Trade and Developmen Review Vol. 7, Issue 2, 2014, 1-17 hp://www.drju.ne Empirical Approaches o he Pos-Keynesian Theory of Demand for Money: An Error Correcion Model of Bangladesh Nobinkhor Kundu Muhammad Musharuf Hossain Mollah The demand for money is crucial imporan ool of moneary policy o deal wih he macroeconomic problems and o prescribe appropriae policy of he economy. This paper invesigaes o empirically explore he long-run equilibrium for demand for real money balance as well as shor-run dynamics in he conex of moneary policy in Bangladesh. Using ime-series annual daa for he period 1981 o 2012 and applying he mehods of coinegraion and error-correcion, he sudy find a single coinegraing equaion showing long-run sable relaionship beween demand for money and explanaory variables in he model. The sudy also finds convergence of shor-run dynamics owards saisically significan long-run equilibrium and concludes ha he resuls have imporan implicaions for he conduc of moneary policy in Bangladesh. JEL Classificaion: E41, C22, C52 Keywords: Demand for Money, Coinegraion, Bangladesh I. Inroducion The demand for money is one of he mos imporan componens of he ransmission mechanism of moneary policy on he macroeconomic views. A sable money demand funcion is a condiion in he conduc of moneary policy as i enables a Deparmen of Economics, Comilla Universiy, Comilla-3503, Bangladesh, Tel: +88-01911662226, Email: nobinkundu@yahoo.com School of Business Sudies, Souheas Universiy, Dhaka-1213, Bangladesh, Tel: +88-01716200047, Email: mhmru@yahoo.com Jadavpur Universiy.
policy-driven change in moneary aggregaes o have predicable influences on oupu, ineres rae, price and ulimaely exchange rae. The analysis of he demand for money plays an imporan role in he decision-making process of cenral banks which has been working on a demand-for money analysis inensively. Empirical sudies have been conduced in boh developed and developing counries o evaluae he deerminans and he sabiliy of he demand for money funcion for various moneary aggregaes. For developed counries, Friedman (1959), Chow (1966) and Goldfeld (1973) found ha permanen income o be he superior scale variable over measured curren income even for developing counries. The sable and predicable money demand funcions in developed counries are in mos cases suppored by well managed daa se and insiuional seings. The scenery is almos opposie in developing counries. The choice of he scale variable, curren income as measured by real GDP (gross domesic produc) is due o he empirical findings of Adekunle (1968), ha is, income and price expecaions are saic in Leas Develop Counries (LDCs). I migh no be inappropriae ha he demand for real balances should be relaed o curren income and curren rae of inflaion, hus we use nominal GDP measure as a proxy for curren income and curren inflaion rae as a proxy for expeced inflaion rae. Taslim (1984) inroduced expeced inflaion rae as a measure of opporuniy cos of holding money insead of using nominal ineres rae ha was significan. He criicized using of nominal ineres rae as an explanaory variable as i was deermined by insiuional seings raher han marke forces. Conversely, for a developing counry wih immaure capial marke and insiuionally regulaed ineres raes, i has been argued ha he use of inflaion (expeced) over ineres raes as a measure of he alernaive cos of holding money is more apposie. Consequenly, i would no be inappropriae ha he demand for real balances depend on curren income in Bangladesh. Ahmed (1977) found coefficiens ha were saisically significan a respecable levels and of he expeced signs. He concluded ha here exised a sable money demand funcion in Bangladesh. Muri and Muri (1978) focused on he appropriae funcional form of he Demand for Money in Bangladesh using he generalized Box- Cox Transformaions. The resuls found by hem were similar o hose of Ahmed's resuls as hey adoped he whole daa se of Ahmed and esimaed he same variables and funcions. Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 2
