Does Consumption Respond to Economic Sluggishness: Some Evidence from Malaysia

Transcription

1 DOI: /IPEDR V46. 9 Does Consumpion Respond o Economic Sluggishness: Some Evidence from Malaysia Zarinah Yusof 1 and Jenny Pereira Graian Peer Pereira 1 1 Faculy of Economics and Adminisraion, Universiy Malaya, 50603,Kuala Lumpur, Malaysia Absrac. This paper adds o he evidence on consumpion behavior by incorporaing economic sluggishness from he recen wo financial crises using daa from Malaysia The finding rejecs excessively volaile consumpion behavior and suppors a sable consumpion paern. The marginal propensiy o consume 0.67 is consisen wih oher findings in he lieraure. Wealh is one of he deerminans of consumpion. Keywords: Consumpion, Financial Crisis, Error Correcion Modelling, Fiscal Simulus. 1. Inroducion In recen years consumpion has been given much aenion o simulae growh. Consumpion is one of he main drivers o he susainabiliy of economic growh. I conribued abou 50-70% of GDP. Based on modern economic heory consumpion is closely lined o lifeime income. Availabiliy of credi allows consumpion o go higher han household s income. As such, increased financial deepening and innovaion is an inegral par in supporing consumpion by faciliaing greaer access o credi. Consumpion is a forwardlooing behavior. The life-cycle and permanen income hypoheses explain ha consumpion responds more o permanen changes in income and rejec excessive volaile paern in consumpion. However a conroversy has developed over he issue as evidence on visible response of consumpion o he decline in income has been found (Gordon, 2012 and Flavin, 1981). Accordingly, he Asian financial crisis and he global financial crisis may have adverse impac on consumpion. Counries suffered coninuous currency depreciaion, soc mares crash, baning crisis, company s banrupcy and rerenchmen of worers. Alhough almos all Asian economies were affeced by he crises bu he magniude differs across he counries. I is he concern of his paper o sudy he effec of hese crises on household consumpion. Specifically, his paper examines he dynamic lin of consumpion-income relaionship in Malaysia and explores wheher consumpion in Malaysia is adversely affeced by he wo financial crises. I also analyzes wheher consumpion responds srongly o changes in acual income or behaves in a smooher paern. The sudy helps policymaers o undersand beer he behavior of household consumpion in Malaysia. I offers he esimae of muliplier effec from he esimaed marginal propensiy o consume. This is crucial for an effecive fiscal simulus program and macroeconomic managemen. Moreover, up o presen ime, research on consumpion behavior in Malaysia is scany and his serves as a srong moivaional drive of his sudy. The res of his paper is organized as follows. Secion 2 discusses lieraure review relaed o he issue of consumpion. Secion 3 presens he mehodology of he sudy and he analysis of he findings is given in Secion 4. Finally, Secion 4 concludes he sudy by highlighing he ey findings of his sudy. 2. Lieraure Review Keynes in his heory explains ha consumpion is srongly affeced by curren disposable income. People consume abou 80-90% from disposable income. However, he modern consumpion heory of life-cycle hypohesis and permanen-income hypohesis mainain ha mos people favor a sable consumpion. In oher words, consumpion is less responsive o curren income bu reac more o income in he long run. Consumpion is a sable funcion and does no flucuae. Some scholars argue ha in he presen of liquidiy consrain or unavailabiliy of credi, consumpion is srongly affeced by curren income and flucuaes wih 43

