Asian Economic and Financial Review IMPACT OF GOVERNMENT SPENDING ON PRIVATE CONSUMPTION USING ARDL APPROACH

Transcription

1 Asian Economic and Financial Review ISSN(e): /ISSN(p): journal homepage: hp:// IMPACT OF GOVERNMENT SPENDING ON PRIVATE CONSUMPTION USING ARDL APPROACH Khalid Khan --- Chen FEI Muhammad Abdul Kamal Badar Nadeem Ashraf 4,2,3,4 School of Economics, Huazhong Universiy of Science and Technology, (HUST), Wuhan, P.R. China ABSTRACT This sudy invesigaes he impac of governmen spending on privae consumpion a case sudy of China. We used Auoregressive Disribued Lag (ARDL) approach o esimae he long and shor run effecs of he model using annual daa from 985 o 23. The resuls of he sudy revealed ha governmen spending have posiive impac on privae consumpion. Therefore, governmen spending is a very good insrumen o boos economy and encourage aggregae demand in China during recession. The negaive and significan Error Correcion Term (ECT) suggess ha Chinese economy will adjus relaively quickly in response o an exernal shock o he overall developmen. 25 AESS Publicaions. All Righs Reserved. Keywords: Auoregressive disribued lag (ARDL), Governmen spending, Error correcion erm (ECT). Conribuion/ Originaliy To he bes of my knowledge, his sudy is firs of is kind o examine long and shor run effecs of governmen spending on privae consumpion in China. Secondly, his sudy uses more efficien and less resricive approach o co-inegraion (ARDL) which improves he findings on he subjec no only for China bu for overall lieraure on he relaionship beween governmen spending and privae consumpion.. INTRODUCTION Impac of governmen spending on aggregae economic aciviies is one of he conroversial issues in economics, differen school of houghs have differen views on he issue. Similarly, here is no clear evidence on he impac of governmen spending on privae consumpion. Since privae Corresponding auhor DOI:.8488/journal.aefr/25.5.2/ ISSN(e): /ISSN(p): AESS Publicaions. All Righs Reserved. 239

2 consumpion is one of he big componens of Gross Domesic Produc (GDP) and aggregae demand, i is very imporan o undersand he impac of governmen spending on privae consumpion o boos aggregae demand when economy face recession. Keynes s heory of Absolue income hypohesis posulaed ha households curren consumpion are response o curren disposable income, hus he increase in governmen spending leads o increase oupu and employmen, which furher influence households aggregae consumpion. Bu laer on, neoclassical consumpion heories believe ha households are concerns wih consumpion smoohing and response o fuure expeced income raher han curren income. Boh sandard RBC and Keynesian models are suppored ha governmen spending have muliplier effec and increase aggregae oupu, however, he debae of he effeciveness of governmen expendiure is based on he size of muliplier, and he size of muliplier based on he response of aggregae privae consumpion o governmen spending. Keynesian model predic posiive effec of governmen spending on privae consumpion while RBC model suppored negaive wealh effec. Blanchard and Peroi (22); Baxer and King (993) Generally he sources of financing governmen spending are: axes, selling bound and prining of money. The increase of governmen spending hrough raise in axes and selling bound are associaed wih crowding ou effecs, however, increase of governmen spending hrough prining of money associae wih problems of inflaion and increase of ineres rae. Therefore, i is sill a debaable issue ha weaher he increase of naional income is due o money supply or governmen spending or boh are responsible for i. Thus, in his sudy we ried o invesigae ha weaher privae consumpion and governmen spending is subsiue or complemenary o each oher. Lieraure on he relaionship beween governmen spending and privae consumpion presens mixed resuls. For insance, Bailey (97) provides he evidence for subsiuabiliy beween governmen spending and privae consumpion and suggesed ha governmen spending leads o crowding-ou effec. Similarly, Barro (98) assumed he uiliy funcion of ypical household in form of ( U C G, I) ) where he impac of governmen spending on privae consumpion depend upon he coefficien of governmen spending. Likewise, Kormendi (983) suppored he subsiuabiliy beween governmen spending and privae consumpion. On he conrary, some oher empirical sudies found suppor for posiive impac of governmen spending on privae consumpion. For insance, Faas and Mihov (2), Heppke-Falk e al. (26), found ha governmen spending and privae consumpion are posiively relaed o some exen. De Casro Fernández and Hernández De Cos (26), invesigaed he shor and long run effecs of governmen spending in case of Spain and found ha in he shor run he expansionary fiscal policy leads o low oupu and high inflaion while in long run i booss oupu. Tagkalakis (28) used he daa of OECD counries and esablished ha o simulae privae consumpion, fiscal policy is much beer in recession. Linnemann and Schaber (26) suggesed ha governmen spending is no waseful compleely and i improve he produciviy of privae secor. Ho (2) re-examined he crowding ou effec for 24 OECD counries by applying dynamic OLS (DOLS) and panel co-inegraion model. The resuls of he sudy rejec permanen income hypohesis and suppored he crowding ou 25 AESS Publicaions. All Righs Reserved. 24

