Inernaional Quarerly Financial Journal Crisis of Quaniaive and Money Economics, Demand in Winer Iran 2012, 8(4): 127-147 127 Inernaional Financial Crisis and Money Demand in Iran: ARDL Approach Moslem Alboosoveilem(M.Sc.) and Hossein Karimi(Ph.D.) Received: 2011/9/18 Acceped: 2012/3/8 Absrac: Becoming pervasive he inernaional crisis in he world and heir affecs leads o exreme researches which sudy he affecs of hese crisis on he imporan economic variables. One of he variables is money demand. Inernaional crisis influences on money marke via change in exchange and ineres rae, i causes o insabiliy in real income and inflaion rae and influences inernal economic condiions. This paper has invesigaed he money demand for Iran economy in 1974-2008 and he affecs of financial crisis on money demand and is changes on real income, inflaion and exchange rae. Also his sudies he rend of changes in money demand sabiliy and he oher variables via Auo Regressive Disribued Lag (ARDL) and Error Correcion Mehod (ECM) in dynamic analysis. The coefficien of Gross Domesic Producion (GDP) is posiive and significanly more han 1. Inflaion rae variable is negaive expecedly. The significan posiive coefficien relaed o he las financial crisis demonsraes ha he money demand will increase. Of course, he shorrun affec is negligible and is affecs become obvious in long-run. JEL classificaion: E41, C32, C22 Keywords: money demand, inernaional financial crisis, Auo Regressive Disribued Lag (ARDL), Error Correcion Mehod * Ph.D. suden of economics a Shahid Bahonar Universiy of Kerman and assisan professor of economics a Isfahan Universiy, respecively, Iran. Email: (moslem_albu@yahoo.com)
128 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 1. Inroducion Knowledge and recogniion abou he facors affecing money demand can help he policy makers o making decision more convenien. Price sabiliy, economic growh, full employmen, bank ineres rae, exchange rae and balance in paymen are he mos imporan goals in money policies decision making. Cenral bank in every counry as a money policy maker performs he menioned policies. Accurae recogniion of money demand funcion is essenial for general macroeconomic procedure, economic policy making and macroeconomic policies effeciveness. Money demand funcion as an imporan variable in major macroeconomic heories is always subjeced o discuss. The finding resuls in developed counries demonsrae ha money demand is funcion of he kinds of ineres raes, income and exchange rae, bu in developing ones, due o inconvenien financial marke, inflaion rae beside real income, exchange and ineres rae play role as deerminan facors of money demand. Financial crisis affec money marke via exchange and ineres rae. In his paper, he affecs of wo recen grea crises on money demand funcion in Iran are sudied. One is he grea Eas-Souh Asia crisis in las 1990s which iniiaed via foreign invesmen markes and he oher is curren crisis which overcomes he world economy. Since Iran economy is relian on oil income, financial crises can affec money demand by decreasing oil price and finally change in naional income. Also recen crisis can influence money demand via change in ineres rae and making budge defici leading inflaion. One of he ways of financial crisis effeciveness on he counries economy is exchange rae (he linkage beween counry economy and world economy). In his sudy, i has been ried o analyze he affec of recen crises on he money demand in Iran via new brand economeric mehods. 2. Theoreical base 2.1. inernaional financial crisis Generally, financial crisis is he posiion ha he currency value decreases or he inernaional reserves diminish severely or boh of hem. Financial crisis can be led of he disurbance in a
Inernaional Financial Crisis and Money Demand in Iran 129 financial secor and i ransfers o he oher pars due o he linkage beween financial and real economic secor. Porfolio marke crisis, balance paymen, bank and insurance company bankrupcy, severe changes in exchange rae and decrease in currency value are some crises in his financial field. Perhaps i would be beer o define financial crisis as a self-fulfilling or communicaive disurbance in financial sysem funcion. Probably he resource of self- fulfilling disurbance would be he imperfecion in counry economic basics. In done researches have ofen menioned money and bank crises or he combinaion of boh as a financial