African Journal of Business Managemen Vol.3 (2), pp. 039-051, February, 2009 Available online a hp://www.academicjournals.org/ajbm ISSN 1993-8233 2009 Academic Journals Full Lengh Research Paper Modeling he impac of financial innovaion on he demand for money in Nigeria Gbadebo Olusegun Odularu and Oladapo Adewale Okunrinboye Deparmen of Economics and Developmen Sudies, College of Business and Social Sciences (CBS), Covenan Universiy, P.M.B. 1023, Km 10 Idiroko Road, Oa, Ogun Sae, Nigeria. Acceped 8 December, 2008 The demand for money is a very crucial in he conduc and deerminaion of he effeciveness of moneary policy. This sudy aemps o analyse wheher financial innovaions ha occurred in Nigeria afer he Srucural Adjusmen Programme of 1986 has affeced he demand for money in Nigeria using he Engle and Granger Two-Sep Coinegraion echnique. Though he sudy revealed ha demand for money conforms o he heory ha income is posiively relaed o he demand for cash balances and ineres rae has an inverse relaionship wih he demand for real cash balances, i was also discovered ha he financial innovaions inroduced ino he financial sysem have no significanly affeced he demand for money in Nigeria. Based on he resuls obained, a policy of aracing more paricipans (non-governmen) and privae secor funds o he money marke is necessary as his will deepen he marke and make he marke more dynamic and amenable o moneary policy. Key words: Money demand, deposi, narrow money, ineres rae, M, DD, Nigeria. INTRODUCTION AND PROBLEM STATEMENT The concep of financial innovaion is no an enirely new phenomenon in economics bu is pace over he las wo decades of he wenieh cenury has hrown up new challenges o perhaps one of economics mos holy debaed opics: he demand for money. The empirical sudy of he demand for money is one of he mos popular subjecs in applied economerics (Melnick, 1995). The search for a sable demand for money has been a very conenious issue since he grea inellecual debaes beween Keynesians and Moneariss of he 1960s and 1970s, as no demand for money model se forh by any of hese wo schools as well as heir conemporaries has wihsood he es of ime. The insabiliy of he demand for money in he 1970s and in he 1980s has been aribued primarily o changes in he performance of financial markes in he area of new financial producs arising ou of financial innovaions. Financial innovaion is becoming increasingly imporan in he 21 s cenury as i poses a serious problem for mo- *Corresponding auhor. E-mail: gbcovenan@yahoo.com. Tel: 2348034736572; 234-01 - 8179667 Ex. 2121. Fax: 234 01 8179668. neary policy, as wih new financial producs he abiliy of moneary policy o be effecive diminishes, as i changes one variable vial for effecive moneary policy; he demand for money. Financial innovaion refers boh o echnological advances which faciliae access o informaion, rading and means of paymen, and o he emergence of new financial insrumens and services, new forms of organizaion and more developed and complee financial markes (Solans, 2003). Wih new financial producs, conracionary moneary policy for insance, argeed a reducing excess liquidiy as economic agens can easily move money from less liquid holdings o more liquid packages being offered by financial inermediaries. In he process, undermining moneary policy, he reverse occurs vice-versa. In effec financial innovaion has also raised serious problems in he definiion and measuremen of money. This sudy seeks o replicae empirical works carried ou in he Wesern world in Nigeria o see if financial innovaion has had significan effecs in alering he demand for money in Nigeria. There is, and has always been, considerable dis-agreemen among economiss over wha deermines he levels and raes of growh of oupu, prices and employmen. The appropriae ool for macro-economic sabilizaion depends on he underlying heory in use. Keynesians would
040 Afr. J. Bus. Manage. would go for fiscal policy while moneariss would clamour for moneary policy. Moneary policy refers o he use of ineres raes, money supply and credi availabiliy o achieve macro-economic objecives. The use of moneary policy as a ool for macro-economic sabilizaion depends largely on he behaviour of he demand for money or real cash balances in he hands of economic agens. This brings in he demand for money funcion which expresses a mahemaical relaionship beween he quaniy of money demanded and is various deerminans; ineres rae, income, price level, credi availabiliy, frequency of paymens ec. The sabiliy of hese relaionships (elasiciies) is vial for deermining he appropriaeness and effeciveness of he ools or insrumens of moneary policy. The year 