3 effec. Kwan (26) empirically invesigaed he relaionship beween governmen spending and privae consumpion for Eas Asia counries using panel co-inegraing regression. The resuls of panel regression show ha on average governmen spending and privae consumpion are subsiue in Eas Asia, however, he cross-secion analysis revealed ha he value of elasiciy of subsiue is moderae for China, Hong Kong, Japan, and Korea, while high for Malaysia and Thailand and zero for Philippines. However, in case of Indonesia and Singapore i is complemenary. Shupei and Zhanwei (29) examined he fiscal expendiure and he privae consumpion from China opening up and saing reforms. The resuls of he sudy showed crowing ou effec. Chalk (2) invesigaed he effec of governmen social spending (i.e. spending on healh and educaion) on privae consumpion of China. The resuls of he sudy revealed ha a RMB increase in governmen spending on healh leads o increase urban households consumpion by 2 RMB, while educaion spending have no affec on households consumpion. Ramey (2a) discovers ha consumpion response posiively o defense spending shocks. Ramey (2b) reviewed he recen lieraure on governmen spending mulipliers based on ime series and panel daa. He concluded ha he range of governmen expendiure muliplier lies beween.8 and.5, however, he possible lowes and higher values of muliplier are;.5 and 2. Murphy (23) examined he recen conradicory evidence o new Keynesian and neoclassical models ha an increase in governmen spending boos aggregae privae consumpion. The sudy developed he imperfec informaion framework based on sudies of Lucas and Rober (972) and Lorenzoni (29). The model of he sudy argeed he owners of firms ha increase in governmen spending increase he permanen income of he firms owners relaive o fuure ax liabiliies, when he owners of he firms are no aware abou fuure ax liabiliies in such a case governmen expendiure leads o increase aggregae privae consumpion oherwise permanen income. Lieraure menioned above poins owards he imporance of he relaionship beween governmen spending and privae consumpion, however, his area of research is relaively undersudied for China. Therefore, his sudy ries o invesigae he effec of governmen spending on privae consumpion by using he ARDL approach o co-inegraion in case of China. Moreover, presenly, economic growh of China has showed a declining rend and i dropped o 7% agains he arge of 7.4%, which is an alarming siuaion for governmen auhoriies. There are several plausible reasons for relaively slow economic performance; however, among hem one noable obsacle is low proporion of aggregae privae consumpion in GDP, which is 39% in China. Therefore, i is very imporan for China o boos aggregae privae consumpion o keep he high face of economic growh. The remaining of he sudy is organized is as follow. Secion 2 discusses mehodology and daa. Secion 3 presens resuls and discussion while las secion of he paper offers conclusion. Inernaional Financial Saisics (IFS), AESS Publicaions. All Righs Reserved. 24