crisis. Also in some ohers, decreasing rae he currency value is considered as a deerminan index of financial crisis. In his condiion, economic agens occasionally ransform heir asses from currency ino oher forms and i leads o make financial crisis. Governmen increases money supply via prining money or ax rae for eliminaing he budge defici and financing iself or decreases he currency value via expanding he expor, so invesors ry o ransform heir pecuniary possession from currency ino oher forms. In he oher side, decrease in currency value is made by marke pressure for increasing in exchange rae, because counry can' or don' wan olerae he warrany cos of is currency. In a case, cenral bank has o sale curren money insead of foreign reserves for keeping below ceiling exchange rae. If cenral bank misses all foreign reserves, he governmen mus allow floaing he exchange rae (decreasing he money value). So he inernal goods and services become cheaper han he exernal ones. Decreasing he money value led of speculaive expansion, can be he reason of inflaion and failure in inernal and exernal financial marke. In given circumsances, his crisis moves o oher financial references due o he relaion beween markes. One example of a currency crisis occurred in Russia in 1998 and led o he devaluaion of he ruble and he defaul on public and privae deb. Currency crises such as Russia s are ofen hough o emerge from a variey of economic condiions, such as large deficis and low foreign reserves. They someimes appear o
130 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 be riggered by similar crises nearby, alhough he spillover from hese conagious crises does no infec all neighboring economies- only hose vulnerable o a crisis hemselves. World economy has ever experienced differen crises which have influenced economic indexes such as money demand. In his paper, i is represened he affecs of wo recen grea crises on he money demand funcion. One is he grea Eas-Souh Asia crisis in las 1990s which iniiaed via foreign invesmen markes and he oher is curren crisis which begun since 2006 by exploding he selemen bubble in America and on 2007 his ransferred o financial secor and grew severely and hen climaxed by bank bankrupcy in financial markes and now has pervaded o real economic secor. Insabiliy in exchange marke is led of crisis enrance. By falling he exchange marke and disrus o i, consumers have decreased heir expendiures for purchasing consan goods especially auomobile and selemen and also loaning process has became less and consumers and invesors have missed heir hope (i means if hey loan, hey can' repay i.). So auomobile and selemen which are consan and sensiive o ineres rae, have involved furher by crisis. Fear and disrus o credi markes (from boh invesor side and banks side) and marke sagnaion from banks side and pervading he crisis ino credis marke cause o less efficiency of money policies in order o moivae he demand. Allen N.Berger and Chrisa H.S. Bouwman have sudied he behavior of bank liquidiy creaion around five financial crises in he U.S. from 1984 o 2008 by using a recenly-developed comprehensive measure of aggregae liquidiy creaion by banks. They hen examine he effec of bank capial on a bank s compeiive posiion, profiabiliy, and sock reurn performance around hese crises. We also creae wo fake crises o explore bank behavior in normal imes. They have reached five conclusions based on our analysis of he behavior of bank liquidiy creaion around financial crises.
Inernaional Financial Crisis and Money Demand in Iran 131 - Firs, here seems o have been a significan build-up or drop-off of abnormal liquidiy creaion before each crisis, where abnormal is defined relaive o a ime rend and seasonal facors. - Second, banking and marke-relaed crises differ in wo imporan ways. The banking crises were preceded by posiive abnormal liquidiy creaion by banks, while he marke-relaed crises were generally preceded by negaive abnormal liquidiy creaion. In addiion, he crises hemselves seemed o aler he rajecory of aggregae liquidiy creaion during banking crises bu no during marke-relaed crises. - Third, liquidiy creaion has boh decreased during crises (e.g., he 1990-1992 credi crunches) and increased during crises (e.g., he 1998 Russian deb crisis / LTCM bailou). Thus, liquidiy creaion likely boh exacerbaed and amelioraed he effecs of crises. - Fourh, off-balance shee illiquid guaranees (primarily loan commimens) moved more han semi-liquid asses (primarily morgages) and illiquid asses (primarily business loans) during banking crises. - Fifh, because he subprime lending crisis was preceded by a dramaic build-up of posiive abnormal liquidiy creaion, heir analysis hins a he possibiliy ha while financial fragiliy may be needed o creae liquidiy, oo much liquidiy creaion may also lead o financial fragiliy. 