An Empirical Analysis of Money Demand Function in Nepal

Transcription

1 An Empirical Analysis of Money Demand Funcion in Nepal Birendra Bahadur Budha * Absrac This paper analyzes he money demand funcion for Nepal during he period of he FY 1997/98 o FY 2009/10 using annual daa. The empirical resuls imply ha he coinegraion ess clearly show he exisence of he long-run relaionship beween real money balances and is deerminans, oupu and ineres rae. The vecor error correcion model has proved he shor-run relaionship beween he real money balances and is deerminans. Furhermore, Dynamic OLS esimaion of he money demand funcion indicae ha he sign of coefficiens of he oupu and ineres rae were found o be consisen wih he assumpion of he money demand heories. I. INTRODUCTION The sudy of a money demand funcion is a prime issue since he sable money demand funcion is prerequisie for he conduc of he effecive moneary policy. The demand funcion for money helps o ascerain he liquidiy needs of he economy (Handa, 2009). As a resul, i is exigen for he policy maers o undersand he facors ha deermine his funcion and he exisence of a sable long-run relaionship beween hese facors and he money soc. This pivoal role of he money demand funcion has generaed many empirical researches relaed o he money demand funcion, including is long-run and shor-run sabiliy. Despie he fac ha here is a grea deal of sudies on he money demand in boh developed and developing counries, no in-deph sudy, o our nowledge, has been repored ye on his subjec for Nepal. Taing his fac ino consideraion, his paper ries o fill he gap in he lieraure by esimaing he money demand funcion for Nepal. The main objecive of his paper is o esimae a heoreically consisen model of he money demand funcion of Nepal for he period of FY 1997/98 o FY 2009/10 using annual daa. Two differen definiions of money balances have been employed in he sudy: narrow money (M1) which includes currency and demand deposis, and broad money (M2) which includes M1 and ime deposis. For he model esimaion, his paper * Assisan Manager, Nepal Rasra Ban, Nepalgunj, Nepal. bc_birendra@homail.com. T# Remars: The views expressed in his paper are hose of he auhor and do no necessarily represen hose of he Nepal Rasra Ban. The auhor would lie o han Dr. Bamdev Sigdel, Direcor, Nepal Rasra Ban, and all he saff a he Research Uni, Nepal Rasra Ban Nepalgunj, for heir encouragemen and coninuous suppor in preparing his paper.

2 An Empirical Analyais of Money Demand Funcion in Nepal 55 employed he mehod of coinegraion, error correcion model (ECM) and Dynamic ordinary leas squares (DOLS). The curren moneary policy framewor of Nepal has aen broad money as he inerim arge of he moneary policy (NRB, 2010). Furhermore, several noiceable changes have been occurred in he Nepalese financial sysem and he economy as a whole afer he implemenaion of he financial secor reform program. In his ype of economic milieu, his sudy bears significance for policy maers, especially Nepal Rasra Ban in is fuure policy maing. The brief ouline of his paper is as follows. Secion II deals wih mehodology, which includes models, daa feaures and model esimaion echnique. Secion III presens he empirical resuls of he sudy and analysis of he resuls. Secion IV includes he conclusion and provides he policy implicaions of he findings. II. MODELS AND DATA Daa and Their Feaures This paper uses annual daa of Nepal over he period of FY 1997/98 o FY 2009/10 for empirical analysis. Daa include broad money (M1), narrow money (M2), he urban consumer price index (CPI), nominal GDP, he ineres rae and real GDP. The sources of daa include Nepal Rasra Ban s annual repors and quarerly economic bulleins, and various issues of Economic Survey of Minisry of Finance, Nepal. The daa source of M1 and M2 are he various issues of he Quarerly Economic Bullein of NRB. The ineres rae a he savings deposi a commercial bans has been used as he ineres rae for he empirical analysis. 1 Due o he unavailabiliy of he daa on weighage rae, his sudy uilized he ineres rae calculaed by aing he average of he upper and lower limi of he srucure of he ineres rae for each year. Daa on ineres rae were obained from he various issues of Quarerly Economic Bullein of NRB bullein. Daa on GDP and CPI were obained from he various issues of Economic Survey of Minisry of Finance (MOF), Nepal. The nominal values were deflaed by using CPI in order o compue he real values. Logarihm values are used for money supply, price levels and oupu (GDP). Ineres raes are analyzed in wo ways, aing a logarihm in one case and no in he oher. 1 Near money asses such as savings deposis in commercial bans proved o be he closes subsiues for M1, so ha heir rae of reurn seems o be he mos appropriae variable for he cos of using M1. Bu, if he broader definiion of money were used, he ineres rae on mediumerm or long-erm bonds would become mos appropriae (he alernaive o holding M2 or M3 is longer erm bonds), since savings componens of he broad definiion of money hemselves earn an ineres rae close o he shor rae of ineres (Handa, 2009).

