THE MALAYSIAN PRIVATE CONSUMER EXPENDITURE: RECONCILING BETWEEN THEORIES AND DATA. Eu Chye Tan

Transcription

1 THE MALAYSIAN PRIVATE CONSUMER EXPENDITURE: RECONCILING BETWEEN THEORIES AND DATA Eu Chye Tan Faculy of Economics & Adminisraion Universiy of Malaya I. Overview This paper aemps o examine he Malaysian privae consumer expendiure behavior agains he backdrop of he Keynesian and wo pos-keynesian consumpion heories, namely he Life Cycle Theory and he Permanen Income Hypohesis. Aenion is focussed on hese wo pos-keynesian heories as hey are he principal heories of privae consumpion commonly referred o in macroeconomics exs. They also seem o gain more accepance amongs economiss compared wih Duesenberry s relaive income approach (Branson, 1989). Specifically, he paper aims o relae as exensively as possible he Malaysian annual privae consumer expendiure daa o he heories and o draw policy implicaions if warraned. Privae consumpion accouns for approximaely 43% of Malaysia s GDP in As such, undersanding he rend movemens of privae consumpion is essenial for effecive macroeconomic managemen given ha i is inegral in he Keynesian heory of income deerminaion. The relaionship beween privae consumer expendiure and income is a major aspec in macroeconomics. The res of he paper is configured as follows. Secion II describes he heories while he subsequen aemps o reconcile he heories wih Malaysian annual daa daing back o he 1960s. Policy implicaions in he ligh of he daa analysis are drawn in Secion IV. The chaper concludes wih remarks in Secion V. 1

2 II. The Theories Keynes consumpion funcion may simply be wrien as follows: c = a + by d (1) where a > 0 and 0 < b < 1, c is real consumer expendiure and y d, he real disposable income. Figure 1 is he graphical represenaion of he consumpion funcion ha plos real consumer expendiure agains real income (y). I suggess ha households are inclined o consume a smaller fracion of heir income or correspondingly end o save a larger fracion as heir income surges. Thus, he raio of consumpion o income or he average propensiy o consume (APC) falls wih he rise in income. 1 This absolue income hypohesis would imply ha he rich would save proporionally more han he poor. Anoher implicaion of he hypohesis is ha he marginal propensiy o consume (MPC) is less han he APC, implying ha as income increases from recen levels, consumpion would no rise proporionally. Conversely, consumpion would no fall proporionally wih income as households srive o defend heir consumpion sandards. However, he absolue income hypohesis could no wihsand he empirical observaion ha he long run relaionship beween consumpion and income is a proporional one raher han non proporional as equaion (1) suggess. According o Branson (1989), a heory of consumpion should be able o explain four observed phenomena viz: i) increase in he raio of saving (s) o income (s/y) wih socioeconomic sraum; ii) lower han average c/y in imes of boom bu higher during slumps, hus shorrun flucuaions in income have a bearing on he c/y; iii) c/y is generally consan in he long run as income grows along rend; and iv) he effec of wealh or asses on consumpion. The firs wo phenomena would imply ha MPC < APC while he hird, MPC = APC. The absolue income hypohesis of Keynes failed o explain he hird and fourh phenomena and his promped he developmen of new heories such as he Life-Cycle Theory of Consumpion and he Permanen Income Hypohesis (see e.g. Branson, Froyen & Low, 2001). The life-cycle heory was advanced by Franco Modigliani, Alber Ando and Richard Brumberg (see Modigliani & Brumberg, 1954; Ando & Modigliani, 1963; Modigliani, 1966 & 1986) while he permanen income hypohesis by Milon Friedman (Friedman, 1957). 1 A downward rend in he average propensiy o consume could arouse fears of a secular sagnaion of he economy. This arises from he fac ha as he raio of saving o income increases, aggregae demand may evenually fall shor of full employmen oupu unless he rise in he raio is offse by oher componens of aggregae demand, namely governmen expendiure and privae invesmen. The secular sagnaion hypohesis may be illusraed as follows: If y = c + i + g or 1 = c/y + i/y + g/y is he condiion for equilibrium real oupu y, and if i/y does no rise, hen i is imperaive ha g/y mus be increased o offse he decline in c/y in order o mainain full employmen demand as income grows. Hence unless governmen expendiure increases more rapidly han income, he economy would sagnae. 2

