THE FISHER EQUATION EXAMINED: IMPLICATIONS FOR THE MONEY DEMAND IN TURKEY

Transcription

1 THE FISHER EQUATION EXAMINED: IMPLICATIONS FOR THE MONEY DEMAND IN TURKEY By Oben Donmez Cenral European Universiy Deparmen of Economics In parial fulfillmen of he requiremens for he degree of Maser of Ars Supervisor: Professor Max Gillman CEU etd Collecion Budapes, Hungary 2007

2 Absrac This hesis invesigaes he money demand in Turkey using monhly daa from 1986:1 o 2003:12. The invesigaion sars off by esing he Fisher equaion and resuls sugges ha he Fisher equaion does no hold in Turkey. This resul jusifies he inclusion of he inflaion variable in he esimaion of money demand funcion. The empirical analysis carried ou by means of he Johansen mulivariae coinegraion analysis and, coinegraion analysis indicaes ha here is a saionary long-run relaionship beween money, oupu, inflaion and ineres rae. Robusness of he analysis is checked by esimaing he money demand funcion when excluding he inflaion variable and he resuls are no supporive of exclusion of inflaion rae. Finding a sable money demand funcion in a period characerized as poliically and economically insable migh be surprising, bu i migh also provide suggesion of using he money aggregaes as policy insrumen. Moreover he variables in a sable money demand funcion migh be imporan while forecasing he inflaion as well. CEU etd Collecion ii

3 Acknowledgemens I would like o hank Professor Max Gilman for his help in choosing his opic and for his kind guidance in he furher seps. Special hanks are also addressed o Srdan Bejakovic, my family and my unnamed friends for heir suppor and ensuring my wellbeing. CEU etd Collecion iii

4 TABLE OF CONTENTS 1. Inroducion Turkish economy background Economeric modelling Daa descripion Inegraion Fisher Equaion Money Demand Esimaion Money Demand wihou Inflaion: Conclusion...22 CEU etd Collecion iv

5 1. Inroducion Economic heory has long posulaed a connecion beween money, ineres raes, and inflaion. The Fisher equaion radiionally describes he simple relaionship hypohesized o exis beween nominal ineres rae and he inflaion (Crowder, 1997). Similarly, classes of funcions have been sudied linking he demand for money o he ineres rae such as he consan semi-ineres elasiciy model of Cagan (1956) and he consan ineres elasiciy model of Baumol (1952). Given he general ineres and imporance of inflaion, and he fac ha mos cenral banks place conrolling he price level a he op of heir lis of responsibiliies, i is no surprising ha here has been much research in esimaing and refining hese relaionships, in he hope ha hey will provide insigh ino fuure inflaion rends and a way of influencing he level of inflaion. There are wo possible channels hrough which moneary policy migh influence he price level. The firs one is based on affecing he financial flows beween borrowers and lenders and he ineres rae is generally preferred as a policy insrumen over resricing he amoun of flows. The oher one is based on alering he money aggregae. Cenral Banks can influence boh channels, meaning ha hey are no exclusive. CEU etd Collecion However, ineres raes have mainly come o play a dominan role in cenral bank policies. The Unied Saes Federal Reserve implemens moneary policy conrolling he shor erm ineres rae by open marke operaions. Also one of main ool of disinflaion programs implemened by Turkish Cenral Bank has been ineres rae. For insance, here is a broad empirical lieraure abou he demand for money in he Unied Saed, and his lieraure mosly provides a skepical aiude abou using he money aggregae as an 1

6 informaion variable o guide moneary policy decisions because of is relaively high insabiliy.this obscures he value ha he money aggregae would have as a predicor of inflaion, and reduced is possible value as a policy ool for reducing inflaion. (Esrella and Mishkin, 1997). Sock and Wason (1993) invesigae he long-run money demand in he U.S for he period beween 1900 and 1989 and heir resuls abou he parameers of money demand are inconclusive. In conras, Ball (1998) exends he daa used by Sock and Wason hrough 1996 and he auhor finds fairly precise esimaes. More recenly, researchers have once again begun o focus on money aggregaes. For insance, researches done by conducing he daa of European Union counries poin o he exisence of sable money demand funcion.( e.g. Fagan and Henry, 1999; Coenen and Vega, 1999). For insance, Brand and Cassalo ( 2004) esimaes a demand funcion for real M3 in he euro area, and heir finding suggess no major disorions in he sabiliy of money demand. Also, he European Cenral Bank (ECB), wih he goal of finding sable money demand in he Euro area in 1998, graned money an imporan role in is policy. The role of money requires he ECB o analyze he developmen in moneary aggregaes, and use his informaion while aking policy decisions. Following he decision by ECB, many auhors ried o invesigae he relaion beween moneary CEU etd Collecion aggregae and fuure price developmens.(e.g. Gerlach and Svensson, 2000). This increasing ineres in he money aggregae does no discoun he imporance of he ineres rae, bu suggess ha, paricularly in he middle and he long run, money aggregaes are a useful ool for predicing and conrolling inflaion. In he words of Alvarez (2001: 219), while conrol of moneary aggregaes is he key o long-run 2