2 changes in income. This is suppored by Flavin (1981), she finds ha consumpion is excessively volaile as i reacs srongly o changes in income. In he lieraure, many of he pas sudies on consumpion behavior are exclusively focusing on he US economy (Holbroo and Safford, 1971; Hall, 1978; Sargen, 1978; Balla 1979; Hall and Mishin, 1982; Hayashi, 1982; Flavin, 1981 and 1985, Maniw, 1981 and Zeldes 1985). Time series daa are applied o he raional expecaions permanen income hypohesis in he sudies by Hall (1978), Sargen (1978), Flavin (1981 and 1985) and Maniw (1981). Panel daa are used in he sudies of Holbroo and Safford (1971), Bhalla (1979), Hayashi (1985), Zeldes (1985) and Alonji and Siow (1987). To cie sudies in Malaysia, he recen wor of Eu Chye (2009) and Ismail (2010) are considered here. In boh sudies, models are no esed sufficienly using he laes economeric mehod. Resriced by he availabiliy of daa, Eu Chye (2009) invesigaes privae consumpion behavior by analyzing he average share of privae consumpion in he GDP. Ismail (2010) examines he relaionship beween governmen spending and privae consumpion in Malaysia and concludes ha governmen spending and privae consumpion are complemenary o each oher. Thus, in his sudy he empirical wor on consumpion behavior aes more comprehensive esimaion procedures via VAR and ECM approaches and esimaes wih more recen daa. 3. Mehodology and Daa The analysis uses coinegraion echniques, vecor auoregressive (VAR) and error correcion model (ECM). Regression involving non-saionary ime series is meaningless alhough he good-of-fi of he model is very high. However, if he non-saionary series are coinegraed he esimaes are no dubious. Coinegraion implies ha here always exiss a linear combinaion of hese variables ha is saionary. Therefore, i is imporan o examine firs he ime series properies of he daa and only when each of he series is inegraed he same order, coinegraion is possible. The Augmened Dicey-Fuller (ADF) and Phillips Perron (PP) ess are used o es he order of inegraion of he series. The coinegraion follows Johansen and Juselius (JJ) (1992) mehod. I is a mulivariae coinegraion analysis using a maximum lielihood esimaion procedure. VAR models are esimaed wih unresriced inerceps and no rends. An error correcion model can be wrien as o p 1 Δy = A + Γ Δy + αβ ' y + u i= 1 i Where Γ = αβ ', marices α and β are (n x r) dimension, r is he ran of marix Γ as before. The marix β is he long-run coefficiens (coinegraing parameers) and he marix α represens speed of adjusmens o disequilibrium. The Granger-causaliy is explained hrough i) he shor-run causaliy relaionship in he differenced variables, Δ y i and ii) he long-run dynamic causal lin in he long-run comovemens (error correcion erm) of he variables, y p (Masih and Masih, 1996). The esimaing model ΔLCon = B0 + β 1 ΔLCon i + B2 ΔLYd i + B3 ΔLW i + φ CRIS γ ECT + ε 1 where B 0 > 0, 0 <B 1, B 2, B 3 < 1, u is E(u ) = 0, LCon is consumpion, LYd is disposable income (GDP minus income ax), LW is wealh, CRIS9808 is a dummy variable measuring he financial crises in 1998 and 2008, ECT is error correcion erm (he value mus be negaive for is o be sable), u is whie- noise error erm. L denoes logarihms. LCon and LYd are in real erms. Soc reurns are used as a proxy for wealh. The analysis spans from 1991 o Daa sources are from Inernaional Moneary Fund (IMF) and Monhly Saisical Bullein, Ban Negara Malaysia. 4. Findings The resuls for uni roo es using Augmened Dicey Fuller (ADF) and Phillips Perrones suggess ha all variables are inegraed of order 1, I(1) (Table 1). Nex coinegraion es is performed on he LCon, LYd and LW o deermine he exisence of any long-run relaionship beween he series. The maximum order of lags is =1. The finding shows ha all he series are coinegraed (Table 2). Boh λ-max and race saisics i p 44

3 ess rejec he null hypohesis of no coinegraion (r = 0) a he 5 per cen significan level in mos of he cases. There is a leas one coinegraing vecor a 5 per cen significan level. Table 1.Uni Roo Tess Augmened Dicey-Fuller (ADF) Ho:Uni Roo Phillips-Perron (PP) Ho: Uni Roo Series Level Difference Level Difference η η η LCon b a LYd b a LW a a r b a Noes: 1. a, and b represens significan level a 1 per cen and 5 per cen respecively. 2. The Augmened Dicey Fuller es saisics are compued wih an inercep, a linear ime rend and lagged firs-differences of he series o he series in level. The ADF regression in firs-differences, exclude a linear ime rend. The lag lengh ( ) was seleced based on Aaie Informaion Crieria (AIC). A n=25, he ADF criical values are (1 per cen), (5 per cen] and -2.62(10 per cen) for consan ( ); (1 per cen), (5 per cen) and (10 per cen) for consan and ime rend ( ). Table 2.Coinegraion Tes Hypoheses Tes Saisics Ho H 1 Max Eigen λ max Trace λ Trace Vecor:[LCon, LYd, LW] =1 r = 0 r > * * r 1 r > r = 2 r > *a 0.05 significan level Subsequenly, he model is regressed by applying he error correcion modeling echnique. Ineresingly, finding from he coinegraing vecor or he long-run relaion shows ha consumpion is srongly affeced by long-run income wih marginal propensiy o consume (mpc) of The esimae is consisen wih he recen empirical wor on he impac of fiscal simulus of $100 billion paymens o 130 million households in he Unied Saes where he mpc is 0.65 (Gordon, 2012 and Parer, 2010). Consumpion is also significanly influenced by household wealh and he mpc is as expeced, relaively low value of This finding suppors he modern heory of consumpion and rejecs he simple Keynes heory of consumpion. More finding is evidence from he esimae of error correcion model. All he variables of ineres have he expeced signs and saisically significan. Consumpion reacs lile o changes in income in he shor run and even more he coefficien is saisically insignifican differen from zero. Changes in wealh have some posiive effecs on household consumpion and he response is small. An increase in 1% of wealh drives up consumpion by 13%. Mainaining financial sabiliy is imporan for susainable economic growh. The resul suggess ha he counry is affeced by he financial insabiliy in 1998 and 2008 as consumpion is affeced negaively. Neverheless, he effec on consumpion is small (8%) as people believe ha his is only emporary shocs o he economy. The evidence from long-run relaionship shows ha any disequilibrium in he sysem will be closed by consumpion converging moderaely a a speed of adjusmen 34%. In oher words, in he shor run, deviaions from long-run equilibrium would iniiae consumpion o move bac he economy o is equilibrium. The robusness of he findings is suppored by he diagnosic ess. The residuals are all well-behaved. Coinegraing Vecor: LCon (-1) = LYd (-1) LW (-1) ( ) ( ) Error Correcion Model: LCon = LCon LYd LW -1 45

