ARE THE DETERMINANTS OF MONEY DEMAND STABLE IN SELECTED COUNTRIES FROM SOUTHEASTERN EUROPE?

Transcription

1 5. ARE THE DETERMINANTS OF MONEY DEMAND STABLE IN SELECTED COUNTRIES FROM SOUTHEASTERN EUROPE? Jordan KJOSEVSKI 1, Mhal PETKOVSKI 2 Abstract The purpose of ths artcle s to examne the long and short-run determnants of money demand (M1) and ther stablty n the seven Southeast European (SEE) countres usng monthly data from January 2005 to December The Pool Mean Group Estmaton of ARDL was used to fnd the long-run and short-run dynamc relatonshps n money demand model. Emprcal results provde the evdence that ndex of ndustral producton, exchange rate and dummy varable of effect of the European debt crss explan the most varatons of money demand n the long-run, whle exchange rate s sgnfcant only n short-run. Our fndngs also show that real money demand n the SEE countres, despte ther turbulent transton from socalst to market economy, was relatvely stable n the analyzed perod. Keywords: money demand, determnants, Southeast Europe, panel ARDL JEL Classfcaton: C12, C33, E41 I. Introducton Effectve and stable money demand estmatons are the precondton for the monetary authortes to desgn an effectve monetary polcy. The mportance of the money demand functon has encouraged a wde range of economsts to emprcally study ts determnants. But, whle the money demand lterature has focused on developed countres, there have been relatvely few studes examnng the money demand functon n transton economes and especally n SEE countres. The relatve absence of emprcal money demand studes for SEE economes, s usually explaned by the nstablty of the transton process tself, and wth concerns over the lack of avalable data on selected determnants of money demand over longer perods of tme (Payne, 2003). Although economes of the SEE countres mproved markedly over the begnnng of the last decade; they faced sgnfcant challenges snce the global fnancal 1 Moneymax Fnancal, Ohrd Republc of Macedona.E-mal: koseskjordan@gmal.com 2 Fnancal Management, Faculty of Economcs, Ss Cyrl and Methodus Unversty n Skopje, Republc of Macedona. 84 Romanan Journal of Economc Forecastng XX (4) 2017

2 Are the Determnants of Money Demand Stable n Selected Countres crss began n 2008, and contnued wth the European debt crss, from These factors are suspected to contrbute to the stablty/nstablty of money demand functon n SEE countres. Estmatng a stable money demand functon s essental for the central banks from SEE countres wth respect to ther targets of sustanable growth and prce stablty. Consderng narrower concept of money supply M1 as a measure for money n the selected countres, our research adds to a growng lterature by determnng the varables of money demand n the selected countres from SEE. We focused on seven countres from Southeast Europe (Albana, Bosna and Hercegovna, Bulgara, Croata, Macedona, Serba and Romana), usng monthly data, for the perod The sample has been chosen because these countres are relatvely homogenous, wth common hstory and smlar nsttutons and economes. The analyss of ths study wll contrbute to the exstng level of knowledge n several ways. Frst, t wll gve new nsghts nto the orgns, characterstcs and consequences of long and short determnants and ther stablty n SEE economes as a whole. Namely, most studes focus on ndvdual country cases and accordng to the knowledge of the authors there are none for ths regon. Ths s an ntal effort to reveal the parameters that govern ths key relatonshp n economcs Second, some of the exstng panel studes (Kumar et al. 2010; Nautz and Rondorf, 2010) usually consdered only several varables (e.g. ncome, prces and a measure of the opportunty costs), n the money demand equaton. But, n the perod of transton exchange rate can also play a crucal role n explanng money demand. Namely, durng perods of hgh nflaton, the countres n our study experenced a partal replacement of domestc wth foregn currences, ether as a store of value or as a medum of exchange. Also, all selected countres from SEE are small open economes, and the foregn trade lberalzaton durng the transton process has therefore, affected countres behavor wth respect to ther demand of foregn and domestc fnancal assets. These countres could swtch more easly between foregn and