4 2. METHODOLOGY Asian Economic and Financial Review, 25, 5(2): To see he effec of governmen spending on aggregae privae consumpion, for his purpose a Linnemann and Schaber (26) and Ganelli and Tervala (29) developed a sandard heoreical framework very recenly. Therefore, o develop a heoreical model for our sudy, we follow he proposed models of Linnemann and Schaber (26) and Ganelli and Tervala (29). Thus, he uiliy funcion of he ypical household is: Where: (C)= Privae consumpion (G) = Governmen spending (L) = labor supplied U log( C G) L v V ( G) ( v v = The marginal disuiliy of work. V (G) = The separae impac of governmen spending However, marginal uiliy of privae consumpion is: U C C G The sign of he coefficien of equaion (2) depend upon he response of governmen spending o privae consumpion; i may be negaive or posiive depending upon he relaionship beween governmen spending and privae consumpion. Moreover, he budge consrain of ypical household is: Where: P= Aggregae price level W= Real wage rae = lump-sum ax D = dividends of firm. PC PwL P D( i) di (3) In equaion (3) we only have lump-sum ax and governmen spending is fully financed by ax revenue. Thus, he household opimizaion model becomes: (4) P( C G) (2) L v Pw (5) Subsiuing equaion (4) ino (5) and we obain equaion (6), which represens consumpionleisure rade off; 25 AESS Publicaions. All Righs Reserved. 242

5 L v w ( C G) (6) The producion funcion of ypical firm becomes: Y L G (7) The firs order condiion of profi maximizaion implies: MC w/( L G ) (8) Where MC is marginal cos, in equaion (8) he elasiciy of governmen spending is posiive, which shows ha governmen spending increase privae consumpion. 3. ESTIMATION METHOD Economerics lieraure provides various alernaives o examine he long run relaionships among variables. For example, Engel Rober and Granger (987) and FMOLS procedures of Phillips and Hansen (99) are popular for univariae analysis, while Johansen (988); and Johansen (995) are popular for mulivariae analysis. Auoregressive Disribued Lag Model (ARDL), are relaively new esimaion echnique inroduced by Pesaran and Shin (995, 998); Pesaran e al. (2), has some advanages over previous mehods,. For example, for small samples ARDL gives relaively more robus resuls. Moreover, ARDL approach is applicable irrespecive of he order of inegraion i.e. I()/I()/I()and I(). ARDL equaion akes he following form; ( L, p) y,..., n o k i ( L, q) x i i w (9) Where: 2 ( L, p) L 2L... pl 2 q ( L, q) o i L i2l... ipql i,2..., k Where: ( p y ) is he exogenous variable, ( ) is consan, (L) is lag operaor ( Ly y ), w is s vecor of deerminisic rend, inercep and exogenous variables. Thus he long run equaion of ARDL approach is: k y i xi w i () where : (, P) The long run elasiciies are represened by: 25 AESS Publicaions. All Righs Reserved. 243

6 ˆ i (, qˆ) ˆ i ˆ ˆ i... iqˆ i i,2,..., k () (, pˆ) ˆ ˆ... ˆ However, he long run coefficiens can be wrien as: 2 p ˆ( pˆ, qˆ ˆ, q2...ˆ q ) k ˆ ˆ... ˆ 2 (2) ˆ( ˆ, ˆ ˆ ˆ represens he ordinary leas square (OLS) esimaes of in Where: p q, q,..., q ) 2 k equaion (9). Thus, he ARDL equaion is: y c p c yy y yx x x. i z wx i (3) i H Where he hypoheses of co-inegraion are: H and : yy yx. x : yy yx. x. Keeping in view he objecive of our sudy, he mahemaical and economerics form of our sudy are given below respecively: c y g (4) y g c y g (5) y g i Where: ( c ) is privae consumpion, ( y ) is real GDP,( g ) is governmen expendiure,( ) random error. However, he ARDL srucure of equaion (5) is as follow: c c y g c y g m c y g i i j j k k i (6) i j ko n p Equaion (6) conains boh shor and long run informaion, o es he null and alernaive hypohesis of co-inegraion for equaion (6) is: H while alernaive is c y g H c y g.we used he bound es procedure o es null and alernaive hypohesis of co-inegraion. Annual daa on he variables of privae consumpion and GDP and governmen spending are exraced from Inernaional Financial Saisics (IFS) differen issues, all of he variables are used in real values. 25 AESS Publicaions. All Righs Reserved. 244