3 56 ECONOMIC REVIEW Economeric Models There are various heories concerning he money demand funcion. There is generally a consensus among he money demand heories ha he main deerminans of he quaniy of money demand are he scale variable, which can be real income, wealh, or permanen income and opporuniy cos variables. For example, Kimbrough (1986a, 1986b) and Faig (1988) came up wih he following money demand funcion as a resul of explicily considering ransacion coss: M = L( Y, R ) P LY >0,L <0..(1) R In his funcion, M represens nominal money supply for period ; P represens he price index for period ; Y represens oupu for period ; and R represens he nominal ineres rae for period. Increase in oupu lead o increase in money demand, and increase in ineres raes lead o decreases in money demand. The funcion L is assumed o be increasing in Y, decreasing in hose elemens of R represening raes of reurn on alernaive asses, and increasing in raes of reurn associaed wih asses included in M. Income, GDP in his model, is he choice of he scale variable because of he daa limiaion on wealh. The opporuniy cos of holding money i.e he rae of ineres is he second independen variable ha deermines he money demand funcion. Therefore, he proposed money demand funcion for Nepal specified in a log-linear form corresponding o equaion (1) in order o conduc an empirical analysis are: Model 1: Model 2: ln( M ) β R + µ, β 1 >0, β 2 <0..(2) ln( P ) = 0 + β1 ln( Y ) + β 2 ln( M ) ln( P ) = β 0 + β1 ln( Y ) + β 2 ln( R ) + µ, β 1 >0, β 2 <0..(3) Where, β 1 in boh equaions (2) and (3) is he income elasiciy of money demand, bu β 2 in (2) is he semi-elasiciy of money demand wih respec o ineres rae and in (3) is he elasiciy of money demand wih respec o ineres rae. 2 The posiive sign is expeced for income coefficien, while he domesic ineres rae coefficien is expeced o be negaive. Boh models 1 and 2 are log linear models, bu Model 1 uses he level of ineres raes and model 2 uses he logarihm value of ineres raes. Model 1 is convenional form of he money demand funcion mosly used in he empirical research whereas Model 2 is based on he invenory-heoreic approach o money demand pioneered by Allias (1947), Baumel (1952) and Tobin (1956). 2 Semi-elasiciy used in model 1 shows by how much percen real money demand change in response o a change in he ineres rae of 1 percenage poin (ha is, for insance, he rae rising from 5% o 6%). I can be defined as: β 2 d M = ln = dr dm M dr The elasiciy (such as β 1 in model 1 and β 1 and β 2 in model 2) shows how he demand for money changes in response o a given percenage poin change in he ineres rae (say a one percenage poin change from 5.0% o 5.05%).