domestc currences. Ths may have affected the money holdngs n these economes. The exchange rate on money demand s also mportant for these countres n vew of the fact that they are tendng to jon the euro area sometme n the future and are lkely to focus on mnmzng the volatlty of the domestc currency value aganst the euro. Therefore, the exchange rate mght become an mportant factor explanng money demand behavor n these countres and wll be used n the analyss. Thrd, compared to prevous studes, ths paper uses a newly developed econometrc technque known as panel ARDL model or Pooled Mean Group (PMG) estmaton proposed by Pesaran et al. (1999). Ths model s able to capture the long-run and short-run relatonshp among the varables of nterest, and also to examne the dynamc effects of selected varables on money demand. Also, n ths study we wll nclude two dummy varables, the global economc crss from 2008/2009 and current European debt crss. We thnk that these two dummy varables wll be an mportant factor for explanng money demand behavor. Accordng to the knowledge of the authors, there has been only one study Kjosevsk (2013) that has used global crss 2008/2009 and the European debt crss 2011/2012 n hs model of money demand. The structure of the paper s as follows. After the Introducton, Secton 2 revews the lterature on emprcal fndngs relevant to the determnants of money demand. The sources of the data employed as well as methodology are presented n Secton 3. Secton 4 presents the emprcal results and secton 5 concludes the paper and gves polcy recommendaton. Romanan Journal of Economc Forecastng XX (4)

3 II. Lterature Revew Insttute for Economc Forecastng A consderable body of lterature has nvestgated the demand for money both n developng countres Omotor (2009) and n developed countres (Beyer 1988; Brand and Cassola 2000; Calza, et al. 2001). But, there are only several papers on money demand n SEE countres (Anusc 1994; Sonje 1999; Babc 2000; Maravć and Palć 2005; Karla, 1998; Kjosevsk, 2013). Maravć and Palć (2005) examned the long-term and short-term money demand n Serba for the perod from January 1996 to March 2005 applyng Johansen contegraton technque and the VECM model. The results of ther study ndcate that real money M1 has a strong contegraton relatonshp wth economc actvty, nterest rate on dnar deposts and nflaton. Furthermore, the results of the VECM model show that nflaton and exchange rate were the most mportant determnants of money demand, whle nterest rate on deposts n dnars, were statstcally nsgnfcant n the short term. Among the frst analyses that explore the demand for money n Croata are Anusc (1994) Sonje (1999), and Babc (2000). Anusc (1994) used an ordnary least squares method for the perod from January 1991 to November Hs study ndcates that n the perod of hypernflaton n Croata nflaton and GDP were the man determnants of the demand for money. The results also show that nterest rate had no sgnfcant mpact on the demand for money n the analyzed perod. On the other hand, Sonje (1999), analyzed money demand n the perod after hypernflaton, offerng emprcal evdence that nflaton no longer had a sgnfcant effect. Babc (2000) also suggests that nflaton s statstcally nsgnfcant varable and has no nfluence on the demand for money n Croata. Karla (1998) estmates relatonshp between money, nflaton, prces, exchange rate, and nterest rate n Albana durng usng parsmonous error correcton model. She fnds that n the long run there s a postve relatonshp between the prce level and the exchange rate, and between real money demand and exchange rate expectatons, nterest rates and the level of economc actvty Kjosevsk (2013) apples Johansen contegraton technque and VECM model to estmate the money demand n the Republc of Macedona usng monthly data from January 2005 to October The emprcal results n hs paper provde evdence that exchange rate and nterest rate payable on dnar tme deposts up to one month explan the most varatons of money demand n the long-run, whle nterest rate s sgnfcant only n short-run. Hs fndngs show that real money demand M1 n the Republc of Macedona was stable n the analyzed perod. There s only one study Ozturk and Acaravc (2008) whch examnes a porton of the South Eastern European countres although they ncluded only four SEE countres n ther study: Bulgara, Croata, Macedona and Romana). They estmate the demand for M2 for a panel of 10 transton countres (Bulgara, Croata, Czech Republc, Hungary, Macedona, Poland, Romana, Russan Federaton, Slovak Republc and Ukrane) usng data from wth feasble generalzed least squares. The results show that the demand for money and quas money (M2) s postvely related to real GDP and negatvely to nflaton rate and the real effectve exchange rate. The estmated common long-run ncome elastcty for the ten transton economes s about unty. 86 Romanan Journal of Economc Forecastng XX (4) 2017