1973 is a waershed in he hisory of he various models and specificaions pu forh as regards he demand for money as Sephen Goldfeld published his analysis of pos-world War II quarerly daa on he demand for money using M1 definiion of money (ha is, currency in circulaion + demand deposis) and found ha he real income elasiciy of demand for real M1 balance was posiive bu less han one. The ineres rae elasiciy of demand is negaive and he demand for nominal cash balances is proporional o he price level. Hence he demand for money is he demand for real balances and no money illusion exiss (Miller and Pulsinelli, 1986) However, in recen imes he insabiliy of he previously sable money demand for money funcion has hrown up new sudies a is various deerminans and several oher frons have been explored by economiss and economericians alike. One of heses frons is financial innovaion which has blurred he disincion beween M1 and oher asses. I has blurred he various definiions of money M1, M2, M3 ec (ibid). In Nigeria, i has begun o hi home wih he recen recapializaion of he banking secor, wih he banks now bringing in new financial producs ha have combinaions of savings feaures, higher ineres earnings, easy wihdrawals and ransfers, wih increasingly close subsiues for money being inroduced by he day, good news for cusomers bu a hellish nighmare for moneary auhoriies. The main problem wih he sabiliy of he demand for money began wih one man s work in 1973, American professor; Sephen Goldfeld. Prior o 1973, he evidence ha had accumulaed from he large body of research done over he pos world war period was inerpreed as showing ha a sable demand funcion for money did, in fac, exis. In 1973, Sephen M. Goldfeld compued a demand for money funcion using quarerly poswar daa up o 1973. Alhough Goldfeld's resuls differed in several imporan ways from hose of he earlier lieraure, which were based mainly on annual daa, his preferred specificaion became he sandard formulaion. Goldfeld's findings held ha he quarerly demand funcion for money was mos sable when: (i) A narrow ransacions definiion of money was used. (ii) A shor-erm marke rae of ineres like he Treasury bill or commercial paper rae was used and when he rae on savings deposis was included. (iii) Measured income (real GNP) was used raher han permanen income or wealh. (iv) Lagged money was included o allow for incomplee adjusmen in he shor run. Bu saring in 1974, forecass from Goldfeld s equaion began o seriously over predic real money balances. When he equaion was re-esimaed wih daa including pos-1973 daa, The coefficien on he lagged dependen variable became very large (implying implausibly long adjusmen lags) and someimes i was having values greaer han uniy (implying money demand is dynamically unsable). Soon, he demand for money funcion had become "unsable" in he sense ha i had become more difficul o predic wihou serious accuracy errors. This problem of esimaing a sable money demand funcion as saed earlier has hrown up several lines of research. A line of inquiry sough o look a he pre-1973 agenda of empirical issues ha focused on ineres rae and inflaion. Anoher line of invesigaions have suggesed ha he rouble is linked o changes in he financial marke (Garcia and Pak, 1979; Goldfeld, 1976; Simpson and Porer, 1980). I is argued ha financial innovaions have led o deerioraion in he marginal relaionship beween real money balances and ineres raes. Economic lieraure saing ha financial innovaions affec he link beween ineres raes and money demand are no enirely new (Judd and Scadding, 1982). In fac, his is he building block of he heories of Gurley and Shaw (1960). They posied ha he definiion of money should include all oher asses which can serve as close subsiues for money, hen by aaching respecive weighs o his asse depending on he level of heir subsiuabiliy. Gurley and Shaw hypohesized a wide proliferaion of money subsiues which increased he ineres elasiciy of money demand. The paper has five secions, each dealing wih he differen aspecs of he sudy. Secion wo presens a review of relevan lieraure, while secion hree discusses he heoreical framework and model specificaion of he sudy. Secion four focuses on model esimaion and analysis of he regression resul and fifh secion concludes he paper wih relevan policy proposals. Objecives The broad objecive of his sudy is o examine he role of financial innovaion on money demand in Nigeria. The specific objecives of his sudy include: 1) To examine he degree of he relaionship beween financial innovaion and he money demand in he Nigerian economy. 2) To see how his relaionship affecs he effeciveness of moneary policies in Nigeria.
