Rokeya Sultana*, Murshida Khanam**, Khnd. Md. Mostafa Kamal ***

Transcription

1 Inernaional Journal of Mulidisciplinary Approach Mulivariae Co inegraion & Granger Causaliy under VECM o Idenify he Causal Effec and Effec Direcion of Major Macroeconomic Variables on Inflaion Dynamics in Bangladesh Rokeya Sulana*, Murshida Khanam**, Khnd. Md. Mosafa Kamal *** *Lecurer, Chiagong Governmen College **Associae Professor, Deparmen of Saisics, Biosaisics & Informaics, Universiy of Dhaka ***Associae Professor, Deparmen of Saisics, Biosaisics & Informaics, Universiy of Dhaka ABSTRACT This sudy invesigaes he causal relaionship beween inflaion and some major macroeconomic variables namely money supply, exchange rae, producion and ne domesic asse of Bangladesh over he year 2001 o 2011 using coinegraion and Granger causaliy under VECM. Geing daa saionary a firs difference he sudy finds ha inflaion is coinegraed wih money supply, producion and ne domesic asse, bu no wih exchange rae. Granger causaliy es shows bidirecional causaliy of money supply, producion, ne domesic asses wih inflaion; however, unidirecional causaliy exiss beween inflaion (CPI) rae and exchange rae. Therefore, in order o achieve ulimae susainable macroeconomic growh he moneary auhoriy of Bangladesh should keep focus on hese aforemenioned causal relaionships for argeed inflaion rae below 6%. Keywords: Inflaion, Economic growh, Coinegraion, Granger causaliy, Bidirecional causaliy. I. INTRODUCTION Mainaining low inflaion is an imporan macroeconomic policy aspec of robus and susainable economic growh. Alhough Bangladesh has achieved much progress in economic developmen in recen years, he rising rae of inflaion is also a serious concern for susainable economic growh. There are conroversial views ha, in developing counries, inflaion is he resul of exogenously generaed facors while ohers believe ha inflaion is primarily generaed due o he absence of sound inernal economic policies. However, here exiss consensus ha in developing economies, inflaion is deermined by muliple inerconneced facors. This sudy underakes he empirical esimaion of inflaion models using mulivariae single equaion co-inegraion equaion echnique wih reference o he Bangladesh economy o find he causal relaionship beween inflaion, money supply, exchange rae, ne domesic asse and producion index for monhly daa over he period 2001 hrough Alhough here is disagreemen among economiss on he roles of money supply, producion index, Exchange rae and ne domesic asse as well as heir inerrelaionship, hese variables are considered very imporan for macroeconomic performances and have been exensively invesigaed in boh, heoreical and empirical lieraure for boh developed and developing counries. Their causal relaionships have been an acive area of invesigaion in economics paricularly by Sims (1972) when he developed a es of causaliy and found unidirecional Page : 81