7 average inflaion raes, an ineres-rae policy can improve he shor run behavior of ineres rae and prices. There is an increasing ineres o focus on he sabiliy of money demand in developing counries as well, hough i is sill a challenge o conduc a sudy on money demand in hose counries, because of he lack of confidence in daa qualiy. This fac is also indicaed by Gillman and Cziraky (2005) in a sudy which he auhors examine he money demand in Croaia, an EU accession counry. Their findings are supporive of sabiliy of money demand funcion using inflaion as one of he variables. The inclusion of inflaion in he money demand funcion is jusified by finding a failure of Fisher equaion. This hesis follows he sudy of Gillman and Cziraky(2005), and sars off wih he invesigaion of Fisher equaion. The Fisher hypohesis is based on he relaionship beween nominal ineres rae and inflaion. There is a rich lieraure esing his relaion, and he early sudies are using U.S ime series daa. Alhough overall he resuls are, a bes, raher mixed, i is sill o say ha early findings demonsrae low Fisher effec (e.g. Evans and Lewis, 1995) whereas more recen sudies use more advance economeric CEU etd Collecion mehodology and hey find evidence supporing ha he long-run Fisher effec are very close o he heoreically implied value of 1.0 or greaer. (e.g Crowder and Hoffman, 1996). In case of finding a failure of Fisher equaion, he sandard money demand funcion needs o be modified, because he idea ha ineres rae reflec he inflaion rae changes is no valid anymore. The findings in his hesis are no supporive for he 3

8 evidence of such a relaionship beween ineres rae and inflaion. Tha is why while esimaing money demand funcion, inflaion rae is also included as one of he variables. Afer he exension of money demand specificaion, he finding in his sudy suggess a sable money demand funcion in Turkey. This resul migh be lile bi surprising because of he emphasis of he Turkish Cenral Bank on he ineres rae o conrol he inflaion rae as a policy ool and also because of he exisence of financial crises. All he series used in he esimaions are monhly and, hey exend from 1986:1 o 2003:12. The Fisher equaion is esed by using differen definiions and exensions and, money demand equaion is esimaed by using he Vecor Error Correcion Models (VECM).. For he purpose of checking he robusness of he baseline model, he money demand funcion is also esimaed wih he assumpion ha he Fisher equaion does hold. The remainder of he paper is organized as follows. In secion 2, he macroeconomic developmens beween 1980 and 2003 are described in order o ge beer insigh abou he variables used in he paper. The firs and second par of he secion 3 conains he daa descripion and hen he findings abou he inegraion of variables. The hird and forh par of secion 3 illusraes he esimaion resuls of Fisher equaion and CEU etd Collecion Money demand funcion. Finally, secion 5 conains he resuls. 4

9 2. Turkish economy background In his secion, we will review he macroeconomic developmens as well as policies implemened in Turkey over he las wo decades, which are relaed o his paper because of heir influence on he variables used in he esimaions. Turkey pursued an impor-subsiuion policy unil he end of 1970s. The crude oil shocks in he 1970s and he subsequen balance of paymens problems made he inward-looking developmen sraegy impossible o mainain. A broad sabilizaion and liberalizaion program was inroduced from 1980 onwards and he reforms ook place gradually. One of he major aemps o reverse he impor-subsiuion policy and srenghen he expor incenives was o devalue he Turkish lira o enhance is compeiiveness. The fixed exchange rae policy based on governmen decision for he deerminaion of Turkish Lira value was abandoned by he 1980 sabilizaion program, which iniiaed a managed floaing exchange rae scheme. In 1984, when he exchange rae regime was liberalized, he reforms allowed residens and exporers o hold a porion of heir income as foreign exchange deposis, and commercial banks o operae in foreign exchange marke in proporion o heir foreign exchange liabiliies. Moreover, he permission of he non-residens o purchase foreign denominaed securiies and o hold Turkish Lira accouns was made possible. CEU etd Collecion The Capial accoun liberalizaion which sared in 1980 was compleed by he serious seps aken by governmen in Soon afer he inroducion of capial accoun liberalizaion, banks and brokerage houses sared o compee for deposis by offering high ineres raes, and high ineres raes pushed he economy ino he financial crisis in 1982, because banks could no uilize heir high cos deposis. From 1986, he cenral bank implemened policies conrolling on Turkish lira reserves of he banking sysem 5