2.2. Money demand In he economic lieraure, he success of moneary policy, sable money demand funcion is crucial. And any insabiliy in he demand for money can be a problem for moneary policy (Hamori and Tokihisa, 2001), (Payne,2003). Demand for money arises from medium of exchange and sore of value funcions. Because people need money o smooh ransacions, hey hold i for fuure needs. Money as a medium of exchange is a faciliaor of ransacions and hence an essenial lubrican o he mechanism of exchange. In fac hese wo roles of money are inerrelaed. Unless money is a sore of value, i canno be a medium of
132 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 exchange and vice versa. However, he ransacion demand is more fundamenal. There are oher asses besides money ha are compeing and even beer sores of value bu no beer medium of exchange. In developing counries, hough, money s sore of value role is paricularly significan. Money generally serves as he uni of accoun and he sandard of deferred paymen because i is convenien as well as efficien. However, he medium-ofaccoun role is no logically ied o he medium of exchange. Keynes in The General Theory of Employmen, Ineres, and Money (1936) idenified hree moives for holding money: he ransacion moive, he precauionary moive, and he speculaive moive. The ransacion moive and precauionary moive relae o money s role as he medium of exchange, whereas he speculaive moive relaes o money s role as a sore of value. The ransacion moive arises for exchanging money for goods and services, as i is exremely unlikely o have double coincidence of wans, especially in a modern economy. I may no be possible for me o exchange a few pages of my research paper for a meal in a resauran because my wan and he wan of he resauran owner need no coincide. Holding money involves a rade-off beween forgoing he ineres ha can accumulae wih savings and bearing he inconvenience of no holding money for ransacion purposes. People may hold money o mee fuure paymens, which are uncerain; his is he precauionary moive for holding money. Money is also held for speculaive purposes, ha is, o avoid he risk inheren in oher asses, which may pay higher reurns. Demand for money varies beween developed and developing counries because he former have relaively advanced financial sysems, saes of echnology, and degrees of enforceabiliy of conracs. The volume of ransacions also influences demand for money. In less developed counries cash is used more ofen for ransacions; in more developed naions he use of credi cards reduces he demand for cash. There are several economic variables ha affec he demand for money, including gross domesic produc (GDP), ineres
Inernaional Financial Crisis and Money Demand in Iran 133 raes, inflaion raes, financial innovaions in he economy, degree of moneizaion in he economy, exchange raes, srucure and level of exernal rade, and so on. Various heories explain he relaionships beween hese variables and money. The original quaniy heory of money was followed by he Keynesian heory of liquidiy preference and laer by more modern varians of boh. The Keynsian approach makes ineres rae an explici deerminan of he demand for money. Alhough money demand funcion have ever been sudied in various aspecs, bu all of hem believe ha real opimum money volume has reverse relaion o oupu rae of asses and direc relaion o income. Bu, pracically he paerns are differen in applying he opporuniy cos variable and scale. Since inroducion of coinegraion echnique in 1987 by Engle and Granger (1987) or in 1988 by Johansen (1988) many auhors have re-invesigaed he long-run relaionship beween he demand for money and is deerminans. Examples are Buch (2001) for Poland, Hamori and Tokihisa (2001) for Japan; Bahmani-Oskooee and Wang (2007) for china and Bahmani- Oskooee and Chi Wing Ng (2002) for Iran, Payne (2003) for he Croaia. 3. Lieraure review Many researches have ye been done in evaluaing he money demand funcion in developing and developed counries including Iran. In his par, i is ried o menion some of hem. Bahmani-Oskooee and Wang (2007) have employed CUSUM and CUSUMSQ es in conjuncion wih coinegraion analysis o show ha boh M1 and M2 are coinegraed wih heir deerminans. The resuls of sabiliy ess reveal ha while M1 money demand in China is sable, here is some doub abou sabiliy of M2 money demand. Rao and Kumar (2006) have done an empirical work on he demand for money for Fiji. They have used srucural breaks in he coinegraing equaion, wihin he Gregory and Hansen framework, and found ha here is a coinegraing relaionship