4 Are the Determnants of Money Demand Stable n Selected Countres III. Data and Methodology III. 1 Model Specfcaton and Data Followng earler works on the money demand functon (Arango and Nadr 1981; Stock and Watson 1993; Ercsson 1998; Mark and Sul, 2003; Valadkhan and Alauddn 2003) the emprcal model of the money demand can be summarzed by the followng functon. M FY, OC (1) P where: M denotes nomnal money, P prce level, Y a scale varable representng the transacton volume n the economy and OC denotes opportunty costs of holdng money. Before gong to dentfy potental determnants of money demand t s necessary to dentfy the dependent varable. In the lterature (Payne, 2003; Skrabc and Tomc-Plazbat, 2009) the narrower concept of money supply should be used for economes wth a relatvely underdeveloped fnancal system. Bearng n mnd the current development of the bankng and fnancal systems n the SEE s countres, as a measure for money n ths paper we use M1 monetary aggregate, whch covers currency n crculaton and sght deposts. We then deflate M1 wth consumer prce ndex (CPI) to get the measure of the real money balance of M1. The use of a narrow monetary aggregate has several other advantages. Frst, M1 s a good measure of lqudty n the economy snce t conssts manly of fnancal assets held for transacton purposes. Second, the central bank s able to control ths aggregate more accurately than broader aggregates such as M2 and M3 Dobnk (2011). Thrd, M1 defntons tend to be relatvely consstent across countres and, therefore, allow straght comparsons Bruggeman (2000). Some authors ndcate that usng GDP as the measure of economc actvty leads to overestmaton of the level of transactons n the economy and suggest alternatve measures such as the level of consumpton Mankw and Summers (1986) or the ndex of ndustral producton (Payne, 2003; Skrabc and Tomc-Plazbat, 2009). For the purpose of our study we wll follow Payne (2003) and Skrabc and Tomc-Plazbat (2009) and we wll use ndex of ndustral producton. For ths varable we expect a postve correlaton wth money demand. Accordng to Payne (2003) exchange rate s an mportant factor n the demand for money n transton economes. But, the effect of the exchange rate on money demand s not entrely clear. Namely, Komárek and Melecký (2001) ndcate that deprecaton of the domestc currency s lkely to nduce extra demand for domestc goods from abroad and the nduced rse n domestc producton mples hgher domestc nflaton rate and a need for more money n the economy as the amount of transactons ncreases. Hence, beng the exchange rate denoted as unts of domestc currency per unt of the foregn currency, ts coeffcent should be postve. On the other hand, accordng to the currency substtuton approach Calvo and Rodrguez (1997), deprecaton reduces the confdence n the domestc currency, thereby lowerng money demand va a substtuton effect wth foregn money. Hence, ts coeffcent should be negatve. Accordng to the above studes we expect real exchange rate to be ambguously related to money demand. As a determnant of the exchange rate n ths paper we use the nomnal exchange rate of domestc currences per euro. To measure opportunty cost we use nomnal (domestc) short-run nterest rate from Central Banks of the selected countres. We choose ths varable because accordng to Ercsson (1998) long-run rates should not be ncluded n the model for money demand when M1 monetary aggregate s used. Also, Komárek аnd Melecký (2001) suggest that the portfolo Romanan Journal of Economc Forecastng XX (4)