10 wih he goal of indirecly conrolling money supply by argeing broad money supply M2. Due o liberalizaion of capial accoun and in he absence of inerbank money marke, he cenral bank los is conrol over money aggregaes by Furhermore, moneary policy was highly dependen on fiscal policy. In he words of Pongsaparn (2002: 5), Prioriy was given o financial sabiliy raher han conrolling inflaion in he face of increasing currency subsiuion as seen from a rise in he share of foreign currency denominaed bank deposi in oal deposis from 24% in 1989 o 46% in In May 1985, he aucioning of governmen securiies in he a secondary bills and bonds marke,which se up he Isanbul Sock Exchange, sared for securiies wih one year securiy, and in a lile while he shor-run securiies also sared o be aucioned. Afer hese developmens in he governmen securiies marke, in April 1986, inerbank money marke was esablished. The main ool of he Cenral bank for moneary policy implemenaion has been hrough he open marke operaions by using governmen securiies since The Cenral bank became able o use marke based moneary insrumens in conducing is moneary policy whereas is abiliy o conrol money had been limied before by financial liberalizaion Furhermore, a foreign exchange marke where he value of Turkish lira was deermined according o he demand and supply of CEU etd Collecion foreign exchange rae by he marke paricipans (i.e. banks and auhorized exchange branches) opened in Sepember As a resul of all hese reforms, he Cenral Bank was allowed o conduc exchange rae, moneary and ineres rae policy hrough marke mechanism. High ineres raes followed by he liberalizaion of capial accoun also araced he shor run capial inflows o he counry. The shor erm capial inflows, so called ho 6

11 money, led he exchange rae o rise, hence hindered he invesmen and raised credi cos for real secor. All of hese developmens resuled in an exchange rae crises in he firs half of 1994 and he annual inflaion rae exceeded 100%. An explanaion of he jump in he inflaion rae comes from Rodrik (1991) who examined he premaure liberalizaion and incomplee sabilizaion in Turkey. Rodrik claims ha he liberalizaion of he capial accoun followed by an increase in dollarizaion led o a fall in demand for reserve money which pu he auhoriy o increase inflaion so as o earn seigniorage revenue. The Turkish governmen inroduced new disinflaion measures o sabilize he economy afer he 1994 financial crisis, however his did no las long because of he Russian crisis in 1998, general elecions in 1999 and he earhquake in Augus Increasing subsidies o agriculure raised enormously he duy losses of he banking secor which jumped from 0.7% of GNP in 1993 o 16.7% in 1999 and his also conribued o banking sysem volailiy. Banking secor volailiy was perceived as a facor conribuing o he crisis in The so called hree-year program was inroduced in 1999 and i was essenially esablished upon an exchange sabilizaion program supplemened by fiscal adjusmen and srucural reforms. One of he idenifying characerisic of he hree-year plan is ha CEU etd Collecion i provided more srucural and regulaive reforms o sele a marke-oriened economy and foser growh. Via hese reforms, he aims were o srenghen and regulae he banking secor, diminish he public secor defici and, reach o single digi inflaion level in he medium erm. The main policy ool change in disinflaion program was acually he adopion of a crawling-peg regime which is based on fixing he percenage change of 7

12 Turkish lira value of a baske of foreign currencies for a year and half. Iniially, he program was quie successful and i achieved remarkable resuls such as a dramaic fall in ineres rae, slowed down inflaion, increase in producion and demand. Neverheless, a medium size bank having large holdings of governmen securiies pushed he economy ino anoher financial crisis because of is exremely risky posiion. The governmen had o abandon he crawling peg and swiched o a floaing regime. Over 2001 he GDP conraced by 7.4% in real erms, whole sale price inflaion soared o 61.6%, and he currency los 51% of is value agains he major foreign monies. (Yeldan, 2006). According o Kibricioglu (2004), even hough facors such as he real appreciaion of he Turkish lira may have played a role in he occurrence of crisis, he main reasons were he unsusainable domesic deb of he public secor and he unhealhy srucure of he Turkish financial secor and hese argumens are well acceped by many oher auhors. In order o overcome he negaive impac of crisis, a new agreemen was made wih IMF in May 2001 and i was revised in early 2002 for he period of This program was more sringen han he previous adjusmen and reform program. The new plan mainly aimed o reduce uncerainy in he financial secor, especially hrough insiuional reforms, and o srengh public finance and adminisraion so as o ensure CEU etd Collecion deb susainabiliy and o bring down inflaion permanenly. The insiuional reforms in he banking secor consised of resrucuring he public banks and improving regulaion and supervision of privae banks. However, he banking secor sill does no fulfill he role of being a channel beween financial and real secor o foser invesmen. According o Pongsaparn (2002), he reasury bills and bond are he dominan asses of he balance shees of banks and, 8