7 4. RESULTS AND DISCUSSION The concep of co-inegraion revolves around long-run rends embodied in he macro ime series. I is herefore necessary o es variables for uni-roo. We used Augmened Dickey Fuller (ADF) and Dicky- Fuller Generalized Leas Square (DF-GLS) ess o es for uni-roo. Table presens he resuls of he ADF and DF-GLS ess. The resuls of hese ess confirmed ha all variables had uni-roo. In oher words, all of he variables in he model are inegraed of I() i.e. privae consumpion, real GDP and governmen expendiure. Resuls of ADF and DF-GLS ess were consisen. However, ADF repored ha privae consumpion and real GDP is no saionary wihou rend a he firs difference while DF-GLS resuls revealed ha boh variables are saionary wih and wihou rend a firs difference. Table-. Saionariy Tess on Firs Difference Variables ADF DF-GLS c, c, c, c, Con -2.37(.542) (.8) -2.27(.2) -3.7(.) GDP (.48) -4.4 (.) -2.47(.48) -4.53(.) GE (.4) -4.34(.) -4.55(.7) (.) Noe: where: (c) inercep (consan) and () ime rend, P-values are in parenhesis. Table-2. Co-inegraion Tes F- Sa: Tabulaed LB a 95% UB a 95% LU a 9% UB a 9% Noe: Lower Bound (LB) and Upper Bound (UB) Tables 2 presens he resuls of ARDL bound approach o co-inegraion. The resuls of he bound es revealed ha he F- saisic abulaed value is 3.45 which is grea han UB a 95% and 9% respecively. This shows ha here is a long run relaionship among variables in he model. Afer he exisence of long run relaionship among variables, now we move o nex sep, o esimae he long and shor run parameers of he variables hrough ARDL approach. Table-3. Long Run Resuls of ARDL ARDL (,,) SELECTED BASED AIC Variables Coefficiens P- Values GDP GE Table 3 repored long run resuls of ARDL approach. In he long run, boh coefficiens of GDP and governmen spending are saisically significan and having correc signs. The coefficien of GDP is.444 which shows ha a one uni increases in GDP will leads o amplify he aggregae privae consumpion by 44.4%. Similarly, a one uni change in governmen spending will augmen privae consumpion by 58.2% on average. The shor run resuls of ARDL are repored in able 4, 25 AESS Publicaions. All Righs Reserved. 245

8 where he coefficien of GDP is saisically significan while he coefficien of governmen spending is no. The Error Correcion Term (ECT) shows he speed of adjusmen from disequilibrium o equilibrium in nex period. The value of he ECT is negaive and saisically significan which indicaes ha privae consumpion will approach o equilibrium wih speed of from disequilibrium o equilibrium. Table-4. Shor Run Resul of ARDL ARDL (,,) SELECTED BASED AIC Variables Coefficiens P- Values GDP.2.47 GE ECT (-) R-square.56 F-saisic 9.32 DW.85 DIAGNOSTIC TEST OF THE MODEL LM es for Serial Correlaion, CHSQ().828(.773) Heeroscedasiciy Tes of Residuals, CHSQ().32(.734) JB Normaliy Tes, CHSQ().386(.42) Ramsey RESET Tes CHSQ ().24(.522) We also applied sandard diagnosic es and sabiliy es for goodness of fi of he model. The diagnosic ess used in his sudy are o es for he problems of non-normaliy of residuals, heeroscedasicy, serial correlaion and funcional form of he model. Resuls of diagnosic ess sugges ha our resuls are free from above menioned economeric issues. Moreover, for sabiliy of he model we employed CUSUM and CUCUMsq, i can be seen from Appendix, ha he plo of CUSUM and CUSUMsq falls wihin criical bound of 5% and confirmed he long run associaion among variables as well as sabiliy of he coefficiens in he model. 5. CONCLUSION The resuls of sudy revealed ha governmen spending has posiive relaionship wih privae consumpion in China. Moreover, governmen spending has almos he same impac on privae consumpion in boh long and shor run, bu he coefficien of governmen spending is saisically insignifican in he shor run. Furhermore, he value of ECT is quie high and saisically significan wih negaive signs, which shows he speed of adjusmen from disequilibrium o equilibrium. Presenly, he GDP growh of China become lile slow, herefore, in case of China governmen spending is a very good insrumen o boos economy and encourage aggregae demand o keep he high face of economic growh. REFERENCES Bailey, M.J., 97. Naional income and he price level. New York: McGraw-Hill. 25 AESS Publicaions. All Righs Reserved. 246