134 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 beween real narrow money, real income and he nominal rae of ineres in all he hree ypes of heir models. Erbykal and e al. (2008) have sudied wheher he currency subsiuion, which occurred in he 1980 s reversed wih he Program for Transiion o a Srong Economy ha was applied afer he February 2001 crisis in Turkey. For deermining dedollarizaion, hey esimaed an M2 money demand funcion using he bound es developed by Pesaran e al. (2001) wih he monhly daa spans from 2001:05 o 2006:12. The es resuls show ha de-dollarizaion has occurred in he Turkish economy in he 2001-2006 period. Bahmani Oskooee and Chi Wing Ng (2002) have examined he long-run demand for money of Hong Kong using he auoregressive disribued lag (ARDL) coinegraion procedure on quarerly daa over he period 1985Q1-1999Q4. Esimaion resuls sugges ha HK$M2 is coinegraed wih is deerminans. In addiion, he CUSUM and CUSUMSQ ess confirm he sabiliy of he money demand funcion. Hafer and Jansen (1991) sudied he money demand in Unied Saes and deermined ha wheher in fac ha here exiss a coinegraion relaionship beween cerain combinaions of real money balances, real income and ineres rae. 4. Model variables inroducion In his par, he model variables which have been used for empirical sudy of money demand funcion in Iran are inroduced. Every economeric model has some variables classifying ino wo groups: dependen and independen. There isn' any deermine definiion for independen variable, real money demand, wihou any ambiguous ha any one accep i. This variable usually go by dividing money volume ( M 1) or currency volume ( M 2 ) on a price index. Abou money demand funcion in Iran, Bahmani Oskoei (1996) and some ohers have used such wo variables. In his sudy, he expanded and adjused definiion of money (M2) included M1 and quasi money is used. Dependen variables of money demand funcion classified ino wo caegories: scale and opporuniy variable. Abou he
Inernaional Financial Crisis and Money Demand in Iran 135 scale variable, income level is applied for indicaing he ransacion volume in economy; herefore, i has an imporan role in he sudy of ransacion heories of money demand. Approximaely all researchers implied ha usage of wealh variable is beer han consan income and consan income is beer han curren income in making sable money demand funcion. Because wealh empirical deerminaion is no feasible, so gross domesic producion (GDP) is subsiued insead of i. According o Iran economy and researches experiences, finally usage of GDP wih consan price seems convenien. And abou opporuniy cos variables, his variable includes ineres and inflaion rae heoreically and pracically. According o Iran economy specificaions, using from suiable proxies such as ineres and inflaion rae as opporuniy cos variables seems persuasive. Also exchange rae can be used as possession and suiable subsiue for inernal money based on researches done on he relaion beween money demand and exchange rae; whereas he opporuniy cos of money keeping is deermined by profi which is led of is rae increase. In he done researches abou evaluaion of money demand funcion for Iran, Bahmani- Oskooee (1995) and he ohers reached o differen resuls relaed o effeciveness exchange rae on he money demand. 5. Theoreical paern In his segmen, an evaluaing model for he money demand funcion and some mehods for esimaion and analysis are proposed. 5.1. Designing he money demand model for Iran Since 1930s various views for money demand are menioned ha formed he heoreical base of empirical sudies in his field. Based on hese heoreical views, several variables deermined money demand funcion for each one. The mos imporan ones include wealh, income, oupu rae of money keeping o keeping oher possessions like bond, consan commodiies and ground. Ofen in he macroeconomic lieraures which discussed money demand, he real money demand is a funcion of macroeconomic