13 his conribues he need for liquidiy and sho up ineres raes. Highly volaile oupu and risk also conribues o he reason why banks are secondary source of finance for many corporae firms, while larger corporaes prefer o borrow direcly from abroad. CEU etd Collecion 9

14 3. Economeric modelling 3.1 Daa descripion The daa 1 used in he esimaion are defined as follow: m is he narrow money (M1 money) which includes demand deposis in commercial banks and ime deposis in he Cenral Bank, p is he consumer price index (1995=100), is he monly rae of inflaion which derived as monhly change in CPI. (i.e. log cpi logcpi1 ; where cpi refers o consumer price index.), i is he indusrial producion for oupu variable and i is seasonally adjused. (1995=100) and, r is he ineres rae on governmen bonds (monhly rae- annual compound rae convered o a monhly rae). All he series in he esimaion are used in naural logarihms- excep ineres rae is used wihou naural logarihm in he money demand esimaion. The esimaion sample exends from 1986:1 o 2003:12 and series are wih monhly frequency. I is imporan o use monhly daa because of he hisorically high levels of inflaion in Turkey. In such condiions, economic agens make heir decisions and change heir behavior frequenly, and quarerly daa migh no represen his adequaely. CEU etd Collecion 3.2 Inegraion Before he invesigaion of he Fisher equaion and money demand funcion, he univariae uni roos were performed o deermine he order of inegraion of variables by he Augmened Dickey Fuller (ADF) process. The uni roo ess are given for he level and firs differences of he daa and deailed ADF es is repored in Table 1. The opimal lag lengh was deermined by using he Akaike (AIC) informaion crieria. Uni roos 1 The daa is obained from he Cenral Bank of he Republic of Turkey websie ( I also would like o hank Irfan Civcir (Faculy of Poliical Science, Ankara Universiy) for helping me o exend he daase. 10

15 ess are firs performed for wo roos, and if wo roos are rejeced hen single uni roo is esed. Tess are carried ou wih and wihou rend. Tes resul shows ha none of he variables seems o have evidence of wo uni roos, and all he variables are no able o rejec he null hypohesis of single uni roo a he 5% significance level. Tha is, all he variables are non-saionary in level, bu saionary afer firs differencing. Table 1- Uni Roo Tes Variables k A B Levels m m-p r i Firs Differences (m) * * (m-p) * * ( ) * * (r) * * (i) * * 1% criical value % criical value k is he number of lagged dependen variable. Columns A and B give he -saisics from ADF regression including consan, and rend, and consan, respecively. CEU etd Collecion 3.3 Fisher Equaion ineres rae, The Fisher equaion can be represened as r is real ineres rae and,, where r is nominal is inflaion rae in period,. (Fisher, 1930; see also Gillman and Cziraky, 2005). Wih an addiional assumpion ha (i.e real ineres rae is consan), where ~ i.i.d, he Fisher equaion becomes r. This equaion implies independence of he real ineres rae and 11

16 inflaion, in oher words no direc and consisen connecion of any real significance exiss beween price changes and ineres rae (Pelaez quoed by Dimand, 1999: 745). The equaion is usually esimaed in log levels as ln( ) u, where ln( r ) 0 1 consan 0 can be inerpreed as he long-run equilibrium real ineres rae and, 1 is expeced o be 1 o hold he long run validiy of he Fisher equaion. (Gillman and Czikary 2005, 14). Noe ha, when he variables are in logarihms, inflaion is measured as ln( ) ln( p / p 1 ), i.e., ln( p ). Wih all addiional assumpions and modificaions, Fisher equaion can be shown as ln( r ) 0 1 ln( p ) u, ~ i. i. d., u 1=1. Iniially ignoring he order of he inegraion, he esimaed equaion is ln( r ) ln( p ), (0.385) (0.007) The sandard errors are in parenheses and R-squared is equal o , S.E of regression is equal o 0.302, and DW is equal o I is eviden ha he null hypohesis H 0 canno be rejeced, and in addiion, a low Durbin Wason 0 : 1 CEU etd Collecion saisics implies dynamic misspecificaion and serial correlaion beween residuals. The ADF uni roo es on residual produces a -value of where he highes significan lag is 2, which clearly canno rejec ha u is I (1) (i.e inegraed of order 1). Following Gillman and Czikary (2005), he esimaion of Sargen s (1972) exended Fisher equaion wih n=m=3 is esimaed and he esimaion resul is as follows 12