4 An Empirical Analyais of Money Demand Funcion in Nepal 57 Model Esimaion As a preliminary analysis, he augmened Dicey-Fuller es is carried ou for he logs of real money balances, GDP, and ineres raes (Dicey and Fuller, 1979). ADF es is one of he uni roo ess o deermine wheher each daa series is non-saionary (ha is uni roo exis) or saionary (uni roo do no exis). 3 This es forms he preamble o he economeric analysis of long-run equilibrium proposed by economic heory. Saionariy of he series is a desirable propery for an esimaed model. A sochasic process is said o be saionary if is mean and variance are consan over ime and he value of he covariance beween he wo ime periods depends only on he disance or gap or lag beween he wo ime periods and no he acual ime a which he covariance is compued (Gujarai, 2007). Then, his wor uses he widely used mehod of coinegraion and error correcion echnique in he framewor of he linear mulivariae vecor auoregressive (VAR). A esing procedure suggesed by Johansen and Juselius (1990) is conduced o examine possible coinegraion among he variables. The Johansen echnique provided maximum-lielihood esimaes for esing more han one coinegraing vecor in a se of ime series. This echnique is se o accoun for long-run properies as well as shor-run dynamics, in he framewor of mulivariae vecor auoregressive models (Alsahafi, 2009). If he variables ha have uni roos are coinegraed, i is appropriae o esimae Vecor Error Correcion Models (VECM). The VECMs are designed for use wih nonsaionary series ha are coinegraed (Hamilon, 1994). Therefore, his sudy maes use of VEC models o analyze he money demand. 4 The VECM approach has he 3 An Augmened Dicey-Fuller es can be specified as: p y = µ + η + γ y + 1 * 1 φ j y j + ε j= 1 Where y is a random variable possibly wih non zero mean, µ is a consan, ime rend and ε is a error correcion erm wih zero mean and a consan variance. The null hypohesis of he * uni roo ( γ =1) is esed agains he alernaive of saionariy using criical values provided by MacKinnon disribuion (Greene, 2002). 4 The VECM represenaion is : y = 1 Γi y 1 i= 1 + Πy + µ + ΨD + ε 1 Where, Γi y 1 and Π y are vecor auoregressive (VAR) componens in he firs i= 1 differences and error correcion componens in level, respecively. Y is px1 vecor of he variables ha are inegraed of he same order. µ is a px1 vecor of consans. K is a lag srucure, while ε is px1 saionary random process wih zero mean and consan variance. Γ i is a p χ p marix ha represens shor-run adjusmens among variables. Π is decomposed ino βα, where β is an r χ p marix of coinegraing vecors and α is an p χ r marix of he speed of adjusmens (Alsahifi, 2009).

5 58 ECONOMIC REVIEW advanage of joinly esimaing he long- and shor-run componens of he demand for money, hus faciliaing he as of ensuring ha shor-run specificaions are associaed wih long-run componens wih esablished economic heory (Alsahafi, 2009). Finally, he Dynamic OLS (DOLS) developed by Soc and Wason (1993) has been employed in order o esimae he coefficiens of he boh models (2) and (3). III. EMPIRICAL RESULTS AND ANALYSIS Growh and Velociy of Moneary Aggregaes One of he simple approaches for analyzing he relaionship beween money and he economy is o examine heir graphical relaionship and properies. The figures presened show simply he annual growh raes of he money balances and GDP, and income velociy of he money (VM) balances boh in nominal and real erms. Figure 1 shows he hisorical developmen of he nominal money balances (M1 &M2) and GDP of Nepal during FY 1975/76 o FY 2009/10. Figure 1: Growh rae of Nominal M1, M2 & GDP( FY 1975/76 o FY 2009/10) Annual Growh rae of M1, M2 & GDP( In %) Fiscal Year growh rae of nominal M1' Growh rae of Nominal GDP' Growh rae of nominal M2