5 Insttute for Economc Forecastng motve of holdng such money plays only a mnor role relatve to the transacton motve. For ths varable, we expect a negatve correlaton wth money demand. The next varable used n our research s the nflaton rate. It s used to measure the monetary stablty of the country. Ths varable s expressed by annual ncrease n CPI (annual percentage base 2005 = 100). The negatve mpact of nflaton has been wdely documented n prevous research (Hosen 2007; Mehrotra 2008; Dreger and Wolters 2009). Therefore, we also expect a negatve relatonshp wth money demand. Gven the varables specfed above, we construct a panel data regresson model as shown below. The real money demand and exchange rate were transformed nto natural logs. L( 1 / CPI ) t n 7 1 ( IIPS ) L( EXRS ) ( NIRATE ) ( INFS ) 0 1 t 2 t 3 t 4 t (2) 5( DUM ) t 6( DUM 1 ) t t where: denotes a specfc country varyng from 1 to 7, t s tme startng from January 2005 to December 2014; M1t / CPIt = Real money (M1 deflated wth consumer prce ndex CPI); IIPS = Index of ndustral producton (base 2005=100) (seasonally adjusted); EXRS= Exchange rate of domestcs currences per euro (seasonally adjusted); NIRATE=Nomnal nterest rate; INF = Rate of nflaton. (base 2005=100) (seasonally adjusted); DUM = effect of the 2008/09 global economc crss; DUM1 = effect of the European debt crss; t s a whte nose error process; In order to capture nter-country heterogenetes, one can use the fxed effects model, whch allows 0, ) to vary across countres by estmatng dfferent ntercept terms ( , 05. For our research we focus on factors that determne money demand n seven countres from South- Eastern Europe (Albana, Bulgara, Bosna and Hercegovna, Croata, Macedona, Romana and Serba). In order to obtan more observatons, we used monthly data from January 2005 to December The choce of the countres and the tme perod n ths paper ware contngent upon the avalablty of tme seres data on all the varables ncluded n the model. Some of the above-mentoned determnants, such as: ndex of ndustral producton, exchange rate and nflaton, are seasonally adjusted usng the Tramo-Seats method. Apart from the actual varables n the emprcal model we follow Kjosevsk (2013) and we nclude two dummy varables. Wth DUM we mark the global economc crss that has value 1 for the perod from October 2008 to December 2009 and 0 for all other perods. Wth DUM1 we mark the European debt crss that has value 1 for the perod from January 2011 to December 2012, and 0 for all other perods. Data are obtaned from varous sources. Data of the dependent determnant the M1 s obtaned from the webstes of the Central Banks for selected countres. Index of ndustral producton, the consumer prce ndex and nflaton are taken from the webstes of the State Statstcal Offces and the World Bank. Nomnal nterest rates and the exchange rate of domestc currency per euro are also taken from the webstes of the central banks of selected countres. III. 2 Methodology In order to analyze the determnants that are affectng the money demand n seven selected SEE countres, we use the panel data analyss. Among the man advantages of panel data, compared to other types of data, s that the approach allows the testng and adjustment of 88 Romanan Journal of Economc Forecastng XX (4) 2017

6 Are the Determnants of Money Demand Stable n Selected Countres the assumptons that are mplct n cross-sectonal analyss Maddala (2001). A number of econometrcans state that the use of panel data analyss can be very benefcal n a number of ways, ncludng: () panel data suggest that ndvdual countres etc. are heterogeneous; () panel data gve more nformaton, more varablty, less colnearty among other varables, more degrees of freedom and more effcency; () panel data can capture and measure effects that are not detectable n cross-secton tme-seres analyss, as well as provde a platform on whch to test more complcated behavoral models Hsao (1986). Before proceedng to econometrcs technques,.e. to choose whch method wll be used n our study, we need to verfy statonarty of selected varables. Accordng to Campbell and Perron (1991) standard unt root tests can have low power aganst statonary alternatves for the mportant cases. As an alternatve, recently developed panel unt root tests are appled. Therefore, n the paper we have appled, the IPS test Im, Pesaran and Shn (2003) and Fsher-Type test usng ADF and PP-test (Maddala and Wu,1999). These procedures allow for determnstc and dynamc effects