17 ln( r ) ln( m ) 0.108ln( m 1) 0.04ln( m2 ) 0.69ln( m3 ) (0.91) (0.303) (0.373) (0.375) (0.286) ln ( p ) 0.801ln ( p 1 ) ln ( p 2 ) 0.77 ln( p3 ), (0.568) (0.99) (0.99) (0.60) wih R-square equal o , sandard error of regression is equal o0.1847, and DW is equal o Because DW is significanly smaller han 2, here is a possible remaining residual auocorrelaion, alhough he fi improved and he residuals are saionary. The ADF -value was wih 3 lags included in regression, which is above he 5% criical value of wih consan). This is also seen in he long run relaion: ln( r ) ln( m ) ln( p (0.803) (0.099) (0.102) ), When he long run relaion resricions are imposed on coefficiens, he 2 Wald 1485 ( 2) and, i is highly significan. Even hough he sign of coefficien are as expeced, boh coefficiens are insignifican. Moreover, -saisic from ADF (-2.858) implies uni roo which confirms he previous conclusion abou inegraion order of CEU etd Collecion variables. Alernaively, again following Gillman and Cziraky(2005), I consider a bivariae VECM sysem formed wih nominal ineres rae and inflaion using he Johansen echnique. Iniially, opimal lag lengh was deermined by LR saisics which assures no serial correlaion beween residual in he sysem, and he suggesed lag lengh is 5 for he model. The model represenaion wih vecors is as follows 13

18 ln ln r p 1 2 ( i ) 11 ( i ) 21 ( i ) 12 ( i ) 22 ln ln r p i i ln ln r 1 p 1 The sysem s -race and -max saisics are and and , whereas 95% criical values of -race and -max ess are 19 and These resuls imply ha here is no coinegraion beween he ineres rae and inflaion. The long-run Fisher equaion does no hold in Turkey. Tha is why inflaion will be added ino he sysem of money demand. 3.4 Money Demand Esimaion In his secion, he Vecor Error Correcion Models (VECM) is applied o gain an insigh ino dynamic of he money demand funcion and allow for feedbacks among endogenous variables of he funcion. The sandard VECM is obained from a vecor CEU etd Collecion auoregressive (VAR) model; Here, x k i 1 A i x i x is (nx1) dimensional vecor of endogenous variables, u conains deerminisic erms like consan and ime rend, Ai are (n n) dimensional coefficien marices and u (0, ) is serially uncorrelaed erm. In case of serial correlaion, i can be absorbed ~ u 14

19 by including appropriae number of lags. Subracing x 1 and rearranging erms yields he VECM: 1 1 k 1 x x x u, where x is a vecor of nonsaionary (in levels) variables and, and i1 i i i are funcion of he : A i. The marix can be decomposed ino wo (n r) dimensional marices and where is adjusmen marix conaining adjusmen coefficiens which show he amoun of changes in he variables o bring he sysem back o equilibrium and, comprises he coinegraing vecors which show he long run equilibrium relaionship beween levels of variables, and r is number of linearly independen coinegraing vecors. Following Gillman and Czikary e al. (2005), he vecor x comprises he following variables: real money ln ( m p), inflaion ln ( ), indusrial producion ln( i ) used as oupu variable, ineres rae R. The four variable sysem is represened as CEU etd Collecion ( m p ) y R i1 ( i 11 ( i 21 ( i 31 ( i 41 ) ) ) ) ( i ) 12 ( i ) 22 ( i ) 32 ( i ) 42 ( i ) 13 ( i ) 23 ( i ) 33 ( i ) 43 ( i ) 14 i () 24 ( i ) 34 ( i )) 44 ( m p ) y i i R i ( m y R p ) i 15