6 An Empirical Analyais of Money Demand Funcion in Nepal 59. Figure 2: Growh rae of real M1, M2 & GDP(FY 1975/76 o FY 2009/10) Annual Growh rae of real M1,M2 & GDP( In %) Fiscal Year Growh Rae of real M1 growh rae of real GDP' growh rae of real M2 Similarly, Figure 2 shows he growh of he real GDP, M1 and M2. The figures clearly show ha he changes in he nominal money balances have been closely associaed wih he changes in economic aciviy as represened by GDP. During he sudy period, he growh raes in boh nominal (Figure 1) and real erms (Figure 2) money supply have also flucuaed and nearly capured he major up (for example, FY 1985/86) and downswings (for example, FY 1995/96) of he economy. The income velociy of money (VM) also plays he imporan role in ensuring he effeciveness of he moneary policy since when VM is unpredicable, money demand funcion is also unsable. 5 Figure 3 and 4 show he income velociy of money for boh M1 and M2 of Nepal for he period of FY 1975/76 o FY 2009/10. The velociy of boh M1 and M2 in nominal and real erms has been declining gradually, bu along wih he flucuaion in few years. The velociy of he M1 has been more flucuaing han he M2 in he sudy period. The velociy of M1 became more sable afer FY 1999/00 whereas i flucuaed coninuously before his ime period. I is clear from he figures presened above ha he VM of M2 seem o be relaively more sable han he VM of M1 in boh real and nominal erms. 5 VM is defined as he average number of imes ha a naional currency is spen in a year. Hence, i can be defined as he raio of GDP o money supply i.e and M sands for money supply. VM = Y M, where Y sands for GDP

7 60 ECONOMIC REVIEW. Velociy of M1 & M Figure 3: Velociy of Nominal M1 and M2(FY 1975/76 o 2009/10) Fiscal Year velociy of nominal M1 velociy of nominal M2 Velociy of real M1 & M Figure 4: Velociy of real M1 and M2(FY 1975/76 o FY 2009/10) Fiscal Year velociy of real M1 velociy of real M2 The declining value of VM of Nepal implies he increase in he degree of he moneizaion of he economy. Furhermore, he decline in VM has parly offse he inflaionary poenial of he growh in he money supply. Uni Roo Tes Resuls Before embaring upon he coinegraion analysis, he ime series properies of he variables need o be examined. For his, his sudy maes use of he Augmened Dicey- Fuller (ADF) es in a regression wih a drif, bu no rend o analyze he ime series

8 An Empirical Analyais of Money Demand Funcion in Nepal 61 properies of he daa. Table 1 presens he esimaed es saisics for all variables on he level and firs difference using ADF. Table 1: Augmened Dicey-Fuller (ADF) Tes Variables Augmened Dicey Fuller(ADF) Levels Firs Difference LRGDP ** LRM * LRM * LP ** R Noes: 1.LRGDP, LRM1, LRM2, LP and R are he log of he real GDP, real M1, real M2, CPI and nominal ineres rae respecively. 2. */** sand for significan a 1% and 5%, respecively. 3. Criical values were used of Macinnon (1991). The criical values are and a he 1% and 5% level of significance respecively. The saisics under ADF ess implies ha all levels of he naural logarihms of he menioned ime series variables have uni roos a he 5 per cen level of significance. Furhermore, he ADF saisics show ha he uni roo hypoheses are rejeced a he 5 per cen level of significance for he firs difference of he naural logarihms of he variables excep R. As a resul, he ADF es conduced implies ha he level of each variable was found o have a uni roo, whereas he firs difference of each variable was found no o have a uni roo excep R. Thus, all variables are found o be non-saionary a levels and saionary a heir firs differences excep R. All variables are bes modeled as I (1) wih drif. Since almos variables are inegraed of he order I(1) wih drif, hen one can expec ha hese series may be coinegraed as well. Coinegraion Analysis This paper has used he Johansen and Juselius(1990) mehodology o es he presence of a sable long-run relaionship beween real money balances and heir deerminans. Johansen and Juselius (1990) use boh race eigenvalue saisics(λ race ) and maximum eigenvalue saisics(λ max ) which are employed o deermine he coinegraion vecors. The opimal lag lengh of VAR is deermined by he Sequenial Lielihood Raio (LR), he Final Predicion Error (FPE), and he Aaie Informaion Crierion (AIC). Table 2 presens he resuls of he ran ess of he M1. The resul repored from he race and maximum eigenvalue saisics show ha he null hypohesis of he coinegraing vecor lining real M1 and is deerminans is rejeced a he 5% level of significance for boh λ max and λ race saisics for boh models 1 and 2 since λ max and λ race exceed heir corresponding 5% criical values. In model 1, however, for λ max, he null hypohesis of he, a mos, one coinegraing vecor canno be rejeced a he 5% significance level. I is obvious ha boh saisics yield differen resuls. Bu Johansen and Juselius(1990) sugges he use of λ race saisics in he siuaion of he conflic beween he wo saisics. As a resul, we can conclude ha here exis more han one coinegraing vecors for his M1 a he 5% significance level if we consider model 1. In model 2, he null hypohesis