2 Inernaional Journal of Mulidisciplinary Approach causaliy from money o income for Unied Saes as claimed by he Moneariss. Aferwards, Lee and Li (1983) found bidirecional causaliy beween income and money and unidirecional from money o prices for Singapore. Joshi and Joshi (1985) suppor heir claim from he evidence of Indian economy. However, Khan and Siddiqui (1990) found of unidirecional causaliy from income o money and bidirecional beween money and prices in Pakisan. In he same line of research, Akash, e. al. (2011) claim ha here exis long erm relaionship among money supply, inflaion and indusrial producion for Pakisan. Abbas (1991) performed causaliy es beween money and income for Asian counries and found bidirecional causaliy in Pakisan, Malaysia and Thailand. Chimobi and Uche (2010) find ha money supply Granger causes boh producion and prices. However, Sharma, e.al. (2010) indicae ha oupu and prices do no Granger cause money supply reflecing exogeneiy of money supply. Therefore, for moneary policy sance i is imperaive o undersand he emporal dimensions of income, money and price causal relaionship. Very recenly, Kamal (2015) reveals a uniform direcional causaion beween he supply of money and price movemens for Bangladesh. The Auhors shows ha causal and reverse causal relaions beween money and produc and money and prices vary across frequencies while he causaliy running from money o oupu remains a shor-run phenomenon. The sudy also sipulaes a unidirecional causaliy beween money and prices, wih causaliy running from money supply o prices, which can be regarded as a piece of empirical evidence supporing he monearis claim. The sudy also demonsraes ha shor run causaliy from money supply o oupu, long run causaliy from money supply o prices, as well as lack of long run causaliy from money supply o oupu, all co-exis. This resul is also suppored by Nucu (2011) in he case of Romanian economy. Similarly, Das (2012) shows ha here exiss a bi-direcional causaliy beween per capia elecriciy consumpion and per capia GDP, per capia GDP and per capia income for Bangladesh. Likewise, bi-direcional causaliy beween budge defici and nominal effecive exchange raes exis for Indian economy, alhough he relaionships beween budge defici and GDP, Money supply & CPI are no significan for Indian economy (Srivyal and Venkaa, 2004). Comparable resul follows for Greece (Andreas and Anasasios,2011). Akinbobola (2012) finds a causal linkage beween inflaion, money supply and exchange rae in Nigeria. Nwosa and Oseni (2012) suppor his claim. Asari, e. al. (2011) show ha ineres rae moves posiively while inflaion rae goes negaively owards exchange rae volailiy in Malaysia. Gokal and Hanif (2004) have revealed ha he causaliy beween he wo variables ran one-way from GDP growh o inflaion. In spie of immense body of lieraure, here has been miniaure evidence for causal relaionship beween hese imporan macroeconomic variables ogeher, especially no evidence in he conex of Bangladesh. This paper seeks o redress his gap by examining he causal effec and direcion of causaliy beween he variables inflaion, money supply, exchange rae, producion & ne domesic asse in Bangladesh on monhly daa under Vecor Error Correcion Mechanism (VECM). II. DATA AND VARIABLES In his sudy, he monhly daa of five ime series variables- Consumer price index general (as proxy of Inflaion) using he base as , Broad money M2 (as proxy of money supply) Page : 82

3 Inernaional Journal of Mulidisciplinary Approach in 10 millions BDT, Quanum index for indusrial producion (all indusries) using he base as , Ne domesic asse in 10 millions (crore) BDT & Exchange rae (BDT per US dollar) from July 2001 o Sepember 2011 ha is 122 sample poins have been used. The variable have been named as cpigen, monsup2, qiip, ndmas & usdollar respecively. The daa have been colleced from Monhly saisical bullein of Bangladesh published by Bangladesh Bureau of Saisics. III. METHODOLOGY Graphical analysis and he correlogram analysis along wih Augmened Dickey Fuller uni roo es have been applied o deec he saionariy of he macroeconomic variables o avoid he problem of spurious regression and o mee he requiremens of saionariy properies of he ime series daa. The appropriae lag lengh has been deermined by consuling -es, AIC, Cauchy- Schwarz and Hannan-Quinn crierion. For Tesing for coinegraion Johansen coinegraion rank es has been used. Finally, he working Vecor Error Correcion Model akes he form ( indicaes he firs difference.): (1) (2) (3) (4) ;where is he error obained from he coinegraing regression and is he error in he ECM. Alhough, he exisence of coinegraion implies Granger causaliy, does no poin ou he direcion of he causaliy. Therefore, in order o deec he direcion, following Granger (1988) and Engle and Granger (1987), a VECM has been esimaed using he models 1 hrough 5. Before esing for Granger causaliy he esimaed resuls should be checked by diagnosic ess (serial correlaion (LM es), heeroscedasiciy, and sabiliy es). (5) Page : 83