Hossain (1992) sudied he sabiliy of money demand funcion in Bangladesh. Using boh Chow and CUSUM and CUSUMS'Q es, and he found a sable broad money funcion while he idenified he insabiliy of narrow money demand funcion, which migh have been caused by financial reforms in Bangladesh since he early 1980s. Islam (2000) who employed sophisicaed economeric echniques, Johansen-Juselius Co-inegraion ess, in esimaion of money demand funcion in Bangladesh using quarerly daa. Hossain (2006) over again was invesigaed dynamic money demand behaviour by applying he mehods of coinegraion and error-correcion and using annual daa. Islam and Hossain boh have been employed by he coinegraion and error-correcion echniques. They found a sable single co-inegraing vecor exiss in he long-run equilibrium money demand relaionship in Bangladesh. This paper briefly reviews explains relevan empirical issues in modeling and esimaing money demand funcion for Bangladesh and provide some addiional perspecives no covered in oher reviews. The main purpose of research sudy is o empirically explore he long-run equilibrium money demand relaionship as well as shor-run dynamics (i.e., sabiliy and he speed of adjusmen o he long-run equilibrium) in Bangladesh for real money balance, real income, ineres rae, inflaion rae and exchange rae. The remainder of his paper is organized as follows. Theoreical foundaion is presened in secion 2. Secion 3 describes he economeric mehodology used for he presen sudy as well as he daa. Secion 4 discusses analyical frameworks for empirical esimaes. Resuls are presened in Secion 5. Finally, Secion 6 concludes he paper. 2. Theoreical Foundaion Assumpion of wage-price flexibiliy in he classical macroeconomic sysem resuls in all markes being in equilibrium wih full employmen of labour force. The demand for money in his sysem is represened by he Quaniy heory, which emphasized on a direc and proporional relaionship beween money and he price level. This relaionship was developed in classical equilibrium framework by wo alernaive bu equivalen expressions: One is he Equaion of exchange, associaed wih Fischer s (1911) equaion: MV = PT, where M is he quaniy of money in Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 3
circulaion, V is he ransacions velociy of circulaion of money, T is he volume of ransacions and P is he price level. The oher is he Cambridge approach or cash balance approach, associaed wih he Cambridge Universiy economiss, especially Pigou (1917). Cambridge economiss poined ou he role of wealh and he ineres rae in deermining he demand for money. According o Friedman (1956), among of modern moneariss, sabiliy of developmen of he demand for money depends on he overall wealh of sociey in various forms (money, bonds, securiies, maerial and human resources) as well as on he ase and preferences of holders of he wealh. Keynes (1936), on he oher hand, posulaed ha he demand for money arises no only due o he ransacional moive bu also due o people s desire o hold cash balances for speculaion. This speculaive demand for money is wha Keynes called he liquidiy preference. Formally, he Keynesian demand for money can be wrien as: ( Y ) L ( i) M d = L 1 + 2 where M d is demand for money, L 1 expresses he ransacional and precauionary moive, L 2 expresses he speculaive moive of liquidiy preference, Y is nominal GDP and i is he ineres rae. The Neo-Keynesian inerpreaion of he money demand is based on Keynes s principles. Thus, he demand for money can be expressed as follows: M da = ky and = α βi M ds where, M da is demand for acive balances, M ds is speculaive demand for money, k is he share of acive balances in GDP, α and β are parameers. This approach was developed by Baumol (1952) and Tobin (1956) o an approach based on he possession of money as invenory, where he ransacional moive of liquidiy preference is he well-known formula: M d P = cy/2i Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 4