9 62 ECONOMIC REVIEW of he zero coinegraion is srongly rejeced by he daa a he 5% level of significance for boh λ max and λ race. However, he null hypohesis of he, a mos, one coinegraing vecor canno be rejeced as boh race and maximum eigenvalue saisics are smaller han he criical values repored for each. Thus, i can be concluded ha here exiss a unique coinegraing vecor for he model 2 for M1 a he 5% level of significance. Model Table 2: Johansen Coinegraion Tess for Models (1 &2), M1 Hypohesized no of CE(S) Tes Saisics 5% Criical Values H 0 Eigenvalue λ max λ race λ max λ race Model * 40.94* A mos * A mos * 6.12* * * A mos Model 2 A mos Noe:*denoes he rejecion of he null hypohesis a he 5% significance level. Criical values are from Source: Oserwald-Lenum(1992). Table 3: Johansen Coinegraion Tess for Models (1 &2), M2 Hypohesized Model no of CE(S) Tes saisics 5% Criical Values H 0 Eigenvalue λ max λ race λ max λ race * * A mos Model 1 A mos * * A mos Model 2 A mos Table 3 shows he ran es for M2 for boh hypohesized models 1 and 2. From he repored resul, he null hypohesis of he zero coinegraing vecors is srongly rejeced by he daa in boh models a he 5% significance level for boh λ max and λ rac as boh λ max and λ rac are greaer han heir corresponding 5% criical values. Bu, he null hypohesis of he, a mos, one coinegraing vecor is no rejeced by boh race and maximum eigenvalue saisics because boh race and maximum eigen value saisics are smaller han he criical values repored for each. Therefore, we can conclude ha here is a unique coinegraing vecor for M2 a he 5% significance level. Vecor Error Correcion Model (VECM) Resuls As he variables in boh models esed above are found o be coinegraed, a beer way o explain he dynamic relaionship beween hem is o use VECM. Thus, his sudy employed he VECM o ie he shor-run behavior of each money demand o is long-run

10 An Empirical Analyais of Money Demand Funcion in Nepal 63 equilibrium values. Under his mehod, he simulaneous effec of all he variables in he model on each oher is esimaed. The shor-run error correcion model for money demand for Model 1 using M1 is given by he equaion (4). Here, -saicic and p-values are in round braces and in squared braces respecively. The coefficiens of boh (oupu and ineres rae) have he expeced sign, and are saisically significan. In he shor-run, boh variables oupu and ineres rae have significan effec on he narrow money (M1). Alhough he error-correcion erm is significanly differen han zero, i does no have he expeced sign. This implies ha he dynamic adjusmen o an excess money supply by economic agens would be hrough increasing heir demand for money, which would cause he dynamic sabiliy in he demand for money. lnrm1 = lnrm lny R EC -1.(4) (0.0308) (0.0184) ( ) (0.492) [2.401] [2.811] [-4.488] [0.251] (0.99) R 2 =0.31 F-sasic=0.658 LM(2)=7.17 LM(5)=5.31 (0.73) (0.81) Pormaneau es(1) Adj Q-sa=10.11 Jarque-Bera Normaliy es=9.79 (0.34) Residual Heerosedasiciy Tes χ2=40.76 (0.098) Equaion (5) repors he shor-run error correcion model for money demand of Model 2 using M1, where -saisic and p-values are in round braces and in squared braces respecively. The esimaed resuls show ha he coefficiens of boh GDP and ineres rae have he sign ha confirm o he money demand heory along wih heir saisical significance. Therefore, he shor-run demand for M1 seems o be influenced by he lags of M1, GDP and ineres rae. Bu, he error-correcion erm does no have expeced sign. lnrm1 = lnrm lny lnr EC -1.(5) (0.524) (0.0318) (0.0161) (0.595) [2.393] [2.691] [-4.193] [0.0042] R 2 =0.33 F-sasic=0.983 LM(2)=4.685 LM(5)=5.44 (0.86) (0.79) Pormaneau es(1) Adj Q-sa=6.867 Jarque-Bera Normaliy es=9.695 (0.65) (0.99) Residual Heerosedasiciy Tes χ2=39.5 (0.118)