3 The pos-keynesian heories have a microeconomic foundaion in he heory of consumer choice. I is assumed ha he observed consumer behavior represens an oucome of he ques by raional individuals o maximize uiliy by opimally allocaing a lifeime sream of earnings so as o secure an opimum lifeime consumpion paern. Specifically, a consumer prefers o smooh his acual income sream ino a more or less even consumpion paern. Thus he heories depar from he assumpion ha here exiss a single represenaive consumer wih he following uiliy funcion: U = U c,..., c,..., c ) (2) ( 0 T where his lifeime uiliy (U) is a funcion of his real consumpion (c) in every ime period up o T, he insance before his demise. His objecive is o maximize uiliy subjec o he consrain ha he presen value of his oal lifeime consumpion does no ousrip he presen value of his oal income as follows: c T T 0 0 r) y = (1 + r) (1 + (3) The underlying assumpion of such a lifeime budge consrain is ha he consumer can freely allocae his income sream o a consumpion sream in an opimum manner via borrowing and lending and ha neiher does he receive nor bequeah. Such an opimizaion model would yield he following general consumpion funcion: c = f PV ) (4) ( where f >0 and PV refers o he presen value of his curren and fuure incomes a ime, y i.e. PV =. Thus an individual s consumpion a ime would vary direcly + (1 r) and conemporaneously wih he presen value of his income. Ando and Modigliani however decomposed income ino labor income (y L ) and asse or propery income (y P ). Hence, T L T P y y PV0 = + (5) (1 r) 0 (1 r) where ime 0 refers o he curren period wih ranging from zero hrough T. And if capial markes are efficien, he presen value of he income from an asse may simply be equal o is value a he onse of he curren period, i.e. a 0 T P y = (1 + r) 0 (6) 3

4 where a0 refers o he real ne worh a he sar of he period. Generally if he represenaive consumer were o experience a rise in he presen value a any insance, he would raise more or less proporionaely his consumpion in all periods from hen on. Equaion (4) may be ranslaed more specifically ino c = k(pv ) (7) where k refers o he fracion of he consumer s presen value ha he wans o consume in period. 2 Thus any increase in he presen and or expeced labor income and he value of he curren asses held ha booss his esimae of he presen value would induce he consumer o consume some fracion of he rise in he curren period. Thus he life-cycle hypohesis would imply in he conex of our analysis of he Malaysian economy he following: i) Privae consumpion depends no merely on curren income bu also expeced fuure income and curren asse holdings. Consumpion responds limiedly o variaions in curren income paricularly if hey do no no enail changes in expeced fuure income. In oher words, ransien income changes have nominal impac on consumpion; ii) However, any susained increase in curren income has a considerably greaer impac on curren consumpion; iii) The age profile of he Malaysian populaion has a bearing on he privae consumpion-income raio given ha an individual s posiion in his life cycle would deermine his consumpion and savings. The young in he work force may have relaively low income and hus low if no negaive saving raes. As heir income rises in heir middle-age years, heir saving raes may increase correspondingly. Then upon reiremen, heir incomes would slacken and hey would sar dissaving. This essence of he life-cycle hypohesis is capured in Figure 2. I depics a ypical individual wih an expeced lifespan of T wih an income sream over his lifeime as represened by he y-curve. However based on he opimizing model skeched in he preceding secion, he individual may wish o mainain a generally consan or perhaps moderaely rising level of consumpion over his lifeime as indicaed by he c-curve. Thus, in he early sage of his life, he individual is a ne borrower. In he mid-sage, he has sufficien income o repay debs and make provisions for reiremen afer meeing he consumpion needs. Subsequenly in he lae sage, he dis-saves. Friedman s permanen income hypohesis also yields he relaionship beween consumpion and presen value as in equaion (4). However Friedman differs from Ando- Modigliani in his reamen of he presen value erm by muliplying i by a rae of reurn ( r) as follows: 2 The fracion would depend on his iner emporal uiliy funcion and discoun rae. 4