136 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 variables such as real income, profi rae, inflaion rae, and exchange rae and so on. The main work carried ou for Iran are Bahmani-Oskooee (1996).Therefore, based on (in his sudy was based on Bahmani- Oskooee 1996) and he empirical sudies, he money demand funcion is defined as following: LM p j1 LM DD DG 6 0 7 i j q1 j0 Lp 1 j j q2 j0 LY 2 j j q3 j0 LR 3 j j q4 j0 LE 4 j j DU While LM is logarihm of currency adjused by consumer price index, LP logarihm of inflaion rae, LY logarihm of official marke exchange rae, is residual. As you can see, all he variables are in logarihm, he aim of his work is o exrac relevan elasiciy. The exchange rae is considered he official rae announced by he Cenral Bank. Three oher variables are inerred ino his model. DG is dummy variable which is relaed o criical years afer war and is value for inerval 1981 o 1989 is 1 and for ohers is 0. DU is dummy variable for Eas- Souh Asia crisis which is value for inerval 1998 o 2008 is I and for ohers is 0 and he las one is DD which indicaes he recen crisis and is value for 2006 and 2008 is 1 and for ohers is 0. The goal of enering wo las variables ino model is analyzing he affecs of wo menioned crises on he money demand and is relaion wih gross domesic producion, inflaion, profi rae and exchange rae. 5.2. Research mehod If daa would be ime series, various mehods such as coinegraion, ECM and VAR can be applied for long-run and shor-run esimaion. In recen years ARDL among coinegraion mehods expands for esimaing he money demand models. Some coinegraion mehods like Engel and Granger has a lo of limiaions. For example, when he sample quaniy is less, he resuls of his mehod will be biased. Also disribuion of leas square operaors isn' normal. So doing he hypohesis es is no valid by usual saisics. Engel and Granger mehod is based on 5 (1)
Inernaional Financial Crisis and Money Demand in Iran 137 prehypohesis of exisence he coinegraion vecor. Hence, when here are more han one coinegraion mehod, using Engel and Granger leads o inefficiency. Generally, when here is a small sample, mehods like Engel and Granger isn' valid due o no considering he shor-run dynamic reacions beween variables; because is esimaions are biased and doing hypohesis es wih saisics like will no be valid. So in hese condiions, i is beer o use oher mehods like Auo Regressive Disribued Lag (ARDL) which doesn' have hese limiaions. For esimaing he long-run equaion, ARDL has wo seps: - Tesing he exisence long-run relaion beween whole variables in he sudied equaions. (Non auocorrelaion consrains) - Esimaing he long-run and shor-run variables via ARDL and ECM. The following equaion is ECM model: LM LM LP LY LR ale e q 4 i 0 LR i f q 6 i 0 1 DU DD DG b 6 0 7 1 8 LE i 2 q1 i 1 1 LM 3 j 1 c q 2 i 0 4 LP i 1 d 5 q 3 i 0 LY 5.3. ARDL, ECM and VDCF mehods Whereas ARDL (Auo Regressive Disribued Lag) is used for o long-run analysis and ECM (Error Correcion Mehod) for shorrun and VDCF (Variance Decomposiion Funcion) for ineracion affecs analysis in his sudy, so i is essenial o describe hese mehods in his segmen. 5.3.1. ARDL The oal form of ARDL ( LM 0 p j1 LM i j q 1 j0 P, q1,..., q ) is: Lp 1 j j 5DU 6DD 7 DG (3) The opimum lag for each explanaory variable can be earned by k q 2 j0 2 j LY j q 3 j0 LR 3 j 1 i j q 4 j0 (2) 4 j LE j
138 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 AKAIKE informaion or Schwarz Bayesian or Hannan-Quinn 2 crierion or R. MICROFIT sofware provides he possibiliy for esimaing he ARDL mehod wih expanded lag as follow: As an example: Q( L, P) Y k i ( L, qi ) x i1 i w u And 2 Q ( L, P) 1 Q L Q L... Q p L i i 1 2 2 B ( L, q ) 1 L L... L i1 i2 p iq i 1,2,...,k, L is lag operaor, W is a vecor of deerminisic variables like inercepion, rend variable, dummy variables or exogenous variables wih fixed lag. Firs, MICROFIT esimaes he equaion 1 by OLS for possible combinaions P 0,1,2,..., M, q 0,1,2,..., m, i 1,2,... k and for 1 ( m 1) k imes. The maximum lag is deermined by researcher o esimae in ime inerval m 1 o n. In he nex sep, 1 researcher can selec one of he esimaion beween ( m 1) k 2 based on he 4 index including AIC, SBC, HQ or R. In he final sep, MICROFIT calculaes coefficiens of long-run paern and laeral sandard deviaion relaed o long-run coefficiens based on ARDL. 5.3.2. ECM When here is coinegraion beween some variables, i mus be use ECM (Error Correcion Mehod. These mehods in experienced research have expanded inensively. The mos principle reason for using his mehod is o relaing he shor-run vibraions of variables o long-run equilibrium values. When wo variables like X and Y is coinegraed, here is e long-run equilibrium relaion beween hem. Of course here is some nonequilibraion in shor-run. So i can be consider he error erm of following equaion as an equilibrium error. Y X u & u Y X (6) q (4) (5)