11 64 ECONOMIC REVIEW Equaion (6) is he error correcion model of money demand for Model 1 using M2. The resuls obained here are also very similar o hose of M1. Here, boh oupu and ineres rae coefficiens are saisically significan and have he expeced sign. The coefficien of he GDP is posiive i.e 0.04 and he coefficien of he ineres rae is negaive i.e This implies ha he shor-run demand for M2 is also influenced by lag of he boh oupu and ineres rae. lnrm2 = lnrm lny R EC -1.(6) (0.0258) (0.0081) (0.312) (0.2645) [3.429] [5.321] [-4.293] [2.03] R 2 =0.35 F-sasic=2.049 LM(2)=7.96 LM(5)=8.62 (0.54) (0.473) Pormaneau es(1) Adj Q-sa=5.69 Jarque-Bera Normaliy es=10.22 (0.72) (0.99) Residual Heerosedasiciy Tes χ2=40.46 (0.096) lnrm2 = lnrm lny lnr EC -1.(7) (0.0259) (0.0082) (0.0629) (0.278) [3.413] [5.289] [-3.966] [2.019] R 2 =0.361 F-sasic=2.036 LM(2)=5.67 LM(5)=10.24 (0.73) (0.33) Pormaneau es(1) Adj Q-sa=6.22 Jarque-Bera Normaliy es=9.795 (0.72) (0.99) Residual Heerosedasiciy Tes χ2=36.28 (0.198) From VECM, he esimaed money demand funcion for M2 for Model 2 is given by equaion (7). This equaion also repors he same resuls as earlier. The coefficiens of he oupu and ineres are boh saisically significan and have he expeced sign. Bu he error correcion erm does no have he expeced sign. Dynamic OLS Resuls As he exisence of he coinegraion relaion was suppored by he Johansen coinegraion ess, he money demand funcion can be esimaed by using he Dynamic OLS (Soc and Wason, 1993). The Soc-Wason approach is a robus single equaion approach which correcs for regressor endogeneiy by he inclusion of he leads and lags

12 An Empirical Analyais of Money Demand Funcion in Nepal 65 of firs differences of he regressors, and for serially correlaed errors by GLS procedures. 6 In his sudy, he number of leads and lags is chosen arbirarily o be 1. Table 4 presens he esimaion resuls obained from DOLS wih respec o Model 1 for M1. From his able, i is clearly observed ha he coefficien of GDP is significanly esimaed o be posiive i.e 2.19 and he ineres rae coefficien is esimaed o be negaive i.e Table 4: Dynamic OLS (M1, Model 1) ln 1 ln P = β 0 + β1 ln y + β 2 M R + γ ln y + γ R + u Variables Coefficiens SE -sasic P-value R 2 Consan lny yi lnr Table 5: Dynamic OLS (M1, Model 2) ln 1 ln P = β 0 + β1 ln y + β 2 ln R + γ yi ln y + γ ri M ln R + u Variables Coefficiens SE -sasic P-value R 2 Consan lny lnr Table 5 shows he DOLS resuls of Model 2 for M1. This able also clearly shows ha he oupu coefficien is 2.68 and he ineres rae coefficien is Thus, he sign condiion of he money demand (M1) holds for boh of he cases. In his way, i becomes apparen ha no only he coinegraing relaion was suppored, bu also he exisence of he money demand funcion wih respec o he M1 was saisically suppored. Nex, we ae he money demand funcion using M2 componen. In he case of M2 also, as he exisence of he coinegraion was suppored, he DOLS has been used o esimae he money demand funcion. ri 6 The DOLS esimaors is based on he following augmened coinegraing regression, which includes he pas, presen and fuure values of he change in X, = β 0 + β1x + γ i Y X + u Where represens he leads and lags of he variable, β 0 and β 1 are he parameers needs o be esimaed. Y and X are he coinegraed variables.