4 Inernaional Journal of Mulidisciplinary Approach IV. RESULTS AND DISCUSSIONS Descripive saisics presened in able 1 shows ha primarily he daa is valid enough o employ he specified echniques for analysis. Table1: Descripive Saisics of he five macroeconomic variables. Variable Mean Sandard deviaion Minimum Maximum Observaion Consumer price index general Broad money M Quanum index for indusrial producion TK per us dollar Ne Domesic Asse The saionariy of he considered variables have been checked by Graphical analysis, correlogram es and Augmened Dickey Fuller uni roo es. The correlogram analysis is presened in figure 1 hrough figure Lag Barle's formula for MA(q) 95% confidence bands Figure1:ACF of consumer price index Lag Barle's formula for MA(q) 95% confidence bands Figure2: ACF of quanum index for indusrial producion Lag Barle's formula for MA(q) 95% confidence bands Lag Barle's formula for MA(q) 95% confidence bands Page : 84 Figure3: ACF of money supply Figure4: ACF of exchange rae.

5 Inernaional Journal of Mulidisciplinary Approach Page : Lag Barle's formula for MA(q) 95% confidence bands Figure5: ACF of ne domesic asse. The figures1 hrough5 of auocorrelaion funcion (ACF) of all he series show an exponenial rend. The ACF for he firs lag has is highes value and declines very slowly as lag increases. Therefore he auocorrelogram analysis shows all series are nonsaionary. The saionariy of he considered variables have been furher checked by Graphical analysis and have been presened in figures 6 hrough 15. Even number figures indicae he graphical presenaion of he variables a level while he odd numbers represens he firs difference values of he variables F igure6: General consumer price index (CPI) Figure7: Firs difference of consumer price index (CPI). Figure8: Money supply M2 Figure9: Firs diff. supply M2 of money

6 Inernaional Journal of Mulidisciplinary Approach qiip D.qiip 0 50 Fi gure10: Quanum index for indusrial producion Figure11: Firs difference of quanum index for indusrial producion Figure12: Exchange rae Figure13: Firs diff. of exchange rae Figure145: Ne domesic asse Figure15: Firs diff. of ne domesic asse From he graphs i is seen ha over he sudy period all he variables are showing an upward rend, suggesing perhaps ha hey has been changing wih he change of ime. This implies ha he series are nonsaionary. The firs differences of he series do no show any upward or downward rend along he ime. The plos of firs differences have a consan mean and variance (i.e. independen of ime variable) which indicaes ha he firs difference have made he daa saionary. Page : 86

7 Inernaional Journal of Mulidisciplinary Approach Table2: Lag selecion order crieria Lag LL LR df p FPE AIC HQIC SBIC e e * * e e+16* * e * e From he lag lengh selecion crieria (able 2) he appropriae lag lengh is 3. Furhermore, he Augmened Dickey Fuller (ADF) es has been employed and he es resuls are presened in able 3. Table 3: Resuls of Augmened Dickey Fuller (ADF) uni roo es Series Level Firs differences Conclusion CPI General Index *** I(1) Money Supply (M2) *** I(1) Producion index *** I(1) Exchange Rae *** I(1) Ne Domesic Asse ** I(1) Noe: Each ADF es uses a rend and no inercep, he lag lengh has been chosen based on minimum AIC. ***, **, * denoes a es saisic is saisically significance a 1%, %5, & 10% levels respecively. From he Augmened Dickey-Fuller es i is seen ha daa are nonsaionary in heir levels bu are saionary afer firs differencing indicaing all he variables are inegraed of order one. Aferwards, he Johansen coinegraion rank es resuls have been presened in able 4. From able 4 i has been found ha here are wo coinegraing vecors in he VAR sysem. Tha is here exis wo long erm or equilibrium relaionship among he sudy variables. Table4: Resuls of Johansen coinegraion es for mulivariae case Null value Eigen value Trace 5% criical 1% criical value saisic value r = r ** r * r Noe: No resricion is imposed in he coinegraion es. ** and * denoes es is saisically significance a 1% and 5% levels respecively. Opimum lag lengh for Johansen coinegraion es based on minimum AIC. Page : 87 Table 5 shows ha inflaion (CPI) has pair wise coinegraion wih money supply M2,