dfferng across the panel members. Maddala and Wu (1999) further dsagreed wth the average ADF statstcs method and nstead, they employed a Fsher test to combne the p-values from unt root tests for each cross-secton. Ther tests have more advantages because: (1) the cross-sectonal dmenson can be ether fnte or nfnte; (2) each group can have non-stochastc and stochastc components; and (3) the tme-seres dmenson can vary for each cross-secton Baltag (2001). Also, the advantage of the Fsher test s that unlke the IPS test, t does not requre a balanced panel, and allows the use of dfferent lag lengths n the ndvdual ADF regresson. In our study we prefer Fsher-type tests but we also report the results of the IPS tests to provde an addtonal check for robustness. Test Varable LM1 Level (0.2577) Table 1 Unt Root Tests IPS ADF-Fsher Ch square PP-Fsher Ch square Frst Level Frst Level Frst Dfference Dfference Dfference (0.1898) IPIS (0.1734) INFS (0.0047) LEXRS NIRATE (0.0982) Source: Authors calculatons (0.1739) (0.0002) (0.0035) (0.0005) (0.0132) Table 1 shows the results of the IPS, and ADF Fsher panel unt root tests. Whle the exchange rate of domestc currences per euro, rate of nflaton and nomnal nterest rate are statonary n ther levels, the results are not consstent for the monetary aggregate M1 and ndex of ndustral producton. Namely, the unt root tests show that these varables have unt root at ther levels, and become statonary at the frst dfference. Romanan Journal of Economc Forecastng XX (4)

7 Insttute for Economc Forecastng In vew of ths problem, we perform panel ARDL model. Pesaran et al. (1999) suggested two estmators for the panel ARDL model: mean group estmator (MGE) and pooled mean group estmator (PMGE). The dfference between these two estmators s that the MGE seems to be more consstent under the assumpton that both slope and ntercepts are allowed to vary across country, whle PMGE s consstent under the assumpton of a long-run slope homogenety Ndambenda, Njoupouogngn (2010). Accordng to Pesaran et al., (1999) the basc assumptons of the PMG estmator are 1) the error terms are serally uncorrelated and are dstrbuted ndependently of the regressors, 2) there s a long-run relatonshp between the dependent varable and explanatory varables, 3) the long-run parameters are the same across countres. Furthermore, ths estmator s partcularly useful when the long run s gven by condtons expected to be homogeneous across countres whle the short-run adjustment depends on country characterstcs such as vulnerablty to domestc and external shocks, monetary and fscal adjustment mechansms, fnancal-market mperfectons, and relatve prce and wage flexblty Loayza (2004). The PMG estmator s suffcently flexble to allow for long-run coeffcent homogenety over only a subset of varables and/or countres. After these explanatons, we employ the PMGE estmator n ths study. Equaton 2 can be rewrtten for panel ARDL (p, q) form as follows: Y t p q jy, t j j X, t j t j1 j1 X where: t j s a dependent varable (LM1),, s the (k x 1) vector of of regressors for group and represents the country specfc-effects (fxed effects).ths model can be reparametrzed as a VECM system: t ( Y, t 1 X, t 1) p1 q1 yj, t j yjx, t j j1 j1 where: are the long-run parameters and are the error correcton coeffcent measurng the speed of adjustment towards the long-run equlbrum. The PMGE estmator restrcts the long-run coeffcents to be equal over the cross-secton, but allows for the shortrun coeffcents and error varances to dffer across groups on the cross-secton; that s, for all. Next, we examne the structural stablty of the error correcton model of money demand usng cumulatve sum (CUSUM) and cumulatve sum of squares (CUSUMQ) of recursve resduals test. These tests are commonly used by authors who explore the demand for money (Bahman-Oskooee and Shn, 2002). Both the cumulatve sum and the cumulatve sum of squares statstcs le wthn 5% of crtcal values, suggestng the long-run model stablty. IV. Emprcal Results Table 2 presents the results from the Pool Mean Group estmaton of ARDL. Maxmum lags are based on Akake nformaton crtera (AIC) t (3) (4) 90 Romanan Journal of Economc Forecastng XX (4) 2017