13 66 ECONOMIC REVIEW ln 2 ln P = β 0 + β1 ln y + β 2 Table 6: Dynamic OLS (M2, Model 1) M R + γ ln y + γ R + u Variables Coefficiens SE -sasic P-value R 2 Consan lny yi lnr Table 7: Dynamic OLS (M2, Model 2) ln 2 ln P = β 0 + β1 ln y + β 2 ln R + γ yi ln y + γ ri M ln R + u Variables Coefficiens SE -sasic P-value R 2 Consan lny lnr Table 6 shows he esimaion resuls of Model 1 for M2. I is eviden from he able ha he sign of he oupu coefficien is posiive (2.73) and he sign of he ineres rae coefficien is significanly negaive ( ). Table 7 shows he DOLS esimaion oucome for he M2 wih respec o Model 2. In his able also, he oupu coefficien was significanly esimaed posiive values of and he ineres rae coefficien was significanly esimaed negaive values of From his we can conclude ha along wih he exisence of he coinegraion relaion, he exisence of he money demand funcion wih respec o M2 was saisically suppored. IV. CONCLUDING REMARKS This paper empirically analyzed he money demand funcion for Nepal using annual daa for he period of FY 1997/98 o FY 2009/10. The empirical resuls obained from he coinegraion analysis indicae ha he real money balances M1 and M2 are coinegraed wih he oupu as represened by GDP and ineres rae, implying ha a long-run relaionship beween he real moneary aggregaes and independen variables (GDP and R) is esablished. The VECM S were employed o show he shor-run dynamic relaionship among moneary aggregaes and scale variables. The coinegraion and error correcion resuls clearly show ha here exis a long-run and shor-run dynamic equilibrium beween moneary aggregaes (M1 & M2) and scale variables, GDP and ineres rae. Furhermore, esimaed resuls from Dynamic OLS also implied he saisical suppor for he exisence of he money demand funcion wih respec o boh M1 and M2 under boh ri

14 An Empirical Analyais of Money Demand Funcion in Nepal 67 models. The declining value of velociy of money observed in his paper clearly reflecs he growing moneizaion of he economy. As velociy of M2 was observed relaively sable han M1, i simply indicaes he superioriy of broad money over narrow money for policy purpose. These derived empirical resuls from his paper imply ha NRB can focus on boh M1 and M2 conrol in order o achieve hese goals. Fuure research on he money demand funcion may include differen ineres raes from money mare in explaining money demand in shor and long-run. In addiion, he sabiliy of he Nepal s money demand funcion, aing ino accoun he currency subsiuion issue, may be he anoher issue for he researchers in order o sugges he ways for he effecive formulaion and implemenaion of he moneary policy of Nepal.