5 y = r. PV (8) P P Friedman erms his produc as permanen income ( y ). The permanen income is derived based on he consumer s presen value defined o include human capial, i.e. he presen value of his curren and fuure sream of labor income and income generaed by asses or propery. I is generally he expeced average long erm income derived from boh human and non human wealh. Friedman acknowledges he exisence of random or ransiory componens of consumpion and income by disinguishing beween permanen consumpion and ransiory consumpion and beween permanen income and ransiory income. Acual P oal consumpion (c ) in any period is a summaion of permanen consumpion ( c ) and a random ransiory componen ( c T ) which could be posiive or negaive or even zero as follows: c = c + c (9) P T Hence acual consumpion may be greaer or less han permanen consumpion a any one ime. Likewise, acual oal income (y) a any ime period is consiued by permanen income T and a random ransiory componen ( y ) which could also be posiive, negaive or zero as below: y = y + y (10) P T Thus acual income may exceed or fall shor of permanen income. I is assumed ha correlaions do no exis beween ransiory and permanen income, beween ransiory and permanen consumpion, and beween ransiory consumpion and ransiory income. I is he relaionship beween permanen consumpion and permanen income ha maers and cenral o Friedman s model. Thus adaping equaion (7) o capure his relaionship yields he following. c = ky (11) P P This simply says ha permanen consumpion is proporional o permanen income. 3 However, Friedman does no expec his equaion o be a perfec predicor of consumpion as he recognizes he fac ha apar from consumpion driven by permanen income, here is in any period a random componen of consumpion. 3 The individual s raio of permanen consumpion o permanen income (k) would presumably depend on he reurn on savings or he ineres rae, iner-emporal preferences of he individual and he variabiliy of expeced income. 5

6 One facor ha could spell a limiaion o he realisicness of he llife-cycle hypohesis is he omission of bequess as a facor deermining saving. Kolikoff and Summers (1981) mainain ha he desire o make bequess is an imporan facor driving saving. However saving for beques purposes perhaps o some exen could be likened o saving for reiremen. In he conex of our empirical analysis, he model would imply he exisence of a proporional long run relaionship beween consumpion and income, hus he APC is consan and equal o MPC over a long run, hough in he shor run, he relaionship may be non proporional, causing he APC o decline wih he rise in income. This is explicable by he possibiliy ha in he long run, growh of income is dominaed by movemens in permanen income wih posiive and negaive ransiory income movemens offseing each oher. Since consumpion is mainly driven by permanen income and since years of high income would generally correspond wih he presence of a posiive ransiory income componen, he raio of consumpion o acual measured income could be low. Conversely, he raio would be high during low income years as ransiory income ends o be negaive hen. 4 The pure iner-emporal maximizing models above do no acknowledge he presence of liquidiy consrains. Even if a consumer is definie abou fuure income, opporuniies o borrow for a long period based on expeced fuure income may no exis. In his respec, consumpion may acually be more responsive o curren income changes wheher ransien or oherwise, han wha is suggesed by he models. Thus he iner emporal budge consrain [equaion (3)] may have o be modified and his would yield a liquidiy-consrain model. The iner-emporal budge consrain of equaion (3) assumes ha he consumer has borrowing and lending opporuniies a he same rae of ineres. This assumpion may however be generally unrealisic as many are in no posiion o borrow freely for consumpion agains fuure income. Their consumpion decisions are consrained by curren liquidiy which may be defined as curren income plus asses a hand. An exreme case of liquidiy consrain would involve consumpion being limied by curren liquidiy such as follows: c y + a i i i where y is curren income ne of deb repaymen obligaions. The presence of liquidiy consrains may however no render in paricular he life-cycle heory oally invalid as he consrain may only be one-sided. Specifically, only borrowing a he young age may be consrained while saving for reiremen in he middle age may remain opimal and feasible. So he consumpion pah may be as refleced in Figure 3 wih consumpion maching income, i.e. c =y unil poin A. Beyond ha poin, he life-cycle pah of Figure 2 would be followed. Generally, liquidiy consrains are mos binding amongs youh and 4 Remember ha consumpion may have a ransiory componen bu unrelaed o income. 6