Odularu and Okunrinboye 041 3) To make policy recommendaions based on he findings of he sudy. Jusificaion This sudy is of umos imporance as i looks a several reasons accouning for he insabiliy of he demand for money which is very vial in deermining he effeciveness of moneary policy. These are as follows: I is imporan o know if money demand funcion is unsable as a resul of financial innovaion. Knowing his is vial o he relaionship beween ineres raes and aggregae expendiure as his is imporan for choosing insrumens for conducing moneary policy. If he demand for money is significanly affeced, hen he case for conducing macro-economic sabilizaion by regulaing he growh of he money supply and ineres rae changes may be seriously hreaened. As such his research projec will be of imporance o moneary policy makers (Cenral Bank of Nigeria). Research hypohesis In esing he relaionship beween financial innovaion and he demand for money wo hypoheses will be drawn is as follows: Financial innovaion has a posiive impac on he demand for money. SAP era financial secor liberalizaion policies favourably influenced he demand for money. Review of relevan lieraure Over he las wo decades, an enormous body of lieraure has documened he coninuing insabiliy of sandard economeric money demand specificaions and aribued he insabiliy o innovaion in he privae financial secor, Ireland (1995). The quesion of wheher he demand for money funcion is sable is one of he mos imporan recurring issues in he heory and applicaion of macroeconomic policy. Wha is being sough in a sable demand funcion is a se of necessary condiions for money o exer a predicable influence on he econo-my so ha he cenral bank's conrol of he money supply can be a useful insrumen of economic ha is moneary policy. Wha hen is he demand for money? The demand for money can be defined as he desire o hold money in liquid form raher han oher forms of wealh such as socks, bonds, ec. I ofen sems from hree main mo-ives, which are; ransacionary, precauionary and speculaive which are influenced by several facors; levels of income and wealh, raes of ineres, expecaions of eco- nomic agens and insiuional feaures of an economy. (Bannock e al., 1998). Pu differenly, i is he desire o hold cash or liquid asses raher han he equivalen in de- mand deposis. I is also known as liquidiy preference. The convenional money demand equaion has been one of he mos widely sudied relaionships in macroeconomics. I generally feaures real money balances being affeced by conemporaneous levels of real income as a proxy for ransacions, and a nominal ineres rae ha describes he opporuniy cos of holding money. The variables ha ener he demand funcion for money, and he definiion of he quaniy of money appropriae for he demand funcion, has received subsanial aenion in economic lieraure. Firs, here is he quesion of he consrain ha is imposed on money balances, wheher he appropriae consrain is a measure of wealh or income, or some combinaion of he wo. The second issue in mos lieraure has cenered on he imporance of ineres raes and price changes as argumens (independen variables) in he demand funcion. The hird issue is he quesion of he definiion of money balances. Is a more sable demand funcion obained if money is defined inclusive or exclusive of ime and/or savings deposis, and perhaps oher asses ha have value fixed in money erms? Tha is eiher M1 or M2. A rich radiion exiss on he esimaion of money demand in he Unied Saes han in any oher counry. Going by economic lieraure, he differences in he specificaion of he variables in he money demand funcion have produced imporan differences in implicaions or resuls. Tobin (1956) and Baumol (1952) separaely considered he ransacionary demand for money as a problem in capial heory and each obained a demand funcion for cash balances which depends on coss and yields. Boh Baumol and Tobin deduced from heir models ha here are economies of scale in holding ransacion balances. An income or wealh elasiciy less han uniy would confirm his implicaion. Friedman's empirical findings however sugges ha money is a "luxury" and ha he relevan elasiciy is in he neighborhood of 1.8 (Friedman, 1959). However going by empirical lieraure, mos economiss seem o accep Friedman's empirical resul in preference o hose of Baumol (1952) and Tobin (1956), hough here seems o be some debae over he specificaion of he variables in Friedman s money demand funcion. Specifically, Friedman's use of per capia permanen income combines wealh, ineres raes, populaion, and lagged income ino a single variable which combines and masquerades heir separae effecs. Tobin (1958) accorded he raes of reurn on financial and non-financial asses an imporan role in his heory of asse choice. Friedman's essay on he quaniy heory sresses a view of he quaniy heory as a heory of he demand for money. He uses bond and equiy yields as direc argumens in he demand funcion. Bu his empirical findings sugges he imporance of per capia permanen income and exclude ineres raes as direc argumens of he funcion or assign hem a role of second order of imporance. Bronfenbrenner and Mayer (1960