4 ( ) ( ) ( ) Crisis ECT -1 ( ) ( ) R-squared = Adj. R-squared = Jarque-Bera = ( ) Breusch-Godfrey LM Tes F(4,8) = (0.6929) Heerosedasiciy ARCH Tes F(1.15)= (0.9526) 5. Conclusion This paper examines he behavior of household consumpion in Malaysia using daa from In paricular, i analyzes wheher households are excessively sensiive o changes in income and he financial crises in 1998 and 2008 or in shor economic sluggishness. The findings reveal ha household consumpion in Malaysia remains relaively sable over he pas years. Households are more responsive o income in he long-run and less o curren income. In general, people end o smooh ou any change in income. Wealh is one of he facors ha affec consumpion bu he effec is relaively low. This finding suggess ha any fiscal policy simulus such as cash paymens or income ax refunds o he people mus be made large for i o have significan effec on he economy. Since here is endency ha people may no be spending he paymens on goods and services insead for paying deb or saved, money paymen using voucher ha lin o specific purchases is anoher way o do. The recen money paymen o he lower-income group is a righ move o increase domesic spending. This is because lower-income group s mpc is higher han high-income group. Therefore, he governmen should also consider exending he money o he middle-income group based on heir abiliy o consume. 6. References (This is Header 1 syle) [1] A. P. Jonahan, S. S. Nicholas, S. J. David and M. Rober. Consumer spending and he economic simulus paymens of NBER Woring Paper 16684, Naional Bureau of Economic Research, Inc. [2] F. Hayashi. The permanen income hypohesis and consumpion durabiliy: analysis based on Japanese panel daa. The Quarerly Journal of Economics. 1985, November: [3] J. G Alonji and A. Siow. Tesing he response of consumpion o income changes wih (noisy) panel daa. The Quarerly Journal of Economics. 1987, 102(2): [4] M. A. Flavin The adjusmen of consumpion o changing expecaions abou fuure income. Journal of Poliical Economy. 1981, LXXXIX: [5] M. A. Flavin. Excess sensiiviy of consumpion o curren income: liquidiy consrains or myopia? Canadian Journal of Economics. 1985, XVIII (February): [6] N. A. Ismail. Does governmen spending crow ou privae consumpion in Malaysia. Jurnal Kemanusian. 2010, 16. [7] N. G. Maniw. The permanen income hypohesis and he real ineres rae. Economics Leers. 1981, VII: [8] R. E. Hall. Sochasic implicaions of he life cycle-permanen income hypohesis: heory and evidence. Journal of Poliical Economy. 1978, LXXXVI: [9] R. Holbroo and F. Safford. The propensiy o consume separae ypes of income: a generalized permanen income hypohesis. Economerica, 1971, XXXIX (January): [10] R. J. Gordon. Macroeconomics. Pearson, [11] R. Masih, and A. M. M. Masih. Macroeconomic aciviy dynamics and granger causaliy: New evidence from small developing economy based on a vecor error correcion modeling analysis. Economic Modelling. 1996, 13: [12] R.E Hall and F. S. Mishin. The sensiiviy of consumpion o ransiory income: esimaes from panel daa on households. Economerica. 1982, March: [13] S. S. Bhalla. Measuremen errors and he permanen income hypohesis: evidence from rural india. American Economic Review. 1979, LXIX (June):

5 [14] S. Zeldes. Consumpion and liquidiy consrains: an empirical invesigaion. Woring Paper No , Wharon School, Universiy of Pennsylvania. [15] T. Eu Chye. Privae consumer expendiure in Malaysia: heories and evidence. Inernaional Journal of Applied Business and Economic Research. 2009, 7 (2): [16] T. J. Sargen. Raional expecaions, economeric exogeneiy and consumpion. 1978, LXXXVI: