Revenue Causaliy and Coinegraion: The Experience of Pakisan The Pakisan Developmen Review, 48(4), Al-Qudair, K. H. (2005). The Relaionship beween Governmen Expendiure and Revenue in he Kingdom of Saudi Arabia: Tesing for Coinegraion and Causaliy. Journal of King Abdulaziz Universiy, 19(1), Amoah, B., & Loloh, F. W. (2008). Causal Linkages beween Governmen Revenue and Spending: Evidence from Ghana. Working Paper, WP/BOG-2008/08. Bank of Ghana. Baainah, K. I. (Undaed). Governmen Revenues and Expendiures: Causaliy Tess for Jordan. Causaliy and Coinegraion: The Experience of Pakisan The Pakisan Developmen Review, 48(4), Eia, J. H., & Mbazima, D. (2008). The Causal Relaionship beween Governmen Revenue and Expendiure in Namibia. hp://mpra.ub.uni-munchen.de/9154/mpra Paper No. 9154, posed 16. June, 16:04 Elyasi, Y., & Rahimi, M. (2012). The Causaliy beween Governmen Revenue and Governmen Expendiure in Iran. Inernaional Journal of Economic Sciences and Applied Research, 5(1), Engle, R. F., & Granger, C. W. J. (1987). Coinegraion and error correcion: Represenaion, esimaion, and esing. Economerica, 2, Fasano, U., & Wang, Q. (2002). Tesing he Relaionship beween Governmen Spending and Revenue: evidence from GCC Counries. IMF Working Paper, WP/02/201, Inernaional Moneary Fund. Granger, C. W. J. (1969). Invesigaing he Causal Relaions by Economerics Models and Cross-Specral Mehods. Economerica, 37(3), Gujarai, D. N., & Porer, D. C. (2009). Basic Economerics. McGraw-Hill Inernaional Ediion. Harris, R. I. (1995). Using Coinegraion Analysis in Economeric Modeling. Pearson Educaion. Masenyese, R. F., & Moelle, S. I. (2012). Governmen Revenue Expendiure Nexus in Lesoho: he Decline in SACU Revenue. American Journal of Economics, 2(1), Minisry of Economics and Commerce. (1990). Economic Developmen ( ). New Prining Company. 147

7 Inernaional Journal of Economics and Finance Vol. 4, No. 9; 2012 Nyamongo, M. E., Sichei, M. M., & Schoeman, N. J. (2007). Governmen Revenue and Expendiure Nexus in Souh Africa. SAJEMS NS 10(2), Peanlar, S. K., & Sadeghi, S. (2012). Relaionship beween Governmen Spending and Revenue: Evidence from Oil Exporing Counries. Inernaional Journal of Economics and Managemen Engineering, 2(2), Pndyck, R., & Rubinfeld, D. (1998). Economeric Models and Economic Forecass, 4 h Ed. McGraw-Hill. Qaar Cenral Bank. Annual Repor. Various issues. Qaar Saisics Auhoriy. Ravinhirakumaran, K. (2011). The Relaionship beween Governmen Revenue and Expendiure in Sri Lanka. Proceedings of Second Inernaional Research Conference on Business and Informaion. Faculy of Commerce and Managemen Sudies, Universiy of Kelaniya, Sri Lanka. Revenue and Expendiure in Namibia. (2008). hp://mpra.ub.uni-munchen.de/9154/mpra Paper No. 9154, posed 16. June 2008/ 16:04. Sadiq, T. (2010). The Causaliy beween Revenues and Expendiure of he Federal and Provincial Governmens of Pakisan. The Pakisan Developmen Review, 49(4), Subhani, M. I., Hasan, S. A., Osman, A., & Rafiq, T. (2012). An Invesigaion of Granger Causaliy beween Tax Revenues and Governmen Expendiures. European Journal of Scienifics Research, 68(3), Tsen, W. H., & Kian-Ping, L. (2005). The Relaionship beween Governmen Revenue and Expendiure in Malaysia. Inernaional Journal of Managemen Sudies, 12(2),