9 Barro, R.J., 98. Oupu effecs of governmen purchases. Journal of Poliical Economy, 89(6): Baxer, M. and R.G. King, 993. Fiscal policy in general equilibrium. American Economic Review, 83(3): Blanchard, O.J. and R. Peroi, 22. An empirical characerizaion of he dynamics effecs of changes in governmen spending and axes on oupu. Quarerly Journal of Economics, 7(4): Chalk, N., 2. China: Does governmen healh and educaion spending boos consumpion? Inernaional Moneary Fund, Working Paper, No. WP//6. De Casro Fernández, F. and P. Hernández De Cos, 26. The economic effecs of exogenous fiscal shocks in Spain: A SVAR approach. Working Paper No. 647.European Cenral Bank. Engel Rober, F. and C.W.J. Granger, 987. Co-inegraion and error correcion: Represenaion, esimaion, and esing. Economerica, 55(2): Faas, A. and I. Mihov, 2. The effecs of fiscal policy on consumpion and employmen. Theory and evidence. Mimeo: INSEAD. Ganelli, G. and J. Tervala, 29. Can governmen spending increase privae consumpion? The role of complemenary. Economics Leers, 3(): 5 7. Heppke-Falk, K.H., J. Tenhofen and G.B. Wolff, 26. The macroeconomic effecs of exogenous fiscal policy shocks in Germany: A disaggregaed SVAR analysis. Deusche Bundesbank, Discussion Paper No. 4. Ho, T.-W., 2. The governmen spending and privae consumpion: A panel coinegraion analysis. Inernaional Review of Economics & Finance, (): Johansen, S., 988. Saisical analysis of co-inegraed vecors. Journal of Economic Dynamics & Conrol, 2(2): Johansen, S., 995. A saisical analysis of I(2) variables. Economeric Theory, (): Kormendi, R.C., 983. Governmen deb, governmen spending and privae secor behavior. American Economic Review, 73(5): Kwan, Y.K., 26. The direc subsiuion beween governmen and privae consumpion, in Eas Asia NBER Working Paper No. 243, JEL No. E6, H5. Linnemann, L. and A.A. Schaber, 26. Producive governmen expendiure in moneary business cycle models. Scoish Journal of Poliical Economy, 53(): Lorenzoni, G., 29. A heory of demand shocks. American Economic Review, 99(5): Lucas, J. and E. Rober, 972. Expecaions and he neuraliy of money. Journal of Economic Theory, 4(2): Murphy, P.D., 23. How does governmen spending simulae consumpion? Federal Reserve Bank of Dallas Globalizaion and Moneary Policy Insiue, Working Paper No. 57. Pesaran and Shin, 995, 998. An auoregressive disribued lag modeling approach o coinegraion analysis. DAE Working Papers, No Pesaran, Shin and Smih, 2. Bounds esing approaches o he analysis of level relaionships. Journal of Applied Economerics, 6(3): AESS Publicaions. All Righs Reserved. 247

10 Phillips, P.C.B. and B.E. Hansen, 99. Saisical inference in insrumenal variable regression wih i () processes. Rev. Econ. Sudies, 57(): Ramey, V.A., 2a. Idenifying governmen spending shocks: I s all in he iming. Quarerly Journal of Economics, 26(): -5. Ramey, V.A., 2b. Can governmen purchases simulae he economy. Journal of Economic Lieraure, 49(3): Shupei, L. and B. Zhanwei, 29. The dynamic analysis of impacs of governmen spending on privae consumpion since he reform and opening-up in China: An invesigaion based on TVP model. Journal of Finance and Economics, 29(9): Tagkalakis, A., 28. The effecs of fiscal policy on consumpion in recessions and expansions. Journal of Public Economics, 92(5-6): APPENDIX-. Figure-. Figure-2. Views and opinions expressed in his aricle are he views and opinions of he auhors, Asian Economic and Financial Review shall no be responsible or answerable for any loss, damage or liabiliy ec. caused in relaion o/arising ou of he use of he conen. 25 AESS Publicaions. All Righs Reserved. 248