Industrial Output Response to Inflation and Exchange Rate in Nigeria: An Empirical Analysis

Transcription

1 Indusrial Oupu Response o Inflaion and Exchange Rae in Nigeria: An Empirical Analysis Yakubu Musa * and Jibrin A. Sanusi 2 Saisics Uni, Deparmen of Mahemaics, Usmanu Danfodiyo Universiy Sokoo, Nigeria 2 Deparmen of Mahemaics and Saisics, Nuhu Bamalli Polyechnic, Zaria, Nigeria * Corresponding auhor ykmm2000@yahoo.com Absrac This sudy invesigaes he response of aggregae indusrial oupu o relaive change in prices and exchange rae in Nigeria using daa from A vecor error correcion (VEC) model was employed and he dynamic correlaions of he variables have been capured by he analyses of impulse response and variance decomposiion. The response of indusrial oupu o he shock o exchange rae was significanly posiive more specifically in he iniial years, while shock o prices changes, he indusrial oupu responds negaively alhough wih small magniude a he beginning. From variance decomposiion; he sudy shows ha alhough he main source of variance in oupu are own shocks, innovaion in he exchange rae accoun for a higher proporion in he variaion of indusrial oupu han ha of prices. The sudy concludes ha inflaion and exchange rae has he poenials of causing significan changes in indusrial oupu in Nigeria. This sudy herefore suggess ha more policy aenion should be given o proper managemen of he exchange rae and inflaion. Keywords: Indusrial oupu, exchange rae, inflaion, VEC model.0 Inroducion The pah o economic recovery and growh may require increasing producion inpus - land, labour, capial and echnology and or increasing heir produciviy. Increasing produciviy should be he focus because many oher counries ha have found hemselves in he same predicamens have resolved hem hrough produciviy enhancemen schemes. For insance, Japan from he end of he World War II and he Unied Saes of America from he 970s have made high produciviy he cenre poin of heir economic planning and he resuls have been resounding. Also, middle income counries like Hong Kong, Souh Korea, Singapore, he Philippines, India, Mexico and Brazil have embraced boosing produciviy schemes as an inegral par of heir naional planning and oday hey have made significan in-roads ino he world indusrial markes. Given he imporance of high produciviy in boosing economic growh and he sandards of living of he people, is measuremen canno bu be of imporance o boh policy makers and researchers. Produciviy measuremen can be used o evaluae he efficiency of an economy in relaion o ohers. I will also be useful in asceraining he relaive efficiency of firms, sub-secors and secors. A knowledge of he relaive efficiency of indusries and heir profiabiliy could aid governmen in planning is programmes and policies, especially in deciding on which indusries should be accorded prioriy. In addiion, i will help he governmen in deciding he wage level as he inpu and oupu of labour will be well quanified. A he micro level, produciviy measuremen will, among ohers, aid producion planning and sales, especially in checking cos, Including wages, subsiuion of facors of producion, reducion of wases, ec. High produciviy in he Nigerian manufacuring indusry is a necessary condiion for he secors recovery, achieving compeiiveness, boosing he GDP and uplifing he sandards of living of he people. In view of he foregoing, he objecive of his paper is o invesigae he response of aggregae indusrial oupu o relaive change in prices and exchange rae in Nigeria using Vecor error correcion (VEC) model Approach. 2.0 Lieraure review The indusrial producion, exchange rae and inflaion have been sudied exensively in many lieraures. Zhang (2008) applies Hamilon s (200) approach o invesigae he relaionship beween oil price shock and Japanese indusrial producion growh using quarerly daa from 957:I o 2006:IV and finds ha he oil price changes and macroeconomic aciviy in Japan appear o be affeced by a non-linear relaionship. Adenikinju and Olofin (2000) focus on he role of economic policy in he growh performance of he manufacuring secors in African counries. They uilize panel daa for seveneen African counries over he period 976 o 993. Their economeric evidence indicaes ha governmen policies aimed a encouraging foreign direc invesmen, enhancing he exernal compeiiveness of he economy, and mainaining macroeconomic balance have significan effecs on manufacuring growh performance in Africa. Folorunsho and Abiola (2000) examine he long run deerminans of inflaion in Nigeria beween 970 and 998, using he economeric mehods of coinegraion and error correcion mechanism. They find ha inflaion 74