20 The lag order of he esimaed VECM was deermined by using Akaike (AIC) informaion crieria. Gordon (1995) suggess ha according o various crieria differen number of lagged differences may arise. I follow he sraegy o selec he minimal lag lengh suggesed by saisics which ensures he error erm is no serially correlaed. The suggesed lag order by AIC is 14 for he model, and i also jusified by sequenial modified LR saisics a 5% level. I sared esimaing a VAR(14) wih consan and erm. The diagnosics resuls in he form of vecor saisics indicae ha our VAR model is saisfacorily a close approximaion o he acual daa process, wih he excepion of some non-normaliy of residuals: The resul of he VAR sabiliy es shows ha no roo lie ouside he uni circle bu here were wo quasi-uni roos (modulus>0.98) and, he LM 2 ( 14) resuls suggess ha he model residuals are no serially correlaed whereas 2 normaliy es implies non-normaliy of residuals. ( 8) In order o deermine he number of he coinegraing relaionship, he CEU etd Collecion Johansen(1988) maximum likelihood mehod is preferred insead of Engle-Granger wosep mehod, because Johansen mehod provides more robus resuls, especially when more han 2 variables are included in he model.(gonzalo, 1994). The Johansen approach is based on he relaionship beween he rank of a marix derived from coefficien marices and is characerisic roos. The Johansen coinegraion es provides wo differen ess o deermine he number of coinegraing vecors which are race and 16

21 maximum eigenvalue ess. The null hypohesis of he race es is ha here are a mos r coinegraing vecors while he alernaive hypohesis is a general one. In he maximum eigenvalue es, he null hypohesis is ha here are r coinegraing vecors and i is esed agains r+1 coinegraing vecors. Boh he maximum eigenvalues and race es saisics srongly rejec he null hypohesis of no coinegraion in favor of one coinegraion relaionship which is given in Table-. On he oher hand, hey show inconsisen resuls for more han one coinegraing relaionship. Because in he model wih oo many variables or lags he Johansen procedure ends o over esimae he number of coinegraing vecor, I prefer o go wih 1 coinegraing relaionship which was approved by boh es saisics. Table 2- Coinegraion es resuls Hypoheses Eigenvalue race 95% Criical Value max %95 Criical Value r = * * r<= * r<= r<= * * 3.84 CEU etd Collecion The esimaion using he Johansen maximum likelihood echnique does no conain he rend erm. Firs of all rend coefficien was insignifican in VAR(14) and his resul is also visually jusifiable by individual graphs of variables. Finally, he esimaion resuls for he unresriced coinegraing and adjusmen coefficien marices are as follow: 17

22 For he purpose of invesigaing he possibiliy of long run weak exogeneiy of he variables wih respec o he coinegraing parameers, various resricions were imposed on he parameers of alpha marix and, he only hypohesis ha canno be rejeced was ha inflaion is weakly exogeneous wih respec o he long-run parameer wih a es saisic 2 (1) Wih he valid resricion of weak exogeneiy of inflaion imposed he esimaes of and becomes: All he coefficiens in he coinegraing vecor have anicipaed signs and, hey CEU etd Collecion provide economically meaningful represenaion of a money demand funcion. Namely, he coinegraing vecor suggess a posiive relaionship beween money demand and oupu whereas negaive relaionship can deduced for money demand and ineres rae and inflaion. Afer normalizing he coinegraing vecor relaion o m p and wriing he long run relaionship in equaion forma, he long run money demand equaion becomes 18

23 m p 1.16( y ) 0.046( ) R Inflaion has a semi elasiciy of 0.05 percen and, even when he weak exogeneiy of inflaion is imposed on coinegraing vecor, he coefficien on he inflaion does no change significanly. This relaion suppors he view ha agens decrease heir money holding in favor of real asses when he inflaion is expeced o rise. The equaion also indicaes long run uni income elasiciy which means ha here is a one-for-one relaion beween changes in money demand and changes in income VELOCITY R Figure 1 Velociy and Nominal Ineres Rae CEU etd Collecion However, hese resuls on income elasiciy and ineres elasiciy are conradicory wih Baumol s sudy(1952) which predics ha boh he income elasiciy and he ineres elasiciy of money demand is one half. 19

24 Moreover he ineres rae coefficien is no so big which implies a weak relaionship beween money demand and ineres rae and, anoher way o inerpre his resul would be relaing i wih velociy which is defined as oupu divided by real money.. As graphs also indicae, if he ineres rae coefficien was large, a decreasing ineres rae would cause a decrease in money velociy hroughou Bu as we can also see from he graph above, he sharp decrease in ineres rae did no affec he velociy much and velociy coninued o rise. The adjusmen coefficiens for he money demand and inflaion are small. Hence, hey sugges small reacions by money demand and inflaion o a deviaion from he long run equilibrium, whereas ineres rae coefficien is much bigger in he shor-run han long-run Money Demand wihou Inflaion: Afer invesigaing ha he Fisher equaion does no hold for Turkey, including inflaion rae ino he money demand funcion was jusified. For he purpose of checking he robusness of he baseline model, he money demand funcion is now a hree variable CEU etd Collecion VECM insead of four variable one wih he assumpion ha he Fisher equaion does hold. The opimal lag order suggesed by AIC is 15 in his case, and he es saisics of he Johansen procedure sugges no coinegraion. The es saisics wihou he rend in CE is he firs one, and laer is he resul wih he rend in CE: 20