Inernaional Financial Crisis and Money Demand in Iran 139 Now we can use his error for linking he shor-run rea of Y o long-run equilibrium value. In order o achieve his, following mehod is used: 2 y 0 1x 2u 1 & IID(0, ) (7) While u 1 is error erm of esimaion he equaion 1 wih one ime lag. This paern is called ECM ha changes in y would be relaed o before period equilibrium error. When x and y, boh of hem are I(1), would be coinegraed, u in equaion 1 will be I(0) ha means i is saionary. Since x and y are saionary, all variables of equaion 2 of ECM are I (0). So i can be esimaed by OLS mehod wihou any worry of geing spurious regression and uilize from and F saisics for hypohesis es. Above paern implied on he wo-sep modeling as followed: Firs sep: he coefficiens of long-run relaions are esimaed via using of saisics relaed o variables level and hen hypohesis zero which is no exisence of coinegraion beween variables, is esed. Therefore, a se of variables are goen which are coinegraed wih ogeher and finally an equilibrium long-run relaion is earned. Second sep: error correcion erm which is such error erm of saic long-run regression is used as an explanaory variable in ECM. Then he dynamic shor-run srucure is deermined by doing necessary ess. ECM coefficien indicaes he adjusmen ino equilibraion. 5.4. Daa The applied daa in his sudy are including: logarihm of adjused currency variable, logarihm of inflaion rae, logarihm of gross domesic producion, logarihm of exchange rae and hree-monh ineres rae. Whole daa are goen from Cenral Bank of Islamic Republic of Iran. The sudy period is 1973-2008. 6. Empirical resuls This sudy has ried o sudy he relaion beween he affecive and imporan variables on money demand and he affecs of financial crises on hem via shor-run and long-run analysis in
140 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 Iran. ARDL for long-run analysis, ECM for shor-run and VDCF for analysis he dynamic ineracion affecs which led of chocks in paern. One of he mos imporan advanages of ARDL is ha his mehod can be used wihou considering being variables saionary or nonsaionary. I means ha i is no necessary o classify variables ino correlaed ones wih degree 0 or 1. The invesigaion he exisence of equilibrium long-run relaion beween money variables is based on ARDL mehod and he analysis of long-run coefficiens are done by his mehod oo. The maximum ime lag is deermined by researcher according o he number of observaions. In his sudy, Schwarz-Bayesian scale (SBC) is uilized for deerminaion he opimum lag. For surveying he long-run relaion beween variables, we can use saisics. In his mehod, hypohesis zero implies on he exisence he long-run relaion beween variables. Because he consrain of endency of shor-run dynamic relaion ino long-run equilibraion is ha sum of coefficiens would be less han 1. For esing his one, digi 1 is subraced from sum of lag coefficiens of dependen variable and divided o sum of such coefficiens deviaion. If would be less han criical value of Banerjee, Dolado & Maser, he hypohesis zero is rejeced and i means ha here is a long-run relaion beween variables. According o earned values for in his paper, hypohesis zero isn' acceped, so here is a long-run relaion beween variables. Whole ables, SC, RESET, NOR and H which indicaes serial correlaion, funcional form of paern,being normaliy and variance scedasiciy, respecively, confirm he esimaed paern. Beside, Ramsey es implicaed paern correcness. The resuls show ha variable affecs inflaion wih a lag period Brqazay money he main reason, perhaps, is ha people are showing less willingness o ask for money due o he devaluaion of heir money. Souh-Eas Asian crisis also conribues o a coninuous variable. However, his effec was no significan.