where M d is demand for real balances, c is ransacional coss, Y is real GDP and P i is he ineres rae. However, he Bauman-Tobin model assumpion of cos sabiliy in a ransacion (c) is no realisic in he long run. The Pos-Keynesian economics differs from neo-keynesian especially in he inclusion of he financial moive in he demand for money. In his approach, he demand for money is usually expressed in nominal erms. For ransformaion o he real demand for money form, i is necessary o consider inflaion. In his conex, he approach of Philip (1988) o he demand for money in a small open economy can be expressed using he following equaion: M d = K a e b c e ( Y ) ( P ) ( CR) ( ER ) d μ r where M d is real money balances, K is he Cambridge coefficien funcion, Y r is real GDP, P e is he expeced rae of inflaion, CR is an esimaed variable for credi limiaions, ER e is he expeced appreciaion or depreciaion rae of he currency, μ is a non-sysemaic componen and a, b, c, and d are elasiciy values. Our esimaion of he demand for money in Bangladesh is based on his approach. 3. Economeric Model and Daa 3.1 Empirical Mehodology In general, he empirical esimaion of demand for money or real cash balances funcion can be expressed using he following equaion: M * P * β β 4 β1 β2 3 μ = α Y I P R e (1) 0 * where M denoe he desired sock of nominal money balance, P * is he price level used o conver nominal money balance o real money balance M, Y is aggregae real naional income as scale variable, I is he long-erm nominal ineres rae (%) as opporuniy cos variable, P is he rae of inflaion as measured by he consumer price index, and R is he exchange rae as raio of Bangladesh Taka and U.S. Dollar Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 5
respecively. The above demand for money funcion can be rewrien in naural log (ln) linear form as, ln M = α 0 + β1 lny + β 2 ln I + β3 ln P + β 4 ln R + μ (2) For a ime series esimaion, saionariy is an imporan facor. However, empirical analysis in erms of firs differencing o induce saionariy has been quesioned by Engle and Granger (1987). They argue ha he radiional approach of firs differencing o induce saionariy disregards poenially imporan equilibrium relaionships among he levels of he series o which he hypoheses of economic heory are usually aken o apply. The behavioral assumpions require ha β 1 > 0, β 2 < 0, β 3 < 0, β 4 < 0 and ha he μ sequence μ = ln R (3) ln M α 0 β1 lny β 2 ln I β 3 ln P β 4 is saionary, so ha any deviaions from long-run money marke equilibrium are emporary in naure. We argue ha real money balances, real income, nominal deposi rae, inflaion rae and exchange rae are mos likely inegraed of order one, so ha heir changes are saionary. However, saionariy in μ would esablish (3) as a plausible long-run relaionship, wih he shor-run dynamics incorporaed in μ, usually referred o as he equilibrium error. Then he inegraed variables M, Y, I, P & R are said o be coinegraed and equaion (2) is referred o as he coinegraing regression, as in Engle and Granger (1987). In marix noaion, an equilibrium money demand model requires ha M 1 Y μ = β X = 1 1 2 3 β 4 = I P R [ α β β β ] Saionary Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 6
The vecor β = [ 1 α β β β β ] 1 2 3 4 is called he coinegraing vecor for he nonsaionary sochasic process X, corresponding o[ M ], Y, I, P & R. This coinegraing vecor isolaes (in he presen conex) he saionary linear combinaion, μ. 3.2 Daa A definiion of broad money (M 2 ) is used, which a beer measure han a narrow money (M 1 ), and nominal deposi rae as proxy for long-erm ineres rae is used ha is beer han he shor-erm ineres rae o measure he opporuniy cos of holding money in considering he long-run economic impacs of changes in moneary policy in developing counry. Afer compilaion, all daa series are ransformed ino naural log form. This can reduce he problem of heeroskedasiciy because i compresses he scale in which he variables are measured, hereby reducing a enfold difference beween wo values o a wofold difference. The broad money (M 2 ) deflaed by he price level is considered here as he real money balance; gross domesic produc (GDP) is aken as proxy for aggregae real naional income (Y), and he rae of inflaion (P) is proxied by consumer price index. The exchange rae (R) is he raio of Bangladesh Taka and U.S. Dollar. The daa for hese series are for he pas four decades saring from year 1981 o 2012 as aken from he World Developmen Indicaors (WDI) 2013 daa base of he World Bank (WDI, CD-Rom) and Bangladesh Bank. 4. Analyical Framework 4.1 Uni Roo Tess The Augmened Dickey-Fuller (ADF) es is generalized o allow for higher-order auoregressive dynamics, in case ha an AR(1) process is inadequae o render ε whie noise (Dickey and Fuller, 1981). The ADF es for a uni auoregressive roo ess he null hypohesis H o : δ = 0, agains he alernaive H 1 : δ < 0 in he following regression equaion wih an inercep of he form: ΔY m 0 + δy 1 + θi ΔY i ε = 1 1 (4) = α + Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 7
If Y is saionary around a deerminisic linear ime rend, hen he rend i.e., he number of observaion, mus be added as an explanaory variable. Alernaively (4) can be wrien as he following regression equaion wih an inercep and a rend of he form: ΔY m 0 + β0 + δy 1 + θi ΔY i ε = 1 1 (5) = α + where ΔY = Y Y 1 is he firs difference operaor of Y series, and Y is he variable under consideraion, is chosen by Schwarz Informaion Crierion (SIC), α is inercep erm, β is rend variable and ε is a whie noise error erm, m is he number of lags in he dependen variable. The opimal lag lengh, m, can be chosen using daa dependen mehods ha have desirable saisical properies when applied o uni roo ess. In he equaion (4), Y corresponds o modeling a random walk wihou drif (inercep α = 0 and coefficien of rend β = 0 ), whereas in (5), Y is a random walk wih drif ( β = 0 ) around a sochasic rend. A es for nonsaionary of he series Y amouns o a -saisic of δ = 0. Alernaively, hypohesis of saionariy requires ha δ be negaive. 4.2 Coinegraion Tes The Johansen s (1988) maximum likelihood (ML) approach is sufficienly flexible o accoun for long-run properies as well as shor-run dynamics, in he conex of mulivariae vecor auoregressive models. Le us consider he following p- dimensional vecor auoregressive (VAR) model of order k k A i i X u = 1 i (6) X = + where X is a p 1 vecor and u is an independenly and idenically disribued. In he case of he sochasic process X =[ ln M, ln Y,ln I,ln P, ln R ]. Johansen and Juselius (1990) sugges wriing equaion (6) as k i = 1 ΔX = ΓΔX +ΠX + u i i k (7) Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 8
where Γ i = i k ( I = A ) j 1 j and Π = ( I i ) A = and, Δ = 1 L, where L is he lag operaor; I is he n n ideniy marix; A and elemens of X will be given by he rank of Π, denoed as r. λ is an eigenvalue of esimaed Π. i 1 Johansen proposes wo ess for he number of disinc coinegraing vecors. In he race es, he null hypohesis ha here are a mos r coinegraing vecors is esed (agains a general alernaive) by calculaing he es saisic p λrace() r = T ln 1 λi (8) i= r+ 1 In his case, each ln( 1 ˆ λi ) will be equal o zero (since log 1 = 0), and λ race will also be equal o zero. However, he farher he esimaed eigen values are from zero, he more negaive is each of he expressions, and he larger he λ race saisic. In he maximum eigen value es, he null hypohesis of r coinegraing vecors is esed agains he alernaive of (r + 1) coinegraing vecors by calculaing he es saisic. ( r r + 1) = T ln λ λ max, 1 r+1 (9) Again, if he esimaed eigen value, λ r+1, is close o zero, λ max will be small, and he null hypohesis ha he number of coinegraing vecors is r will no be rejeced. 4.3 Vecor Error Correcion Model (VECM) The coinegraion and error-correcion frameworks have proved o be successful ools in he idenificaion and esimaion of aggregae money demand funcions. This ype of approach o he demand for money capures he long-run equilibrium relaionship beween money and is deerminans as well as he shor-run variaion and dynamics. In fac, here may be disequilibrium in he shor run. To invesigae he shor run dynamics among he concerned ime series variables, Vecor Error Correcion Model (VECM) is used according o Enders (2009). An unresriced VECM considering up o ρ lags for demand for money funcions is respecively as follows: Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 9
Δ ln M = δ ρ ρ ρ ρ 0 + θ k Δ ln M j + η k Δ lny j + φk Δ ln I j + ϕ k Δ ln j= 1 j= 1 j= 1 j= 1 P j [ ln M ˆ α ˆ 1 0 β1 lny ˆ 1 β 2 ln I ˆ 1 β 3 ln P 1 ˆ β 4 ln R ] + ε ρ + ψ k Δ ln R j + λ 1 j= 1 (10) where Δ is he firs difference operaor, λ depics he speed of adjusmen from shor run o he long run equilibrium, ε is a purely whie noise erm. In paricular, if he variables are inegraed and coinegraed, hen here is an error-correcion represenaion ha enables he esimaion of long-run equilibrium relaionships wihou simulaneously having o ake a srong posiion on how o model shor-run dynamics. 5. Resuls 1 The sandard ADF es has been used o perform he uni roo es o he M, Y, I, P and R series separaely of he model and examine heir order of inegraion. The ADF es used here includes a consan and linear rend in he es regression since i has more general specificaion. The es has employed auomaic lag lengh selecion using a Schwarz Informaion Crierion (SIC) and a maximum lag lengh of 7. SIC is considered o be more appropriae because of small numbers of observaions in he sudy (32 observaions). Table-1 repors he es saisics for he model wih a ime rend and inercep in level and in firs differences respecively. The esimaed saisic for all he variables a level does no exceed ADF es saisics. I shows ha he null hypohesis of uni roo canno be rejeced a 5 per cen level of significance for all variables a level. To es for he presence of more han one uni roo in all hese variables, he uni roo ess of he variables a firs difference have o be checked. The resuls, of able-1, show ha he uni roo hypohesis is rejeced a he firs differences for all variables. 1 Afer operaing analysis in sofware EViews-5.1 version, we go he resul significan and observing he obained resul we can illusrae ha he macroeconomic variables in Bangladesh. Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 10
TABLE -1: ADF STATISTICS FOR TESTING FOR UNIT ROOTS IN LEVEL AND FIRST DIFFERENCES Variables -ADF(Inercep and Trend) Level ( ) P-Value Firs Difference ( ) P-Value lnm -1.981 (1) 0.587-4.483 *** (0) 0.006 lny -0.478 (0) 0.979-6.019 *** (0) 0.000 lni -2.792 (1) 0.210-3.598 *** (0) 0.029 lnp -2.724 (1) 0.234-5.938 *** (1) 0.000 lnr -4.295 *** (1) 0.010-4.723 *** (0) 0.003 Noe: (i) *, ** and *** denoe rejecion of he uni roo hypohesis a he 10%, 5% and 1% level of significance respecively. (ii) Figures in he parenheses represen he opimal lag lengh as deermined by Schwarz informaion crieria. This resul from uni roo ess provide srong evidence of non-saionariy a levels and saionariy a firs difference for all variables, hese series are inegraion o degree one, I(1). The residuals are also found o be saionary using a Schwarz Informaion Crierion (SIC) and a maximum lag lengh of 7 and 32 observaions. The resul provides he basis for he es of long-run relaionship among all variables ha are saionary. The coinegraion beween variables reveals he exisence of he sable long-run (equilibrium) relaionship. To es for coinegraion among he variables, Johansen Maximum Likelihood procedure has been applied o a vecor auoregressive (VAR) version. The resuls show haλ race and λ max indicae 1 coinegraing equaion a he 5 percen level of significance, presened in able -2. Hypohesiz ed No. of CE(s) Eigen value TABLE-2: THE COINTEGRATION ANALYSIS SERIES: LN_M LN_I LN_P LN_Y & EXOGENOUS SERIES: LN_R Lags inerval (in firs differences): 1 o 1 Unresriced Coinegraion Rank Tes (Trace and Maximum Eigen value) λ 0.05 race Criical Value Prob. ** λ 0.05 max Criical Value Prob. ** None 0.701961 64.22940 54.07904 * 0.0048 36.31589 28.58808 * 0.0042 A mos 1 0.458623 27.91351 35.19275 0.2452 18.40919 22.29962 0.1602 A mos 2 0.169893 9.504316 20.26184 0.6883 5.586027 15.89210 0.8325 A mos 3 0.122440 3.918289 9.164546 0.4244 3.918289 9.164546 0.4244 Noe: Trace es and Max-eigenvalue es indicae 1 coinegraing equaion(s) a he 5% level, * denoes rejecion of he hypohesis 5% level, ** probabiliy values. Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 11
The resuls provide evidence ha he null hypohesis of no coinegraion, i.e., r = 0, is rejeced for he esimaed demand for real cash balance of Bangladesh. So here is a single coinegraing vecor relaionship among he variables of equaion (2) a he 5 percen level of significance. The parameer esimaes represening he coinegraion beween he demand for real cash balance and he endogenous facors in he model, is specified as: ln(m) 0.683 ln(y) + 0.602 ln(i) + 0.305 ln(p) + 5.124= 0 Or, ln(m) = 5.124 + 0.683 ln(y) 0.602 ln(i) 0.305 ln(p) (11) (-6.231) (3.285) (-4.452) (-7.338) Wih he exisence of coinegraion esablished, equaion (2) is re-parameerized as an error correcion model (ECM) o esimae a model for improved forecasing, including he effecs of exogenous variables. The coinegraing equaions are generally inerpreed as he long run equilibrium relaionships characerizing he daa, wih he error correcion equaions represening shor-run adjusmen owards such equilibria. The error correcion model alone also can make direc inference boh abou he long-run and shor-run relaionships. Since here is a single coinegraing equaion, he vecor auoregressive (VAR) needs o include an error correcion erm involving levels of he series, and his erm appears on he righ-hand side of each of he VAR equaions, which oherwise will be in firs differences. The error correcion model for he real money balance is including he exchange rae (R) o capure he effecs of currency appreciaion or depreciaion on demand for money of Bangladesh. The esimaed equaion of he model in error correcion form for he demand for real money balance (M) is: Δ ln(m ) = 0.377Δln(M -1) + 0.257Δln(M -2 ) + 0.410Δln(Y -1 ) + 0.698Δln(Y -2) (3.356) (2.724) (2.184) (2.343) 0.183 Δln(I -1) 0.104 Δln(I -2) 0.101Δln(P -1) + 0.111Δln(P -2) 0.571ln(R ) (-3.735) (-2.985) (-3.438) (2.382) (-2.957) 0.319 [ln(m) 0.683ln(Y) + 0.602ln(I) + 0.305ln(P) + 5.124] (-3.759) (-3.285) (4.452) (7.338) (6.231) (12) In he shor-run he sign of he esimaed coefficiens of he real naional income (Y) Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 12
a 1 and 2 period ime lag boh are posiive and a he 5% level of significance, which shows an increased demand for real cash balance (M) a boh period lag. As a consequence, he posiive effec of he rise in real naional income on real cash balance of Bangladesh is consisen. Conversely, he sign of he esimaed coefficiens of he nominal rae of ineres (I) a 1 and 2 period ime lag are boh negaive and a he 5% level of significance, which shows a reduced demand for real cash balance (M) a boh period lag. I measure he opporuniy cos of holding money in considering he long-run economic impacs of changes in he moneary policy in a developing counry like Bangladesh. Wih respec o he rae of inflaion (P), he demand for real cash balance of Bangladesh responds in a cyclical paern, ha is, deerioraion in real cash balance a 1-period ime lag, improvemen in he nex period which is followed by deerioraion again. The effec of changes in he rae of inflaion on he real cash balance of Bangladesh is negaive in he beginning and posiive from he second period. This implies ha effec of changes in he rae of inflaion on he demand for real cash balance of Bangladesh decrease immediaely afer he decrease in heir real cash balance, which is consisen wih he idea of he inflaionary approach, and laer i reverses. The key finding from he shor-run dynamics above is ha of a negaive and saisically significan speed of adjusmen coefficien (he error correcion erm). This means ha he speed a which he rae of variaion of he demand for real cash balance, Δln(M), he dependen variable in he firs equaion of he vecor error correcion (VEC) sysem, adjuss owards he single long-run coinegraing relaionship differs from zero. In oher words, he equaion for he demand for real cash balance Δln(M) conains informaion abou he long-run relaionship since he coinegraing vecor does ener his equaion. According o he esimaes, shor-run demand for real cash balance disequilibrium is correced a he rae of 32 percen per annum. The speed of adjusmen coefficien indicaes ha he demand for real cash balance converges o he equilibrium and he convergen sign indicae ha his is saisically significan in he long run. An Exension of he Long run relaionship Solving equaion (12) he long-run relaionship beween he variables in he model can be esimaed as (while all he Δ s equal zero a equilibrium): Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 13
ln (M) = 1.634 + 0.218ln(Y) 0.192ln(I) 0.097ln(P) 0.571ln(R) (-6.231) (3.285) (-4.452) (-7.338) (-2.957) A his poin he exchange rae (R) eners in he equaion as an exogenous variable. The equaion reveals ha he esimaed coefficien of real naional income (Y) has a posiive sign wih high level of significance. Accordingly, an increase in he real income of Bangladesh leads o an increase in he demand for real cash balance (M). The resuling esimae of β 1 is 0.218, wih a -value of 3.285. I indicaes ha a one basis poin (percen poin) increase in he real naional income, keeping he oher variables consan, leads o an average 0.218 basis poin increase in he real cash balance. Tha is, he poin esimae suggess ha a change in real naional income is associaed wih an increase in he real cash balance. The negaive sign of he esimaed coefficien for nominal ineres rae (I) variable is consisen wih he moneary view and is saisically significan, denoing significan negaive effec of he change in nominal ineres rae on he change of he demand for real cash balance (M) in he long run. The resuling esimae of β 2 is 0.192, wih a -value of -4.452. Similarly, negaive sign of he esimaed coefficien for rae of inflaion (P) variable is consisen wih he resuling esimae of β3 is 0.097, and a - value of -7.338. I provides a raionale for he idea ha moneary policy should be concerned wih managing inflaion expecaions in order o keep real ineres raes a a sable level ha promoes saving and invesmen. The sign of he esimaed coefficien of exchange rae (R) is negaive and saisically significan. The resuling esimae of β 4 is -0.571, wih a -value of -2.957. I indicaes ha a one basis poin increase in he exchange rae, assuming oher variables remain consan, leads o an average -2.957 basis poin reducion of he demand for real cash balance. As a consequence, his ineracion suggess ha depreciaion of exchange rae is associaed wih reduced demand for real cash balance of Bangladesh. 6. Conclusion The paper sheds some lighs on he impac of some key variables on he demand for real balances in Bangladesh. The opporuniy cos variables carry he expeced sign according o economic heory. I is also observed ha he demand for real balances in he economy is srongly dominaed by he ransacions moive for holding money. Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 14
The model is esimaed using sandard ime series economeric echniques, coinegraion and error-correcion mehod afer esing for he saionary of he daa series and coinegraion among variables of he model. Coinegraion analysis reveals ha here is a saionary long run relaionship beween he demand for real cash balance, he real naional income, he nominal ineres rae, he rae of inflaion and he exchange rae. The esimaion resuls show ha boh he long- and shor-run demand for real cash balance model of Bangladesh are well specified, sable and saisically significan. One major policy implicaion is ha he esimaed effec resuling from variaion in he exchange rae plays a key role in Bangladesh s money demand behavior. The speed of adjusmen coefficien is negaive and indicaes ha he demand for real cash balance converges o he equilibrium in he long run. Finally, he resul also indicaes ha demand for real money balances in Bangladesh remained sable hroughou he period under invesigaion. References 1. Adekunle, J. O. (1968). The Demand for Money: Evidence from Developed and Less Developed Economies. IMF Saff Papers 2: 251-265. 2. Ahmad, S. (1977). Transacions Demand for Money and he Quaniy Theory. Quarerly Journal of Economics 91: 327-335. 3. Bangladesh Bank (Various issues). Scheduled Bank Saisics. Banking and Saisics Division. Saisics Deparmen. BB. 4. Bangladesh Bank (Various issues). Monhly Economic Indicaors: Monhly Updae. Moneary Policy Deparmen. BB. 5. Baumol, W. J. (1952). The Transacions Demand for Cash: An Invenory Theoreic Approach. Quarerly Journal of Economics 66: 545-556. 6. Chow, G. C. (1966). On he Long-Run and Shor-Run Demand for Money. Journal of Poliical Economy 74: 111-131. 7. Dickey, D. A. and Fuller, W. A. (1981). Likelihood Raio Tess for Auoregressive Time Series wih a Uni Roo. Economerica 49: 1057-1072. 8. Enders, W. (2009). Applied Economeric Time Series, 3rd ediion, Wiley. Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 15
9. Engle, R. and Granger, C. W. J. (1987). Coinegreaion and Error Correcion: Represenaion. Esimaion and Tesing. Economerica 55: 251-276. 10. Fisher, I. (1911). The Purchasing Power of Money. London: Macmillan. 11. Friedman, M. (1956). The Quaniy Theory of Money: A Resaemen. In Sudies in he Quaniy Theory of Money. M. Friedman (ed.). Chicago: Universiy of Chicago Press. 12. Friedman, M. (1969). The Opimum Quaniy of Money and Oher Essays. Chicago: Universiy of Chicago Press. 13. Goldfeld, S. M. (1973). The Demand for Money Revisied. Brookings Papers on Economic Aciviy 3: 577-638. 14. Hossain, A. (1992). Financial reforms, sabiliy of he money demand funcion and moneary policy in Bangladesh: An economeric invesigaion. Indian Economic Review 16: 85-100. 15. Hossain, A. (2006). The Money Demand Behaviour in Bangladesh, 1973-2003: An Applicaion of he Coinegraion and Error-Correcion Mehods. Indian Economic Review 41(1): 55-80. 16. Islam, A. M. (2000). Money Demand Funcion for Bangladesh. Bangladesh Developmen Sudies 16(3):1-30. 17. Johansen, S. (1988). Saisical Analysis of Coinegraing Vecors. Journal of Economic Dynamics and Conrol 12: 231-54. 18. Johansen, S. and Juselius, K. (1990). Tesing Srucural Hypoheses in a Mulivariae Coinegraion Analysis of he PPP and UIP for he UK. Journal of Economerics 53: 211-244. 19. Keynes, J. M. (1936). The General Theory of Employmen, Ineres, and Money. London and New York: Macmillan. 20. Miller, S. M. (1991). Moneary Dynamics: An Applicaion of Coinegraion and Error Correcion Modeling. Journal of Money, Credi, and Banking 23: 139-54. Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 16
21. Muri, G. V. S. N. and Muri, S. (1978). The Funcional Form of he Demand for Money in Bangladesh. Bangladesh Developmen Sudies 6(4): 443-460. 22. Philips, A. (1988). The Demand for Money in Small Developing Economies: An Applicaion of he Error Correcion Mechanism. Conemporary Issues in Money and Banking. Chalenham, Edward Elgar. 23. Pigou A. C. (1917). The Exchange Value of Legal Tender Money. Quarerly Journal of Economics 110-129. 24. Taslim, M. A. (1984). On he Rae of Ineres and Demand for Money in LDCs: The Case of Bangladesh. Bangladesh Developmen Sudies 12(3): 19-36. 25. Tobin, J. (1956). The Ineres Elasiciy of he Transacions Demand for Cash. Review of Economics and Saisics 38: 241-247. 26. Tobin, J. (1958). Liquidiy Preference as Behaviour Towards Risk. Review of Economic Sudies 25: 65-86. 27. World Bank (2013). World Developmen Indicaors-2012. World Bank CD- ROM. Washingon. Trade and Developmen Review, Vol. 7, Issue 2, 2014 Jadavpur Universiy. 17