8 Are the Determnants of Money Demand Stable n Selected Countres Pool Mean Group Estmaton of ARDL(1, 1, 1, 1, 1, 1, 1) Table 2 Varable Coeffcent Std. Error t-statstc Probablty Long Run Equaton IPIS INF NIRATE LEXRSA DUM DUM Short Run Equaton COINTEQ D(IPIS) D(INF) D(NIRATE) D(LEXRSA) D(DUM) D(DUM1) C Country-specfc ntercept Albana Bosna and Hercegovna Bulgara Croata Macedona Serba Romana Source: Authors calculatons. The proposed long-run model suggests that demand for money M1 depends on the ncome (real ndustral output) and foregn exchange rate. Also, n the selected countres the second dummy varables DUM1 whch were ntroduced n the model have statstcally sgnfcant mpact of the demand for money. The coeffcent of ndex of ndustral output has a postve sgn, meanng that an ncrease of 1% n the ndustral output generates an ncrease of 0.008% n the demand for real M1. Ths s an expected result, snce based on the underlyng theory the ncome (GDP) elastcty of money demand should be postve. The coeffcent on the effectve exchange rate s negatve and statstcal sgnfcant. The results are consstent wth Kjosevsk (2013) and ndcate that after deprecaton of the exchange rate and f the publc expects further deprecaton, then the publc would demand more foregn currency and less domestc currency, leadng to a decrease n M1 money demand. Ths relatonshp between money demand and exchange rate s also consstent wth the work of Bahman- Oskooee (1996) whch argues that f a deprecaton of domestc currency results n an ncrease n expectatons of further deprecaton, the publc may decde to hold more foregn currency and less domestc money. The sgnfcance of ths determnant n the model s confrmed by the hgh degree of eurozaton n selected countres. Romanan Journal of Economc Forecastng XX (4)

9 Insttute for Economc Forecastng The results of the second dummy varable are n lne wth the result of Kjosevsk (2013). The coeffcent s statstcally sgnfcant and ndcates a hgher demand for real M1 money, solely as a result of the European debt crss. Ths result s not surprsng, because European debt crss and lack of confdence n the euro trggered a rebound n demand for domestc currency Kjosevsk (2013). Furthermore, we estmated the short-run model. Almost all estmated coeffcents of a shortrun dynamc model are small and not sgnfcant but dsplay sgns as expected from theory. Changes n ndex of ndustral producton, the nomnal nterest rate, rate of nflaton and the two dummy varables are nsgnfcant n the short run model. Gven ths result, the exchange rate s the only varable that holds both n the long and n the short run. The estmated coeffcent of the error-correcton term s hghly sgnfcant, valdatng the sgnfcance of the contegraton relatonshp of the common components n the short-run model for money demand. A negatve sgn mples that the money demand adjusts n the current month followng dsequlbrum n the prevous month. In other words, f there s an excess of money n the current month, n the next month the agents wll reduce ther money holdngs. In terms of sze, the adjustng parameter s small, whch means that ether the cost of dsequlbrum s reduced or the cost of adjustment s hgh. A comparson of the three further panel data studes applyng an error-correcton model by (Valadkhan 2008; Nautz and Rondorf 2010; Dobnk 2011) leads to the followng conclusons. Frst, ther estmated short-run dynamcs are also smaller than the long-run coeffcents. Second, Valadkhan (2008) supports that changes n the exchange rate are nsgnfcant n the short run. Thrd, the other short-run coeffcents estmated n ths study are smaller compared to the other three studes, except the mpact of changes of lagged real money, whch s wthn the range of both studes. The coeffcent of the error-correcton model mght be bgger n ths analyss because t measures the speed of adjustment towards overall equlbrum path. Namely, real money s adjustng faster to an equlbrum relaton whch reflects long-run money demand of not decomposed varables, whch s the case n study of Dobnk (2011), who n addton to the common factors also nclude the country-specfc dosyncratc components. For example, n the study of Valadkhan (2008) coeffcent of the error-correcton was -0.26, whle n the studes of Nautz and Rondorf (2010) and Dobnk (2011) the coeffcents were and , respectvely. Snce we are partcularly nterested n whether the estmatons acheved are stable over tme and therefore