15 68 ECONOMIC REVIEW REFERENCES Alsahafi, Mamdoon Linear and Non-linear Techniques for Esimaing The Money Demand Funcion: The Case of Saudi Arabia. Unpublished PhD disseraion, Universiy of Kansas, USA. Aa-mensah, Joseph Money demand & Economic Uncerainy. Woring Paper , Moneary and Financial Analysis Deparmen, Ban of Canada, Onario, Canada. Bae, Youngsoo & Kaav, Vias Money Demand in Japan & he Liquidiy Trap. Woring Paper 04-06, Deparmen of Economics, Ohio Sae Universiy, USA. Bele, Ansgar & Pollei, T Moneary Economics in Globalised Financial Mares. Berlin: Springer-Verlag, Germany. Chuwu, O. J, Agu, Cleus C & Onah, Felix E Coinegraion & Srucural Breas in Nigerian Long-Run Money Demand Funcion. Inernaional Research Journal of Finance & Economics, Issue 38, Available a Inerne address: hp// Enders, Waler Applied Economeric Time Series. John Wiley and Sons.Inc Greene, William H Economeric Analysis. New Yor: Pearson Educaion Ld. Gugarii, Damador Basic Economerics. New Delhi: Taa McGraw-Hill Publishing Company Limied, Hamilon, James D Time Series Analysis. Princeon: Princeon Universiy Press, USA. Handa, Jagadish Moneary Economics. New Yor: Rouledge Taylor & Francis Group. Haug, Alfred A Canadian Money Demand Funcion: Coinegraion Ran Sabiliy. Woring Paper, Deparmen of Economics, Yor Universiy, Torono, Canada, Available a he Inerne address: hp://dep.con.yor.ca/research /woringpapers/woring_papers/moneyca.pdf Hayashi, Fumio Economerics. Princeon: Princeon Universiy Press, USA. Ho, W.S Money Demand in Macao & is Esimaion. Woring paper, Moneary Auhoriy of Macao, Macao. Ihide, Sylvanus Esimaing he Demand for Money in Nambia. Occasional Paper, Ban of Nambia, Windhoe, Nambia. Kollarova, Ing. Vera The Esimaion of Money Demand in he Slova Republic. Woring paper-09, Ban of Slovaia, Slova Republic, Available a Inerne address: hp:// Minisry of Finance Economic Survey 2009/10. Kahmandu: Minisry of Finance, Governmen of Nepal, Nepal.

16 An Empirical Analyais of Money Demand Funcion in Nepal 69 Nepal Rasra Ban Moneary Policy for Fiscal Year 2010/11. Kahmandu: Nepal Rasra Ban, Nepal Quarerly Economic Bullein, Volume XXXVII, Number 3. Kahmandu: Nepal Rasra Ban, Nepal Quarerly Economic Bullein, Volume 46, Number 3 & 4. Kahmandu: Nepal Rasra Ban, Nepal. Osooee, Mohsen B.& Wang Y How Sable is he Demand for Money in China? Journal of Economic Developmen, vol 32, Number 1, Universiy of Wisconsin, Milwauie, USA. Oserwald-Lenum, M A noe wih quiniles of he asympoic disribuion of he maximum Lielihood coinegraion ran es saisics: four cases. Oxford Bullein of Economics & Saisics, 54. Shigeyui, Hamori Empirical Analysis of Money Demand Funcion in Sub- Sahara Africa. Economic Bullein, Vol 15. Kobe Universiy, Available a Inerne Address: hp:// 15/EB a.pdf Shin, Sungwon Sabiliy of he Demand for Money in Korea. Inernaional Economic Journal, Vol 16, Universiy of Wisconsin, USA. Sovannroeun, Samreh Esimaing Money Demand Funcion in Combodia: ARDL Approach, Munich Personal RepEc Archive Paper No 17274, 2009, Available a Inerne address:hp://mpra.ub.uni-muenchen.de/ Soc, James H. & Wason, Mar W Inroducion o Economerics. New Yor: Pearson Educaion.Inc, Publishing as Addison Wisley Series. Wash, Carl E Moneary Theory & Policy. Massachuses: The MIT Press, USA. Yu, Han & Gan, Pei-Tha An Empirical Analysis of he Money Demand Funcion in ASEAN-5. Inernaional Research Journal of Finance & Economics. Issue 33, Available a he Inerne address: hp://

17 70 ECONOMIC REVIEW Appendix 1: Daa se used in he esimaion of he Money Demand Funcion Fiscal Year GDP (Rs. In millions) CPI M1 Mid july (Rs. In millions) M2 Mid july (Rs. In millions) R Mid july 1975/ / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / GDP: Gross Domesic Produc, CPI: Urban Consumer Price Index, M1: Narrow Money, M2: Broad Money, R: Ineres rae srucure of he Commercial Bans a Savings Deposi.