8 Inernaional Journal of Mulidisciplinary Approach producion index (QIIP) and ne domesic asse. Therefore, i can be concluded ha inflaion has long run relaionship wih hese variables. However, exchange rae has no long run relaion wih inflaion. If coinegraion exis among variables hen variable should be Granger cause a leas one direcion wihin coinegraed variables. Table5: Resuls of Johansen coinegraion es for bivariae case Pair wis coinegraion Hypohesis Eigen value Trace saisic Criical value Conclusion cpigen-monsup2 r = Coinegraion r * cpigen-qiip r = Coinegraion r * 3.76 cpigen-usdollar r = * NoCoinegraion r cpigen-ndams r = Coinegraion r * 3.76 Noe: * denoes a es saisic is saisically significance a 5% level of significance. Opimum lag lengh for Johansen coinegraion es based on minimum AIC. Since he series are coinegraed, following Hendry (1995), he Vecor Error Correcion Model (VECM) is esimaed o examine he shor run adjusmen o long run equilibrium, as well as he shor run dynamics among he seleced variables. There are wo coinegraing vecors in VAR sysem according o he mulivariae coinegraion es. So we include wo coinegraing errors in he sysem. Before esimaing he meodel presened in equaion 1 hrough 5, firsly he economeric crieria of auocorrelaion, heeroscrdasiciy, causaliy and he sabiliy of he fied VEC model should be esed. The presen sudy uses he LM es for deecion of residuals auocorrelaion, plo of fied value and residuals o deec he heeroscedasiciy among he residuals and VAR sabiliy checking. Table 6: The resuls of LM es Lag order df χ 2 es saisic p- value From able6, a 5% level of significance he null hypohesis of no auocorrelaion has been acceped up o he lag 3. Bu from he lag 4 he es is significan. Therefore, if lag lengh hree is aken he residuals will free from auocorrelaion. AIC also suggess lag lengh of hree. Afer geing he appropriae lag lengh he economeric crieria of he specified equaions furher should be invesigaed. For five differenced equaions in VECM he plo of prediced value and he squared residuals i is found ha here exis no heeroscedasiciy Page : 88

9 0.5 1 Inernaional Journal of Mulidisciplinary Approach in he residuals in any equaion considered in he VECM. In addiion, he Sabiliy checking for VECM has been repored in he figure 16 and able7. Roos of he companion marix In figure16, since he enire poins lie inside he uni circle so i could be easily concluded ha he model esimaed is sable wih correcly specified coinegraing vecors (Enders, W., 2011) Real The VECM specificaion imposes 3 uni moduli Figure16: Uni circle for he roos of coinegraing equaions. Table7: Eigen value sabiliy condiion Eigen value ±.3748 ±.3743 ± ±.0468 ± Modulus In he presen sudy 5 endogenous variables (cpigen, monsup2, qiip, usdollar, ndmas) have been used. Afer performing he sabiliy es 3 uni modules have been found. The moduli 3 of he remaining eigenvalues given in Table7 are sufficienly below 1 o sugges ha he number of coinegraing vecors have been correcly specified and ha hese are saionary, wihou having o worry abou he lack of disribuion heory for he moduli of Eigen values (Bernard Fingleon, Harry Garresen & Ron Marin, 2010). Afer invesigaing he required economeric crieria now Granger causaliy es can safely be used for deecing he causal direcion and he resuls are presened in able 8. Table8: Pair wise Granger Causaliy es resuls Null Hypohesis χ 2 es saisic monsup2 does no causes cpigen 11.32* cpigen does no causes monsup *** qiip does no causes cpigen 20.37*** cpigen does no causes qiip 27.55*** usdollar does no causes cpigen 5.03 cpigen does no causes usdollar 27.89*** ndmas does no causes cpigen 11.19* cpigen does no causes ndmas 27.74*** Noe: ***, **, * implies significance a 0.01, 0.05, and 0.10 levels respecively From able 8 i could be seen ha bidirecional causaliy exiss beween inflaion ( cpigen) and money supply ( monsup2), Producion index ( qiip) and inflaion ( cpigen), inflaion and Ne domesic asse ( ndmas), however, unidirecional causaliy from inflaion cpigen o exchange rae ( usdollar). Alhough he bidirecional causaliy beween inflaion and money supply is significan a 10%, he es of no causaliy from money supply o inflaion is highly significan a 1% level. Therefore, i could be said ha here is some level of dependency beween inflaion and money supply from cenral bank. Increasing Page : 89