7 he impac of cyclical income flucuaions on consumpion would be fel mos by he younger se of he working populaion. III. Daa Analyses This secion analyzes he Malaysian privae consumpion daa by aking an eclecic view ha all he hree heories of privae consumer expendiure could possibly hold. Table 1 shows he annual average share of privae consumer expendiure in he gross domesic produc of seleced counries generally over he period I is ineresing o noe ha he share of 43.3% recorded in Malaysia as opposed o he higher shares in advanced counries such as Japan (56.6%), Souh Korea (54.1%), Unied Kingdom (66%) and he Unied Saes (69.4%) seems broadly consisen wih he life-cycle heory of consumpion as he laer group of counries have more aged populaion compared wih he relaively young Malaysian populaion. Wih a higher proporion of elderly people, heir naional saving raes could be lower as he elderly end o have low or negaive savings. The oher ASEAN counries namely Philippines, Indonesia, Thailand and Singapore may have a demographic srucure generally more similar o Malaysia han o he advanced counries. However, he Filipinos, Indonesians and Thais appear o consume relaively more han Malaysians and as much if no more han he naionals of he developed counries. Such a phenomenon could plausibly be explained by he Keynesian absolue income hypohesis as Indonesia, Philippines and Thailand have per capia income levels lower han Malaysia. Hence, he average propensiy o consume in hese counries ends o be higher. Table 2 provides he disribuion of he Malaysian populaion by age srucure. I is noeworhy ha he proporion of Malaysians in he generally economically acive populaion age group of has expanded over he years from 52.2% in 1970 o 62.7% by 2003 while he proporion in he unproducive age group of 0-14 shrunk from 44.5% o 33.2% over he same period. The number in he generally less acive group of 65 and above has increased only marginally from 3.3% o 4.1%. The dependency raio has commensuraely dipped from 0.69 in 1991 o in Such demographic changes could have conribued o he downward rend in he Malaysian privae consumer expendiure raio as suggesed by he life cycle hypohesis. Figure 4 depics he decline in he raio hough no seadily from abou in 1967 o abou in 2003 as more Malaysians ener he age group of high earnings poenial. The secular decline could also have been saged plausibly by surging per capia income arising from he naion s remarkable economic developmen brough abou by is expororiened indusrializaion and socioeconomic redisribuion programs. Malaysia s real gross domesic produc per capia has risen by abou four-fold from RM in 1967 o RM in 2003 (Figure 5). So has real privae consumpion increased hough less rapidly by hree imes from RM1,428 o RM over he same period. The Keynesian hypoheses paricularly would predic such a decline in he average propensiy o consume on he basis of rising incomes. As Malaysians become richer, hey save more proporionally. Their consumpion would increase wih income hough by no as much. 7

8 The possible empirical relevance of he life-cycle hypohesis and perhaps Friedman s is also suggesed by he fac ha he esimaed coefficien of correlaion beween he growh of real privae consumer expendiure and he growh of real gross domesic produc is only 0.75 over he period Thus he wo variables do no move very closely wih each oher and his could be an oucome of consumers adoping a long erm earnings prospec as a basis for deciding on he quanum o consume. Hence, consumpion may no rise or fall proporionaely wih curren income as is annual variaion may also be driven by oher facors such as expeced fuure income and asse values. An aemp is made o derive proxies hough crude for permanen consumpion and permanen income by regressing he naural logarihms of real privae consumpion and real gross domesic produc agains a consan and a ime rend. The coefficien of correlaion beween permanen consumpion and permanen income is esimaed a almos uniy, hus supporing Friedman s hypohesis. However, furher exploraion of he daa would cas doubs on is validiy. The implicaion of Friedman s hypohesis ha he long-run relaionship beween consumpion and income is a proporional one and hus he average propensiy o consume is fairly consan in he long run does no appear o show up in he daa. 5 Plo of he observed daa earlier in Figure 4 has revealed a secular decline in he raio. Figure 6 which uses he rend consumpion and rend gross domesic produc as proxies for permanen consumpion and permanen income respecively also shows a similar rend. Moreover, he esimaed coefficien of correlaion beween ransiory consumpion and ransiory income of 0.88 is quie angible. 6 5 In fac, he life-cycle hypohesis could also poenially imply ha he long run average propensiy o consume is consan. The life-cycle hypohesis would imply he following aggregae consumpion funcion: e = b Y + b Y b A where C, Y, Y e and A are consumpion, curren non-asse income, expeced C non asse income and value of asses, all in aggregae erms respecively. Assuming ha A = λ hold in he long run, hen C / Y ) b + λ( b + b / ) Y = which is a consan. ( α Y e = αy and 6 Transiory consumpion (income) is aken as he difference beween acual consumpion (income) and rend consumpion (income). 8