2 in Nigeria could be caused by he level of income, money supply, and public secor balance. The resuls also indicae ha in he long run, exchange rae, money supply, income and fiscal balance deermine he inflaion spiral in Nigeria. The sudy, herefore, concludes ha a reducion in fiscal deficis, an increase in domesic producion and a sable exchange rae should be pursued as means of conrolling inflaion in Nigeria. Morley (992) analyzed he effec of real exchange raes on oupu for weny eigh developing counries ha have devalued heir currencies using a regression framework. Afer he inroducion of conrols for facors ha could simulaneously induce devaluaion and reduce oupu including erms of rade, impor growh, he money supply, and he fiscal balance, he discovered ha depreciaion of he level of he real exchange rae reduced he oupu. Kamin and Klau (998) using an error correcion echnique esimaed a regression equaion linking he oupu o he real exchange rae for a group of weny seven counries. They did no find ha devaluaions were conracionary in he long erm. Addiionally, hrough he conrol of he sources of spurious correlaion, reverse causaliy appeared o alernae he measured conracionary effec of devaluaion in he shor erm alhough he effec persised even afer he inroducion of conrols. Apar from he findings from simulaion and regression analyses, resuls from VAR models, hough no focused mainly on he effecs of he exchange rae on he oupu per se, are equally informaive. Ndung u (993) esimaed a six-variable VAR money supply, domesic price level, exchange rae index, foreign price index, real oupu, and he rae of ineres in an aemp o explain he inflaion movemen in Kenya. He observed ha he rae of inflaion and exchange rae explained each oher. A similar conclusion was also reached in he exended version of his sudy (Ndung u 997). Folawemo and Osinubi (2006) examined he efficacy of moneary policy in conrolling inflaion rae and exchange rae insabiliy. The analysis performed was based on a raional expecaion framework ha incorporaes he fiscal role of exchange rae. Using quarerly daa spanning over 980: o 2000:4 and applying imes series es on he daa used, he sudy showed ha he effecs of moneary policy a influencing he finance of governmen fiscal defici hrough he deerminaion of he inflaion-ax rae affecs boh he rae of inflaion and exchange rae, hereby causing volailiy in heir raes. The sudy revealed ha inflaion affecs volailiy in is own rae, as well as he rae of real exchange. Rogers and Wang (995) obained similar resuls for Mexico. In a five-variable VAR model oupu, governmen spending, inflaion, he real exchange rae, and money growh mos variaions in he Mexican oupu resuled from own shocks. They however noed ha exchange rae depreciaions led o a decline in oupu. Adoping he same mehodology, hough wih slighly differen variables. Copelman and Wermer (996) repored ha posiive shocks o he rae of exchange rae depreciaion, significanly reduced credi availabiliy, wih a negaive impac on he oupu. Surprisingly, hey found ha shocks o he level of he real exchange rae had no effecs on he oupu, indicaing ha he conracionary effecs of devaluaion are more associaed wih he rae of change of he nominal exchange rae han wih he level of he change of he real exchange rae. They equally found ha own shocks o real credi did no affec he oupu, implying ha depreciaion depressed he oupu hrough mechanisms oher han he reducion of credi availabiliy. I is imporan o menion he work of Odusola and Akinlo (200) who examined he linkage among exchange rae, inflaion and oupu in Nigeria. A srucural VAR model was employed which capured he ineracions beween exchange rae and oupu. Evidence from he conemporaneous models showed a conracionary impac of he parallel exchange rae on oupu only in he shor erm. Prices, parallel exchange rae and lending rae were found o be imporan sources of perurbaions in he official exchange rae. In addiion, oupu and parallel exchange rae were significan deerminans of inflaion dynamics in Nigeria. The auhors concluded by suggesing more concered effors by he Cenral Bank owards aming he parallel exchange rae behavior and formulaing moneary policies ha enhance income growh. Largely he findings were informaive. 3.0 Daa and Mehodology 3. Daa Source The daa se for his paper consiss of annual ime series from The variables under consideraion are indusrial oupu (GDP), Inflaion measured as consumer price index (INF) and exchange rae (EXG). The daa are obained from cenral bank of Nigeria Saisical bullein 200 and Model Specificaion Sims s (980) seminal work inroduces unresriced vecor auoregressions (VARs) ha allows feedback and dynamic inerrelaionship across all he variables in he sysem and appears o be highly compeiive wih he large-scale macro-economeric models in forecasing and policy analysis. 75

3 To provide an empirical insigh ino he response of aggregae indusrial oupu o relaive change in prices and exchange rae in Nigeria, we esimaes hree-variable VAR model by using GDP, INF, and EXG. The following model was formulaed: GDP = a0+ ainf + a2exg + u () Where a 0 is he consan and a, a 2 are coefficien o be esimaed and u is an error erm. GDP is he Indusrial oupu, INF is he consumer price index( yearly change in prices), and EXG is he Real official exchange rae. The General basic model of VAR (p) has he following form y = m+ y D + A y Ap y- p+ u (2) where y is he se of K ime series variables y = ( y,..., y K ), Ai ' s are (K K) coefficien marices, m is vecor of deerminisic erms, D is a vecor of nonsochasic variables such as economic inervenion and seasonal dummies and = (,..., K ) is an unobservable error erm. Alhough he model (2) is general enough o accommodae variables wih sochasic rends, i is no he mos suiable ype of model if ineres ceners on he coinegraion relaions. The VECM form D y =P y +GD y G D y + m+ y D + u (3) Where P= ab u u u - - p- - p+ In he VECM model, aenion focuses on he (n r ) marix of coinegraing vecors b, which quanify he long-run relaionships beween variables in he sysem, and he (n r) marix of error-correcion adjusmen coefficiens a, which load deviaions from he equilibrium (i.e. P y - ) o D y for correcion. The G i coefficiens in (3) esimae he shor-run effecs of shocks on D,and herefore allow he shor-run and longrun responses o differ. The erm P y - is he only one ha includes I() variables. Hence, P y - mus also be I(0). Thus, i conains he coinegraing relaions. The G ( j =,..., p ) are ofen referred o as he shor-run or shor-erm js parameers, and P y - is someimes called he long-run or long-erm par. 3.3 Uni Roo Tes Since we are using imes series daa ses for he analysis, i is imporan ha we firs es he daa ses for saionariy properies. Hence, o examine he saionariy properies of he daa ses, we use a variey of unis roo ess. The moivaion behind he assormen of ess is o obain reliable and consisen resuls. Firs, he Augmened Dickey Fuller (ADF) ess and Phillips-Perron (PP) ess are used o check wheher each daa series is inegraed and has a uni roo. The ADF es is based on he following regressions. Where D y = a + a y + ad y + el 0 - i i i= n å D y = a + a y + ad y + d + e 0 - i i= n å y is a ime series, is a linear ime rend, D is he firs difference operaor, 0 y (5) (6) a is a consan, n is he opimum number of lags on he dependen variable and e is he random error erm. The difference beween equaion (5) and (6) is ha he firs equaion includes jus drif. However, he second equaion includes boh drif and linear ime rend. This sudy also employs he Philip-Perron es (988). The regression equaion for he PP es is given by D Y = a+ by + e - (7) 3.4 VAR Coinegraion Tes The resuls of he inegraion ess are hen pursued by Co-inegraion ess. The Exisence of long-run equilibrium (saionary) relaionships among economic variables is referred o in he lieraure as coinegraion. The Johansen procedure will be employed o examine he quesion of coinegraion and provide no only an 76

4 esimaion mehodology bu also explici procedures for esing for he number of coinegraing vecors as well as for resricions suggesed by economic heory in a mulivariae seing. Engle and Granger (987) poined ou ha a linear combinaion of wo or more non-saionary variables may be saionary. If such a saionary combinaion exiss, hen he non-saionary ime series are said o be co-inegraed. The VAR based coinegraion es using he mehodology developed in Johansen (99, 995). Johansen s mehodology akes is saring poin in he vecor auo regression (VAR) of order p given by y = m+d y D p y- p+ e (8) Where y is an nx vecor of variables ha are inegraed of order commonly denoed (I()) and e is an nx vecor of innovaions. This VAR can be rewrien as D y = m+p y- +GD y G p-d y- p+ + e (9) Where P=-( In- A Ap ) and G i =- ( Ai Ap ) To deermine he number of co-inegraion vecors, Johansen (988) and Johansen and Juselius (990) suggesed wo saisic es, he firs one is he race es (λ race). I ess he null hypohesis ha he number of disinc coinegraing vecor is less han or equal o n agains a general unresriced alernaives n = r. he es calculaed as follows: l race n å ( r) =-T ln(- ˆ l ) r+ i Where T is he number of usable observaions, and he ˆi l are he esimaed eigenvalue from he marix. The Second saisical es is he maximum eigenvalue es (λ max) ha is calculaed according o he following formula l ( r, r+ ) =-T ln(- ˆ lr ) max + () The es concerns a es of he null hypohesis ha here is r of co-inegraing vecors agains he alernaive ha r + co-inegraing vecor. 4.0 Empirical Analysis Sofware for he analysis: Eviews 7 4. Uni Roo Tess Before using he daa in he esimaion of VAR/VECM, we need o know ime series properies of all he variables. Accordingly, a series of uni roo es, such as Augmened Dickey-Fuller (ADF, 98) and Phillips- Perron (PP, 988) ess are used o deermine he order of inegraion for each series. The ADF uni roo ess used Akaike informaion crierion for lag order selecion and PP uni roo ess lag lengh are decided based on Akaike s informaion crierion and AR specral GLS derended specra. The null hypohesis of non-saionary is rejeced if he -saisic is less han he criical -value. Using The ADF ess and PP ess (Table ), all oher variables possess uni roos a heir levels since each repored -saisics is no smaller han heir respecive 5% criical values. Hence, he variables are saionary afer firs differencing (Table 2). Table : ADF and PP Tes a Levels Variables Consan Consan & Trend ADF Tes PP Tes ADF Tes PP Tes -Saisic Prob. -Saisic Prob. -Saisic Prob. -Saisic Prob. GDP INF EXG % criical value % criical value Table 2: ADF and PP Tes a firs difference 5% criical value (0) 5% criical value

5 Variables Consan Consan & Trend ADF Tes PP Tes ADF Tes PP Tes -Saisic Prob. -Saisic Prob. -Saisic Prob. -Saisic Prob. GDP INF EXG % criical value % criical value % criical value % criical value Coinegraion es The uni roo ess confirmed ha he series are inegraed hus saisfying he iniial assumpion for co-inegraion analysis. Table 3: VAR Lag Order Selecion Crieria Lag LogL LR FPE AIC SC HQ NA 6.5e * 4.94e+* * * * e e * indicaes lag order seleced by he crierion, LR: sequenial modified LR es saisic (each es a 5% level) FPE: Final predicion error, AIC: Akaike informaion crierion SC: Schwarz informaion crierion HQ: Hannan-Quinn informaion crierion Lag lengh were seleced o be one (see Table 3) using informaion crieria and saisfied he mahemaical sabiliy condiion. Table 4: Coinegraion es (Linear deerminisic rend) Unresriced Coinegraion Rank Tes (Trace) Hypohesized Trace 0.05 No. of CE(s) Eigenvalue Saisic Criical Value Prob.** None * A mos A mos Trace es indicaes coinegraing eqn(s) a he 0.05 level * denoes rejecion of he hypohesis a he 0.05 level **MacKinnon-Haug-Michelis (999) p-values Table 5: Coinegraion es (Linear deerminisic rend) Unresriced Coinegraion Rank Tes (Maximum Eigenvalue) Hypohesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Saisic Criical Value Prob.** None * A mos A mos Max-eigenvalue es indicaes coinegraing eqn(s) a he 0.05 level * denoes rejecion of he hypohesis a he 0.05 level **MacKinnon-Haug-Michelis (999) p-values The resuls of he maximal eigenvalue and race es saisics for he wo models were presened in Tables 4 and 5. The p-value a 5% level of significan indicaes ha he hypohesis of no coinegraion among he variables can be rejeced for Nigeria. Boh Trace es and Maximum Eigenvalue es found one coinegraing relaionships 78