useful for forecastng purposes, we proceed wth CUSUM and CUSUMQ tests. The result of the test statstcs for evaluatng the vector stablty s presented n Fgure 1 and Fgure 2. Based on the CUSUM test results, we can say that the demand for M1 monetary aggregate n countres from Southeastern Europe s predctable and can be used for effectve mplementaton of monetary polcy. The result of CUSUMQ test shows that M1 demand functons were unstable n 2007 and 2008 and n the end of 2010 tll the end of These results may mply that the global economc crss and European debt crss dd have a sgnfcant mpact on the demand for money n the countres from Southeastern Europe. However, ths mpact on stablty was temporary, as stablty of M1 demand s not rejected after the end of Further, M1 stablty s not rejected n the whole-sample perod. These results confrm that the long-run money demand s stable n the countres from Southeastern Europe. 92 Romanan Journal of Economc Forecastng XX (4) 2017

10 Are the Determnants of Money Demand Stable n Selected Countres CUSUM Statstcs Fgure Source: Authors calculatons. CUSUM 5% Sgnfcance 1.2 CUSUMQ Statstcs Fgure Source: Authors calculatons. CUSUM of Squares 5% Sgnfcance V. Conclusons Ths research study examnes factors affectng money demand n seven countres from Southeast Europe (Albana, Bosna and Hercegovna, Bulgara Croata, Macedona, Romana and Serba) for the perod To our knowledge, ths s the frst emprcal study to analyze money demand n the selected countres from Southeast Europe. The research results ndcate that the man forces affectng money demand are real ncome, exchange rate of domestc currences per euro and European debt crss, whch explan the most varatons of money demand n the long-run, whle real ncome and nterest rate payable on domestc currency up to one month are sgnfcant n short-run. Long-run money demand functon ndcates capacty for relatvely quck adjustment and recovery of the equlbrum. Also, based on the presented results, the estmated coeffcents n the model are stable. These results show that despte the turbulent tmes n the regon n the past two decades, demand for money was relatvely stable n the analyzed perod. Romanan Journal of Economc Forecastng XX (4)

11 Insttute for Economc Forecastng The results obtaned n ths paper can provde useful polcy gudelnes to the central banks n ther quest for prce stablty. Namely the central banks of selected SEE countres should carefully montor the exchange rate as most mportant monetary polcy ndcator, because ths determnant s among the most mportant drvers of money demand both n the short and long run. Ths study does not face sgnfcant lmtatons, but ther removal wll certanly contrbute to broader results. The bggest constrant s the lack of avalable data on selected determnants for longer perods. The exstence of long tme seres of data would enable obtanng more accurate and more relable results. The future research on ths ssue should nclude other monetary aggregates such as the M2 and M4, n addton to money supply M1. The further research may also take nto consderaton other determnants, such as neffcency of the bankng system, nterest rates on long-term domestc and foregn currency deposts and nterest rates on treasury blls. Econometrc technques that researchers could use n the future regardng ths topc, should be ether the method of two or three least squares, the generalzed method of moments or the dynamc ordnary least squares-dols. References Anusc, Z., The Determnants of Money Demand n Croata and Smulaton of the Post- Stablzaton Perod. Croatan Economc Survey, 2, pp Arango, S. and Nadr, M., Demand for Money n Open Economes. Journal of Monetary Economcs, 7, pp Babc, A., The monthly Transacton Money Demand n Croata. Workng paper, Croatan Natonal Bank, pp Baltag, B.H. (2001), Econometrc Analyss of Panel Data, 2nd edton. Chchester. John Wley and Sons. Bahman-Oskooee, M. and Shabsgh, G., The Demand for Money n Japan: Evdence from Contegraton Analyss. Japan and the World Economy, 8, pp Bahman-Oskooee, M, Shn, S Stablty of the demand for money n Korea. Internatonal Economc Journa,l 16, pp Beyer, A., Modellng money demand n Germany, Journal of Appled Econometrcs, Vol. 13, Brand, C. and Cassola, N., A Money Demand System for the Euro Area. ECB Workng Paper No.39. Bruggeman, A