Inernaional Financial Crisis and Money Demand in Iran 141 Table 1: he findings of esimaion of money demand funcion ARDL (1,0,1,0,0,0,0,1) saisic coefficien variable 17.93* 0.782 LM(-1) 2.6* 0.251 LY -5.29* -0.633 LP 4.84* 0.596 LP(-1) -3.53* -0.395 LR 2.64* 0.032 LE -1.73** -0.024 DD 2.425* 0.043 DU -0.873-0.013 DG -1.665-0.024 DG(-1) -0.617-0.293 C 2 R =0.993 DW=2.09 F=349.8 * Significan a 5% level of confidence ** significan a 10% level of confidence The obained findings of esimaing he long-run money demand relaed o ARDL (1,0,1,0,0,0,0,1) for res of real money are repored in able 2. As can see, all explanaory variables have expeced sign. Table 2: The esimaion of long-run money demand funcion coefficiens i iaia coefficien variable 2.55* 1.15 LY -1.93** -0.16 LP -3.37* -1.81 LR 2.34* 0.147 LE -1.72** -0.111 DD 2.61* 0.201 DU -2.36* -0.173 DG -0.607-1.348 C * Significan a 5% level of confidence ** significan a 10% level of confidence The calculaion findings show ha long-run income elasiciy of money demand is 1.15, wih oher word, 1% increases (decreases) in gross domesic producion increases (decreases)
142 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 money demand 1.15%. Income elasiciy of money demand is posiive, and i is compaible wih economic heories. Error correcion mehod is esimaed for making a link beween shorrun vibraions and long-run equilibrium. As saed before, in error correcion mehod, he difference values of variables wih lagged disribuive erms namely error correcion erm in long-run relaion and difference values of independen variable are considered. The resuls of esimaing he error correcion mehod relaed o ARDL (1,0,1,0,0,0,0,1) are repored in able 3. Table 3: Esimaion of ECM model of money demand funcion iaiai nfccafao eibairaf 2.6* 0.25 dly -5.3* -0.63 dlp -3.5* -0.39 dlr 2.64* 0.032 dle -1.74** -0.024 ddd 2.42* 0.44 ddu -0.87-0.013 ddg -0.62-0.29 dc -4.9* -0.22 ecm(-1) 2 R =0.89 DW=2.08 F= 23.42 *significan a 5% level of confidence ** significan a 10% level of confidence The ECM coefficien indicaes ha how many percens of shor-run inequaliy for geing o long-run equilibrium mus be adjused in each period o reurn he money demand ino own long-run rend. ECM is 0.22 in his model; i means ha in each period 22% of inequaliy of money demand is adjused and became near o long-run rend. The resuls of variance analysis (VDCF) for real demand of money in a 30-year period is represened in able 4. As finding show, inconsisency share of variable of real demand of money for shor-run is 82.28%, for middle-run 64.75% and for long-run 56.16%. Gross domesic producion variable in shor-run is 11.43%, in middle-run 20.05% and for long-run 22.44%. The
Inernaional Financial Crisis and Money Demand in Iran 143 inconsisency share of oher variables in convincing he res of real demand of money are represened in able 4. By comparing he inconsisency share of wo dummy variables relaed o las crisis, we can said ha 2006 financial crisis has more effec on money demand han he oher one specially in long-run. horizon 1 3 5 The average of 5 periods 7 9 11 13 15 The average of 15 periods 16 18 20 22 24 26 28 30 The average of 30 periods Table 4: The findings of analysis of variance for money demand variable M 1 0.953 0.816 0.705 0.8228 0.628 0.576 0.539 0.513 0.495 0.6475 0.488 0.479 0.474 0.472 0.472 0.4732 0.4749 0.4766 0.5616 Y 0 0.0327 0.1212 0.1831 0.1143 0.2198 0.2412 0.2533 0.2594 0.261 0.2005 0.2606 0.258 0.2539 0.2493 0.2451 0.2418 0.2398 0.239 0.2244 P 0 0.00123 0.00839 0.0187 0.0092 0.03 0.04156 0.0532 0.0651 0.0768 0.0367 0.0825 0.0934 0.1028 0.1105 0.1159 0.119 0.12 0.1194 0.0726 R 0 0.000828 0.0023 0.0024 0.0019 0.002 0.00172 0.00154 0.001486 0.001487 0.0018 0.001495 0.001512 0.001517 0.001504 0.001473 0.001429 0.001376 0.001321 0.0016 E 0 0.00076 0.0009 0.00066 0.0008 0.001577 0.003261 0.00506 0.006653 0.007908 0.0033 0.00841 0.009171 0.009628 0.009812 0.009767 0.00955 0.009223 0.008841 0.0063 DD 0 0.00181 0.00751 0.01271 0.0074 0.017 0.02056 0.02353 0.02602 0.02804 0.0173 0.02886 0.03012 0.03084 0.03101 0.03067 0.02994 0.02892 0.02774 0.0236 DU 0 0.00736 0.04089 0.07509 0.0410 0.09939 0.1143 0.1227 0.1267 0.128 0.0909 0.1279 0.1267 0.1247 0.1225 0.1204 0.1189 0.1181 0.1179 0.1065 D57 0 0.00165 0.00235 0.00155 0.0020 0.00118 0.00109 0.00101 0.00095 0.00101 0.0014 0.00111 0.00153 0.00226 0.00329 0.00456 0.006 0.00751 0.009 0.0029 7. Conclusion The calculaion resuls indicaes ha long-run income elasiciy of money demand are larger han 1, wih oher word, 1% increase