25 Table 3- Coinegraion es resul Hypoheses Eigenvalue race 95% Criical Value max %95 Criical Value r = r<= r<= Table 4- Coinegraion es saisics wih he rend in coinegraing equaion: Hypoheses Eigenvalue race 95% Criical Value max %95 Criical Value r = r<= r<= CEU etd Collecion 21

26 5. Conclusion This hesis models he long run and shor run money demand relaionship in Turkey for he period of 1986:1-2003:12. Throughou he period, Turkey experienced wo financial crises (1994, 2001), many reforms and insiuional changes in he economy; he Capial accoun liberalizaion which sared in 1980 was compleed by he serious seps aken by governmen in 1989, he fixed exchange rae policy based on governmen decision for he deerminaion of Turkish Lira value was abandoned by he 1980 sabilizaion program, and a managed floaing exchange rae scheme was inroduced, amoun of dollarizaion increased dramaically, inroducion of financial innovaion driven mainly by an increasing governmen deb mosly caused by moralhazard problem and financial liberalizaion. The period is also characerized by a high inflaion rae, increasing money supply, unsable exchange and ineres raes. Empirical analysis sared off by invesigaion of Fisher economy in Turkey. Afer using sequence of ess derived by differen approaches, he resuls reveals ha Fisher CEU etd Collecion equaion does no hold in Turkey, meaning ha he nominal ineres rae and inflaion rae does no move ogeher and also hey are no inerchangeable in he money demand funcion. Thus only ineres rae is no appropriae o represen he cos of money and, a modificaion of he sandard money demand funcion by including inflaion rae was needed o model he money demand relaionship in Turkey. 22

27 The analysis abou money demand funcion was carried ou by means of Johansen mulivariae coinegraion analysis mehods. The resuls sugges ha here is a saionary long-run relaionship beween money demand, oupu, inflaion and ineres rae on governmen bonds beween 1986:1 and 2003:12, a period which is poliically and economically less han calm. The coinegraion analysis indicaes ha income elasiciy is close o uniy in accordance wih quaniy heory of money, and ineres rae and inflaion have correc sign in he long run. In he shor run, he adjusmen coefficiens for he money demand and inflaion are small. Hence, hey sugges small reacions by money demand and inflaion o a deviaion from he long run equilibrium, whereas ineres rae coefficien is much bigger in he shor-run han long-run. The robusness of he resuls was checked by analyzing he money demand relaion wihou inflaion variable. The resul indicaes no coinegraion beween money demand, ineres on governmen bonds and oupu and no supporive of sable money demand relaion wihou inflaion rae in he funcion. These resuls migh provide a guide for he moneary auhoriy o use money aggregaes as policy ool, hough his does no necessarily mean ha money should be he only insrumen for moneary policy. The relaionship beween variables would also be CEU etd Collecion helpful for he moneary auhoriies, since one of heir main asks is o forecas he inflaion. 23

28 References Akcay, O.C, Alper, C.A and Karasulu, M. (1997), Currency Subsiuion and Exchange rae insabiliy: Turkish Case, European Economic Review, Vol 41 (3): Alvarez, F., R.E. Lucas, Jr., W. E. Weber (2001), Ineres Raes and Inflaion, American Economic Review, 91 (2): Agenor, P.R., Jensen, H.T., Verghis, M., and Yeldan, E. (2006). "Disinflaion, fiscal susainabiliy, and labor marke adjusmen in Turkey," Policy Research Working Paper Series 3804, The World Bank Alimari N. (2001), Does money lead inflaion in he euro area?, ECB Working Paper No. 63. Ball L. (1998), Anoher Look a Long Run Money Demand, NBER Working Paper No Baumol, W. J. (1952) The Transacions Demand for Cash: An Invenory Theoreic Approach, Quarerly Journal of Economics, 66(4): Brand, C. and N. Cassola (2004), A money Demand Sysem for Euro Area M3, Applied Economics Vol 36: CEU etd Collecion Cagan, P. (1956), The Moneary Dynamics of Hyperinflaion, in: M. Friedman, (ed.), Sudies in he Quaniy Theory of Money, Chicago: Universiy of Chicago Press. Civcir, I. (2003), Money Demand, Financial Liberalizaion and Currency Subsiuion in Turkey, Journal of Economic Sudies, Volume 30 (5): Coenen, G. and J.L. Vega (1999), The Demand for M3 in he Euro Area, ECB Working Paper Series No