The stablty of EMU-Wde money demand functons and the monetary polcy strategy of the European Central Bank. Manchester School, 68(2), pp Calvo, G. and Rodrgez, C., A model of exchange rate determnaton under currency substtuton and ratonal expectaton. Journal of Poltcal Economy, 85, pp Calza, A. Gerdesmeer, D. and Levy,J., Euro Area Money Demand: Measurng the Opportunty Costs Approprately. IMF Workng Paper No. 01/179. Campbell, J. Y. and Perron, P., Ptfalls and opportuntes: what macroeconomsts should know about unt roots. Macroeconomcs Annual, Natonal Bureau of Economc Research, pp Dobnk, F., Long-run Money Demand n OECD Countres. RUHR Economc Paper 94 Romanan Journal of Economc Forecastng XX (4) 2017

12 Are the Determnants of Money Demand Stable n Selected Countres Dreger, C. and Jürgen, W., M3 velocty and asset prces n the euro area. Emprca, 36, pp Ercsson, N. R., Emprcal modelng of money demand. Emprcal Economcs, 232, pp Hosen, Sharf-Renan., Demand for money n Iran: An ARDL approach. MPRA Paper No 8224, Unversty Lbrary of Munch, Germany Hsao, C Benefts and Lmtatons of Panel Data. Econometrc Revews, 4, pp Im, K. S. Pesaran, H. and Shn, Y., Testng for Unt Roots n Heterogeneous Panels. Journal of Econometrcs, 115, pp Karla, S., Inflaton and Money Demand n Albana. IMF Workng Paper Komárek, L. and Melecký, M., Demand for Money n the Transton Economy: The Case of the Czech Republc Warwck Economc Research Papers No.614 Kumar, S. Chowdhury, M. and Bhaskara, B., Demand for money n the selected OECD countres: A tme seres panel data approach and structural breaks. MPRA Paper No Loayza, N. and Rancère, R., Fnancal Development, Fnancal Fraglty and Growth. IMF Workng Paper. Maddala, G.S. and Shaowen, W., A Comparatve Study of Unt Root Tests wth Panel Data and a new smple test. Oxford Bulletn of Economcs and Statstcs, 61, pp Maddala, G.S., Introducton to Econometrcs, 3rd edton. Chchester, John Wley and Sons. Mankw, G.N. and Summers. H., Money Demand and the Effects of Fscal Polces. Journal of Money, Credt and Bankng, 18, pp Maravć, J. and Palć, M., Analza tražnje za novcem u Srbj, Radn Papr, Narodna Banka Srbje. Mark, C.N. and Sul, D., Contegraton vector estmaton by panel DOLS and long run money demand. Oxford Bulletn of Economcs and Statstcs, 65(5), pp Mehrotra, A., Demand for money n transton: Evdence from Chna s dsnflaton. Internatonal Advances n Economc Research, 14(1), pp Nautz, D. and Ulrke, R., The (n)stablty of money demand n the Euro Area: Lessons from a cross-country analyss. Dscusson Papers No Ndambenda, H. and Njoupouogngn, M., Foregn Ad, Foregn Drect Investment and Economc Growth n Sub-Saharan Afrca: Evdence from Pooled Mean Group Estmator (PMG). Internatonal Journal of Economcs and Fnance, 2(3), pp Omotor, D.G., Money Demand and Foregn Exchange Rsk for Ngera: A Contegraton Analyss usng ARDL Test. Journal of Economcs and Management, 18(1), pp Ozturk, I. and Al, A., The Demand for Money n Transton Economes. Romanan Journal of Economc Forecastng, pp Payne, J.E., Post stablzaton estmates of money demand n Croata: error correcton model usng the bounds testng approach. Appled Economcs, 35, pp Romanan Journal of Economc Forecastng XX (4)

13 Insttute for Economc Forecastng Pesaran, M. H. Shn, Y. and Smth, R. P., Pooled Mean Group Estmaton and Dynamc Heterogeneous Panel. Journal of the Amercan Statstcal Assocaton, 94(446), pp Skrabc, B. and Tomc-Plazbat, N., Evdence of the Long-run Equlbrum between Money Demand Determnants n Croata. Internatonal Journal of Human and Socal Scences, Vol.4, No 16, pp Sonje, V., Esej o monetarnm ekspermentma, l: kako zabrat monetarn rezm u kasnoj faz tranzcje. Economc Trends and Economc Polcy, 9(74), pp Stock, J. H. and Mark, W. K., A Smple Estmator of Contegratng Vectors n Hgher Order Integrated Systems. Econometrca, 61, pp Kjosevsk, J., The Determnants and Stablty of Money Demand n Republc of Macedona. Zbornk radova Ekonomskog Fakulteta u Rjec, 1(1), pp.. Valadkhan, A. and Mohammad, A., Demand for M2 n Developng Countres: An Emprcal Panel Investgaton. Dscusson Paper No 149. Valadkhan, A., Long- and Short-Run Determnants of the Demand for Money n the Asan-Pacfc Countres: An Emprcal Panel Investgaton. Annals of Economcs and Fnance, 9-1, pp Romanan Journal of Economc Forecastng XX (4) 2017