10 Inernaional Journal of Mulidisciplinary Approach governmen borrowing from cenral bank is seen as highly inflaionary in he case of Bangladesh. However, here is a general consensus among economiss and policy-makers ha regulaing he growh of money sock is necessary o achieve a fairly sable price level and full employmen of an economy (Sims 1972). From he bidirecional causaliy among producion and inflaion i could be said ha producion and inflaion is srongly relaed o each oher. Accordingly if he producion level of a counry could be made sable hen inflaion could be also made sable. Ne domesic asse in Bangladesh is he aggregae of ne foreign asse, domesic credi governmen (ne) and domesic credi public secor (BBS). Increase in lending means he bank will ge increased reurn from he borrowers by adding he ineres rae wih he reurn. And increase in borrowing for a bank means he bank will have o pay ineres on he reurn. If here is no balance beween hese wo facors hen i resuls high inflaion. As a resul, balance beween lending and borrowing should be mainained o ge a sable inflaion. V. CONCLUSION This sudy invesigaes he causal relaionship of money supply, exchange rae, producion index and ne domesic asse o inflaion rae in Bangladesh over he ime period July 2001 o Sepember 2011 in monhly basis. Resuls show ha he daa are nonsaionary a heir level bu saionary a heir firs difference. Lagged value has been seleced as suggesed by AIC. Johansen coinegraion rank es shows ha here are wo coinegraing equaions i.e. long erm causal relaionship exiss. Based on Johansen coinegraion rank es Vecor Error Correcion Model (VECM) has been applied o find ou he shor erm dynamics of causaliy. Granger Causaliy under VECM is used o find he direcion of causaliy. The sudy is concerned wih he causaliy from he specified macroeconomic variables o he inflaion rae in Bangladesh. The pair wise Granger causaliy es has refleced he bidirecional causaliy beween inflaion (CPI) and money supply (M2), Producion index and inflaion, inflaion and Ne domesic asse, while, unidirecional causaliy from inflaion o exchange rae. The sudy resuls show ha money supply is relaed o inflaion; hence, he auhoriy has o conrol he excess supply of money. And for ha he governmen has o mainain a consisen balance beween consumpion and expendiure so ha he adjusmen in he budge defici could no increase he level of money supply. Moreover, o srenghen local currency Governmen has o increase Domesic Producion, above all, local producion mus be encouraged o boos domesic producion and income and reduce leakages of foreign exchange. Furhermore, a balance beween lending and borrowing should be mainained o ge a sable inflaion. REFERENCES i. Abbas, K. (1991). Causaliy Tes beween Money and Income: A Case Sudy of Seleced Developing Asian Counries ( ). The Pakisan Developmen Review, Vol. 30(4): ii. Akash, R.S.I. e al. (2011), Co inegraion and causaliy analysis of dynamic linkage beween economic forces and equiy marke: An empirical sudy of sock reurns (KSE) and macroeconomic variables (money supply, inflaion, ineres rae, exchange Page : 90