29 Crowder W. J. (1997), The Long-Run Fisher Relaion in Canada, The Canadian Journal of Economics / Revue canadienne d'economique, Vol. 30 (4b): Crowder, W.J. and Hoffman, D.L. (1996), The Long-Run Relaionship beween Nominal Ineres Raes and Inflaion: The Fisher Equaion Revisied, Journal of Money, Credi and Banking, 28(1): Diboolu, S. and Kibriçiolu A. (2004), Inflaion, Oupu Growh, and Sabilizaion in Turkey, , Journal of Economics and Business, Vol 56(1): Dickey, D.A. and Fuller, W.A.(1981), Likelihood Raio Saisics for Auoregressive Tim Series wih a Uni Roo, Economerica, Vol. 49 (4): Dimand, R.W. (1999), Irving Fisher and he Fisher Relaion: Seing he Record Sraigh, Canadian Journal of Economics, 32(3): Ender, W. (1995), Applied Economeric Time Series, New York : Wiley Esrella, A. and F.S. Mishkin (1997), Is There a Role for Moneary Aggregaes in he Conduc of Moneary Policy?, Journal of Moneary Economics 40: Evans, M.D.D. and Lewis, K.K., Do expeced shifs in inflaion affec esimaes of he long-run fisher relaion, Journal of Finance Vol 50 : CEU etd Collecion Fagan, G., Henry, J. and R. Mesre (1999), An Area-Wide Model for he EU11, ECB Working Paper No 42 Fisher, I. (1930), The Theory of Ineres. New York: MacMillan Friedman, M. (1956) The Quaniy Theory of Money: A Resaemen, in M. Friedman (ed.), Sudies in he Quaniy Theory of Money, Chicago: Universiy of Chicago Press. Gerlach, S. and L.E.O. Svensson (2000), Money and Inflaion in he Euro Area: A Case for Moneary Indicaors?, NBER Working Paper No

30 Gillman, M., and Czikary, D. (2005), Money Demand in an EU Accession Counry: A VECM Sudy of Croaia, Cardiff Economics Working Papers E2005/7, Cardiff Universiy, Cardiff Business School, Economics Secion Gillman, M and Michal K. (2004), The Demand for Bank Reserves and Oher Moneary Aggregaes, Economic Inquiry, Vol. 42(3): Gillman, M., and Oo, G. (2003), Money Demand in a Banking Time Economy, HWWA Discussion Paper 254, Hamburg Insiue of Inernaional Economics Gonzalo, J. and Piarakis, J.Y. (1994), Coinegraion Analysis in Large Sysems, Technical Repor, Deparmen of Economics, Boson Universiy Johansen, S. (1988), Saisical Anlaysis of Coinegraing Vecors, Journal of Economic Dynamics and Conrol, Vol. 12 (2): Johansen, S. (1995), Likelihood-based Inference in Coinegraed Vecor Auoregressive Models. Oxford: Oxford Universiy Press. Kibriçiolu, A. (2002), Causes of Inflaion in Turkey: A Lieraure Survey wih Special Reference o Theories of Inflaion, in: Kibriçiolu, Ayku, Libby Rienberg and Faruk Selçuk (eds.) (2002) Inflaion and Disinflaion in Turkey, Aldersho, UK, & Burlingon, VT: Ashgae, pp CEU etd Collecion Kibriçiolu, A. (2005), Macroeconomic Developmens in Turkey: A Long-Term View, Vienna Insiue for Inernaional Economic Sudies (WIIW) Monhly Repor, No : 5-7. Pongsaparn R. (2002), Inflaion Dynamics and Reacion Funcion in High-inflaion Environmen: An Implicaion for Turkey, The Cenral Bank of he Republic of Turkey, Research Deparmen Working Paper No

31 Rodrik, D. (1991), Premaure Liberalizaion, incomplee Sabilizaion: The Ozal Decade in Turkey in Pos-liberalizaion and is Afermah, in M. Bruno e al. (eds.), Cambridge, MA, MIT Press Sargen, T.J. (1972), Anicipaed Inflaion and he Nominal Rae of Ineres, Quarerly Journal of Economics, 86(2): Sock W., and Wason M. W.(1993), A Simple Esimaor of Coinegraing Vecors in High Order inegraed Sysem, Economerica 61: Taylor, J.B. (1999), The Robusness and Efficiency of Moneary Policy Rules as Guidelines for Ineres Rae Seing by he European Cenral Bank, Journal of Moneary Economics, 43(3): CEU etd Collecion 27

32 . CEU etd Collecion 28