144 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 in gross domesic producion leads o 1.15% increase in money demand. Is sign is posiive and his confirms he economic heories. The esimaed coefficien of exchange rae has negaive sign and significan; i means ha in Iran economy here is a reverse relaion beween money demand and exchange rae variable. So he subsiuion effec of exchange rae in economic lieraure is confirmed. If he increase of exchange rae would be expeced, people increase he demand for foreign money for prevening of decrease in purchase power of money and i leads o decrease in domesic money. The comparison beween shor-run and long-run relaions shows ha long-run elasiciies are larger han shor-run ones due o more ime for adjusing o long-run equilibrium. Beside, in long-run relaion on money demand, income elasiciy is larger han 1 which is according o findings of developing counries. The coefficien of dummy variable (DD) which is significan in shor and long-run only 10% level of confidence, indicaes ha small effec of eas- souh crisis of Asia on money demand in Iran, bu DU which is significan in shor and long-run, shows ha las financial crisis has effec on money demand of Iran.he posiive coefficien of his variable indicaes ha by appearing he recession in marke due o high uncerainy and risk, he consumpion expendiures decreases and in his condiion money insead of oher wealh is sored. Therefore demand for money increases; of course is effec on shor-run is negligible and is effec will appear in long-run. Therefore, he sudy recommended he following policies: Given he inverse relaionship beween he exchange rae and money demand, Exchange rae seems o increase he liquidiy of lower demand. Insead i, increases he demand for foreign currency, The damage o he economy is o increase he exchange rae and, herefore, i should be noed. Increased producion will also increase he demand for money. Therefore, we mus carefully prediced annual producion
Inernaional Financial Crisis and Money Demand in Iran 145 rae will increase in proporion o he amoun of money Oherwise, he economy will suffer from several problems. Money demand is also influenced by he economic crisis, Tha i should be considered exper in macro-economic policies ha are essenial.
146 Quarerly Journal of Quaniaive Economics 8 (4), Winer 2012 Reference: Bahmani- Oskooee, M. & R. Chi Wing Ng. (2002). Long-Run Demand for Money in Hong Kong: An Applicaion of he ARDL Model. Inernaional Journal of Business and Economics, 1(2):147-155. Bahmani-Oskooee, M. & Y. Wang. (2007). How Sable is he Demand for Money in China?. Journal of economic developmen, 32(1). Bahmani-Oskooee, M. (1996). The Black Marke Exchange Rae and Demand for Money in Iran. Journal of Macroeconomics,18: 171-176. Bahmani-Oskooee, M. & M. Pourheydarian. (1990). Exchange Rae Sensiiviy of he Demand for Money and Effeciveness of Fiscal and Moneary Policies. Applied Economics, 22: 1377-84. Buch, C.M.(2001). Money Demand in Poland. Applied Economerics, 33:989-999. Erbykal, E., B. Darõcõ & Ö Kadõo lu. (2008). Reverse Money Subsiuion Process: Turkey Case. Inernaional Research Journal of Finance and Economics, 15: 1450-2887. Hafer, R.W. & D.W. Jansen. (1991). The Demand for Money in he Unied Saes: Evidence from Coinegraion Tess. Journal of Money, Credi, and Banking, 23: 155-168. Hamori, S. & A. Tokihisa. (2001). Sesonal Coinegraion and he Demand for Money Funcion: Some Evidence from Japan. Applied Economerics Leers,8:305-310. Payne, J. (2003). Pos Sabilizaion Esimaes of Money Demand in Croaia: Error Correcion Model using he Bound Tesing Approach. Applied Economerics, 35:1723-1727. Rao, B.B. & S. Kumar. (2006). Srucural Breaks and he Demand for Money in Fiji, Working Paper No. 2006, School of Economics, Universiy of he Souh Pacific.
Inernaional Financial Crisis and Money Demand in Iran 147