11 Inernaional Journal of Mulidisciplinary Approach iii. iv. rae, indusrial producion and reserves), African Journal of Business Managemen, vol. 5(27), Akinbobola, T.O.(2012), The dynamics of money supply, exchange rae and inflaion in Nigeria, Journal of Applied Finance & Banking, 2(4), Andreas, G.G. (2011), The macroeconomic effecs of budge deficis in Greece: A VAR-VECM approach, Inernaional Research Journal of Finance and Economics, Issue.79. v. Asari, F.F.A.H. e al. (2011), A vecor error correcion model (VECM) approach in explaining he relaionship beween ineres rae and inflaion owards exchange rae volailiy in Malaysia, World Applied Sciences Journal, 12, vi. vii. viii. ix. Bangladesh Bureau of Saisics (BBS), ( ). Saisical Pocke Book of Bangladesh Dhaka: Bangladesh Bureau of Saisics (BBS), Minisry of Planning, Governmen of he People s Republic of Bangladesh. Bangladesh Economic Updae (2011), Food prices and inflaion rajecory, Bangladesh Economic Updae, 2(1) Bernard Fingleon, Harry Garresen & Ron Marin (2010), Recessionary Shocks and Regional Employmen: Evidence on he Resilience of UK Regions, Deparmen of Economics, Universiy of Srahclyde, UK. Chimobi, O. P., & Uche, U. C. (2010). Money, Price and Oupu: A Causaliy Tes for Nigeria. American Journal of Scienific Research, 8, x. Das, J. (2012), Causaliy relaionship among elecriciy consumpion, energy use, income, expendiure and GDP in Bangladesh: A Vecor Error Correcion Model approach, Unpublished MS Thesis, Session: , Deparmen of Saisics, Biosaisics & Informaics, Universiy of Dhaka. xi. xii. xiii. xiv. xv. xvi. Enders, W. (2011), Applied Economeric Time Series, 2nd Ediion, John Wiley and Sons, Inc. Engel, R.F. and Granger, C.W.J. (1987), Coinegraion and error correcion: represenaion, esimaion and esing, Economerica, 55, Granger, C.W.J.(1988), Causaliy, coinegraion and conrol, Journal of Economic Dynamics and conrol, 12, Gokal, V. and Hanif, S.(2004). Relaionship beween inflaion and economic geowh, Economics Deparmen, Reserve Bank of Fiji, Working Paper-2004/04. Hendry, S. (1995). Long-Run Demand for M1. Bank of Canada Working Paper No Johansen, S. (1988). Saisical analysis of coinegraing vecors, Journal of Economic Dynamics and conrol, 12, xvii. Kamal, K. M. (2014). Invesigaing Long-run Relaionship beween Money, Income and Price for Bangladesh: Applicaion of Economerics and Cross Specra Mehods, Journal of Science Foundaion, 12(2). Page : 91

12 Inernaional Journal of Mulidisciplinary Approach xviii. xix. xx. xxi. xxii. xxiii. xxiv. xxv. xxvi. Khan, A., and Siddiqui A. (1990). Money, Prices and Economic Aciviy in Pakisan: A Tes of Causal Relaion. Pakisan Economic and Social Review, winer, Lee, S., and Li W. (1983), Money, Income, and Prices and heir Lead-lag Relaionship in Singapore. Singapore Economic Review, April, Main, A. (2011). Inflaion in Bangladesh: Driven by global phenomena, Bangladesh: BRAC Sock Brokerage Limied. Nucu, A.E. (2011). The relaionship beween exchange rae and key macroeconomic indicaors. Case sudy: Romania, The Romanian Economic Journal, XIV,l-41. Nowsa, P. I. and Oseni, I. O. (2012). Moneary policy, exchange rae and inflaion rae in Nigeria: A co-inegraion and mulivariae vecor error correcion model approach, Research Journal of Finance and Accouning, 3(3). Sharma A, Kumar A, Haekar N (2010). Causaliy beween Prices, Oupu and Money in India: An Empirical Invesigaion in he Frequency Domain, Cenre for Compuaional Social Sciences, Universiy of Mumbai, Discussion Paper No.3, Sims, C. (1972). Money, income and causaliy, The American Economic Review, 62(3), Srivyal, V. and Venkaa, S.S.(2004). Budge deficis and oher macroeconomic variables in India, Applied Economerics and Inernaional Developmen, 4(1). Tabas, H. M. e. al.(2012). The effec of he real effecive exchange rae flucuaions on macroeconomic indicaors (Gross Domesic Produc (GDP), Inflaion and Money Supply), Inerdisciplinary Journal of Conemporary Research in Business, 4 (6). Page : 92