6 a 5% significan level beween he Indusrial oupu, change in prices and exchange rae variables in Nigeria. Since he variables are coinegraed, hus, i is concluded ha here exiss a long-run equilibrium relaionship among he variables under sudy. 4.3 Impulse Response Funcions for VEC Model Impulse Response Funcions (IRFs) are one of he useful ools of he VAR/VECM approach for examining he ineracion beween he variables in his sudy. They reflec how individual variables respond o shocks from oher variables in he sysem. When graphically presened, he IRFs give a visual represenaion of he behaviour of variables in response o shocks. The responses are for a paricular variable o a one-ime shock in each of he variables in he sysem. As noed by Odusola and Akinlo (200), he inerpreaion of he impulse response funcions akes ino consideraion he firs differencing of he variables as well as he vecor error correcion esimaes. The response forecas period is en years o enable us capure boh he long erm and shor erm responses. 2,000 Response o Cholesky One S.D. Innovaions Response of GDP o INF,500, , Response of GDP o EXG 2,000,500, , Figure : Shocks o Inflaion and Exchange rae Figure, shows he impulse response funcion of Inflaion and exchange rae shocks. The response of indusrial oupu o a sandard deviaion shock from change in prices is negaive bu quie minimal more specifically a iniial periods bu is deepens in he second year before i s sabilize o a new negaive level which coninuous even afer he enh year s period. The response of Indusrial oupu (GDP) o exchange rae innovaions is posiive more specifically a second year bu reduces subsequen o anoher posiive level which coninuous even afer he enh year s period. 4.4 VEC Model Forecas Error Variance Decomposiion The resuls of variance decomposiion a VEC Model reveal he forecas error in each variable ha can be aribued o innovaions in oher variables over en year periods. In VEC Model, he forecas error variances of 79

7 all he variables in he sysem are largely due o heir own innovaions, alhough over ime he innovaions of oher variables show a endency o increase gradually. Forecas error Variance decomposiions are presened in he Table 6, which help idenify he main channels of influence for individual variables. The numbers under each variable represen he percenage of variance of he variable analyzed ha was aribuable o he paricular variable over a 0 year period. Variance of indusrial oupu (GDP) Table 6: Variance Decomposiion of GDP Variance Decomposiion of GDP: Period S.E. GDP INF EXG According o Table 6, indusrial oupu accouned for is conemporary variance from is own innovaions wih abou 00 per cen in he firs year, alhough i shows gradual decline from abou 00% in he firs year o abou 98 % in he long erm. There was some variaion caused by inflaion and exchange rae. However, in laer periods, exchange rae increasingly conribued o variaions of indusrial oupu wih more han %. Alhough, inflaion variable make very lile conribuion o he variance of indusrial oupu bu he impac of exchange rae is more. 5.0 Conclusion This sudy invesigaes he response of aggregae indusrial oupu o relaive change in prices and exchange rae in Nigeria using Coinegraed VAR mehodology. We used Johansen coinegraion es o see if here is presen of long-run relaion beween he variables under sudy. The resuls of which provide evidence of long-run equilibrium relaionship among he variables. Since, here is evidence of long-run relaionship among he variables, a vecor error correcion (VEC) model was employed and he dynamic correlaions of he variables have been capured by he analyses of impulse response and variance decomposiion. For Impulse response funcion: he response of indusrial oupu o a sandard deviaion shock from change in prices is negaive bu quie minimal more specifically a iniial periods while he response of Indusrial oupu (GDP) o exchange rae innovaions is posiive more specifically a second year bu reduces subsequen o anoher posiive level which coninuous even afer he enh year s period. From variance decomposiion; he sudy shows ha alhough he main source of variance in oupu are own shocks, innovaion in he exchange rae accoun for a higher proporion in he variaion of indusrial oupu han ha of prices. The sudy concludes ha inflaion and exchange rae has he poenials of causing significan changes in indusrial oupu in Nigeria. This sudy herefore suggess ha more policy aenion should be given o proper managemen of he exchange rae and inflaion in Nigeria. Reference Adenikinju, A and Olofin, S. O. (2000). Economic Policy and Manufacuring Secor Growh Performance in Africa. The Nigeria Journal of Economic and Social Sudies, 42(): -4. Cenral Bank of Nigeria, Saisical Bullein, 200. Cenral Bank of Nigeria, Saisical Bullein, 20 Copelman, Marina, and Alejandro M. Wermer. (996). The Moneary Transmission Mechanism in Mexico. Working Paper, no. 25. Washingon, D.C.: Federal Reserve Board. Dickey, D. A., & Fuller, W. A. (98). Likelihood raio saisics for auoregressive ime series wih a uni roo. Economerica, 49(4), Engle, R.F and C.W.F Granger (987). Coinegraion and Error Correcion: Represenaion and Tesing. Economerica 55:

8 Folorunsho, B. A and Abiola, A. G. (2000). The Long run Deerminans of inflaion in Nigeria ( ). The Nigerian Journal of Economic and Social Sudies, 42(): Folawewo, A. O. and Osinubi, T. S. (2006). Moneary policy and macroeconomic insabiliy in Nigeria : A raional expecaion approach. Journal of Social Sciences, 2(2), Hamilon, J., (200). A parameric approach o flexible nonlinear inference. Economerica 69, Johansen, S. (99). Esimaion and hypohesis esing of coinegraion vecors in Gaussian vecor auoregressive models, Economerica 59: Johansen, S. (995). Likelihood-based Inference in Coinegraed Vecor Auoregressive Models, Oxford Universiy Press, Oxford. Johansen, S., and Juselius, K. (990). Maximum Likelihood Esimaion and Inference on Coinegraion wih Applicaions o he Demand for Money. Oxford Bullein of Economics and Saisics, 52(2), Johansen, S. (988). "Saisical analysis of coinegraing vecors." Journal of Economic Dynamic and Conrol 2(June-Sep): Kamin, S. B. and Klau, M. (998). Some Muli-counry Evidence on he Effecs of Real Exchange Raes on Oupu. Inernaional Finance Discussion Papers, no. 6. Washingon, D.C.: Federal Reserve Board. Morley, S. A. (992). On he Effec of Devaluaion During Sabilizaion Programs in LDCs. Review of Economics and Saisics 74, no. : Ndung u, Njuguna. (993). Dynamics of he Inflaionary Process in Kenya. Goeborg, Sweden: Universiy of Goeborg. Rober A. Mundell, The Moneary Dynamics of Inernaional Adjusmen under Fixed and Flexible Exchange Raes, Quarerly Journal of Economics, 74, 960, pp Ndungu, N.S. (997). Price and Exchange Rae Dynamics in Kenya: an Empirical Invesigaion ( ). AERC research Paper,58. Odusola, A.F. and A.E. Akinlo (200). Oupu, Inflaion, and Exchange Rae in Developing Counries: An Applicaion o Nigeria. The Developmen Economies XXXIX-2: Phillip, P.C.B. and Perron, P. (988). Tesing for a Uni Roo in Time Series Regression. Biomerika, 75: Rogers, John H., and Ping Wang. (995). Oupu, Inflaion and Sabilizaion in a Small Open Economy: Evidence from Mexico. Journal of Developmen Economics 46, no. 2: Sims, C. A. (980). Macroeconomics and Realiy. Economerica, 48(), -48. Zhang, D.,(2008). Oil shock and economic growh in Japan: A nonlinear approach. Energy Economics 30,

9 This academic aricle was published by The Inernaional Insiue for Science, Technology and Educaion (IISTE). The IISTE is a pioneer in he Open Access Publishing service based in he U.S. and Europe. The aim of he insiue is Acceleraing Global Knowledge Sharing. More informaion abou he publisher can be found in he IISTE s homepage: hp:// CALL FOR JOURNAL PAPERS The IISTE is currenly hosing more han 30 peer-reviewed academic journals and collaboraing wih academic insiuions around he world. There s no deadline for submission. Prospecive auhors of IISTE journals can find he submission insrucion on he following page: hp:// The IISTE ediorial eam promises o he review and publish all he qualified submissions in a fas manner. All he journals aricles are available online o he readers all over he world wihou financial, legal, or echnical barriers oher han hose inseparable from gaining access o he inerne iself. Prined version of he journals is also available upon reques of readers and auhors. MORE RESOURCES Book publicaion informaion: hp:// Recen conferences: hp:// IISTE Knowledge Sharing Parners EBSCO, Index Copernicus, Ulrich's Periodicals Direcory, JournalTOCS, PKP Open Archives Harveser, Bielefeld Academic Search Engine, Elekronische Zeischrifenbibliohek EZB, Open J-Gae, OCLC WorldCa, Universe Digial Library, NewJour, Google Scholar