9 IV. Concluding Remarks Taking all he evidence as a whole, i can be surmised ha he Malaysian privae consumpion behavior could be broadly explained paricularly by he Keynesian absolue income and he life-cycle hypoheses. Though he permanen income hypohesis may have some explanaory role, i is no as well suppored by he available daa compared wih he oher wo heories. I has been observed in he analysis ha here has been a downward rend in he average propensiy o consume over he years. Theoreically, his could be a cause for concern for he naional economy as his could ulimaely lead o is secular sagnaion. However his problem may be foresalled and he coninued growh of he economy remains feasible as he expor secor could be couned upon as an engine of growh. The fac ha he life-cycle hypohesis could be valid would have he following macroeconomic sabilizaion implicaions for Malaysia. Consumpion may respond srongly o curren income if changes in curren income affec expeced lifeime income. Bu ransien changes such as brough abou by emporary changes in ax raes or governmen ransfer paymens may a mos yield only a nominal impac on consumpion. However, he effecs of emporary changes in axes and governmen spending could sill be considerable if hey have a long erm bearing on he economy s growh capaciy. Moreover since real asses or household wealh could be an imporan deerminan of consumer demand under he life cycle hypohesis, he effeciveness of moneary policy may probably be enhanced relaive o fiscal policy as ineres raes could affec asse values. An expansionary moneary policy ha reduces ineres raes could boos he marke value of asses such as governmen and privae bonds and corporae equiies or socks, hus simulaing consumpion. This moneary policy effec is disinc from he effec ha i has on liquidiy consrains faced by consumers. Since governmen and privae bonds are hardly held by Malaysian households, he asse value effec of moneary policy would primarily be ransmied o consumpion via sock or equiy prices. As equiies yield dividends, he prices ha consumers would be willing o pay for equiies for a given dividend yield would rise if marke ineres raes are lower. Hence, a loose moneary policy would enhance wealh and consumpion. In conras, an expansionary fiscal policy could conribue o a rise in ineres raes hus depressing he marke value of asses. The resuling decline in privae consumpion spending may parially offse he iniial impac of a fiscal expansion. 9

10 Bibliography 1. Ando, Alber and Franco Modigliani (1963) The Life Cycle Hypohesis of Saving: Aggregae Implicaions and Tess, American Economic Review, 53 (March), pp Branson, William H. (1989) Macroeconomic Theory and Policy, New York: Harper & Row. 3. Friedman, Milon (1957) A Theory of he Consumpion Funcion, Princeon, N.J.: Princeon Universiy Press. 4. Froyen, Richard T and Linda Low (2001) Macroeconomics: An Asian Perspecive, Singapore:Prenice-Hall. 5. Kolikoff, Laurence J. and Laurence H.Summers (1981) The Role of Inergeneraional Transfers in Aggregae Capial Accumulaion, Journal of Poliical Economy, 89 Augus, pp Modigliani, Franco (1966) The Life Cycle Hypohesis of Saving, he Demand for Wealh and he Supply of Capial, Social Research, 33 (June), pp (1986) Life Cycle, Individual Thrif and he Wealh of Naions, American Economic Review, 76, June, pp and Richard Brumberg (1954) Uiliy Analysis and he Consumpion Funcion: An Inerpreaion of Cross Secion Daa, in K.Kurihara, ed., Pos Keynesian Economics (New Brunswick, N.J.: Rugers Universiy Press), pp

11 Table 1 Privae Consumer Expendiure as a % of GDP by Seleced Counry Annual Average ( ) COUNTRY % Malaysia 43.3 Philippines 70.4 Indonesia* 64.8 Thailand 56.3 Singapore 42.4 Japan 56.6 Souh Korea 54.1 Unied Kingdom 66.0 Unied Saes 69.4 * Source: Compued from IMF Inernaional Financial Saisics Yearbook, various Issues Table 2 Percenage Disribuion of he Malaysian Populaion by Age Srucure Year Age Srucure and above Dependency Raio N/A N/A N/A N/A Source: Various Five-Year Malaysia Plans 11

12 c 0 y Figure 1 Keynes Consumpion Funcion c, y y c 0 T Figure 2 The Life-cycle Model c,y y c A. 12

13 0 T Figure 3 The Liquidiy-consrained Life-cycle Model Raio Year Figure 4 Raio of Real Privae Consumer Expendiure o Real GDP $(million) Real Privae Consumer Expendiure Per Capia 9000 Real GDP per Capia Year Figure 5 Real Privae Consumer Expendiure Per Capia and Real GDP Per Capia 13

14 Raio Year Figure 6 Raio of Trend Real Privae Consumer Expendiure o Trend Real GDP 14