042 Afr. J. Bus. Manage. (1960) esimaed he separae effecs of wealh and ineres raes along wih income and lagged money balances. Their resuls show ha ineres rae, income, and lagged money balances are saisically significan by he usual ess, bu he wealh variable is non-significan. Anoher issue quie common in lieraure is he definiion of money iself which sill remains an open quesion. Gurley and Shaw (1960) suggesed ha moneary heory should be concerned wih a concep broader han he liabiliies of commercial banks. Friedman's empirical work is based on a concep of money ha includes he ime deposi liabiliies of commercial banks while Laane (1954), Bronfenbrenner and Mayer (1960) and ohers have been chiefly concerned wih money defined as he sum of demand deposis and currency. In erms of economeric work, Courchene and Shapiro (1964) idenified cerain dynamic problems wih early lieraure on he demand for money; difficulies wih auocorrelaion arising from he presence of he lagged money sock which possessed a significan role. Thus, he disincion beween he long-run and shor-run de-mands for money surfaced. Chow (1966) argued ha shor-run money demand adjused slowly oward long-run equilibrium; his sock-adjusmen specificaion has wea-hered significan sorms and remains he cenerpiece of many money demand sudies. The sock-adjusmen specificaion did no go unchallenged, however. Feige (1967) demonsraed ha a model of he long-run demand for money produces equaions similar o hose emanaing from he sock-adjusmen model wihou requiring slow adjusmen of money demand when he deerminans of demand are permanen, raher han curren, values. No disincion exiss beween long-run and shor-run demands for money. The long-run money demand depends on permanen (long-run) values of he deerminans of money demand. To he exen ha permanen variables can be modeled wih disribued lags of measured values, he inclusion of measured, raher han permanen, variables ino money demand mimics he sock-adjusmen specificaion. Second, he sock-adjusmen model implies unusual dynamic adjusmen when he money sock is exogenous. The deerminans of money demand mus overshoo heir long-run (permanen) values o clear he money marke on a period-by-period basis (Walers, 1966) and (Sarleaf, 1970). This demand for money specificaion has received renewed aenion in he 1990s wih economeric advances in he area of coinegraion. A large body of lieraure has emerged ha invesigaes long-run properies of he convenional money demand equaion for various counries. Evidence wih regard o a long run money demand relaionship in he Unied Saes, paricularly wih M1 during he poswar period, is mixed. Miller (1991), Hafer and Jansen (1991), Friedman and Kuner (1992), Sock and Wason (1993), and Norrbin and Reffe (1995a) as cied in Dukowsky and Aesoglu (2001) find lile suppor for coinegraion for he convenional saic money demand equaion wih M1. Some sudies, hough, have produced more posiive resuls, especially wih adjusmens in he basic specificaion. Hoffman and Rasche (1991) as cied in Dukowsky and Aesoglu (2001) find evidence supporing coinegraion wih a dummy variable o reflec a shif in he deerminisic rend in money demand during he 1980s. Baba, Hendry, and Sarr (1992) as cied in Dukowsky and Aesoglu (2001) provide suppor for a long-run relaionship wih an augmened model ha includes risk, inflaion, and a measure of he ineres rae spread. Hoffman, Rasche, and Tieslau (1995) as cied in Dukowsky and Aesoglu (2001) presen perhaps he mos supporive empirical findings. Wih a dummy varia-ble included hey obain evidence of a sable long-run saic money demand relaionship for M1 in five indusrial counries. A key o heir resuls is he imposiion of uniary long-run income elasiciy. So far a deliberae gap seems o have occurred in he period beween he 1960s and he 1990s as I skipped ino he 1980s. The omission is deliberae; his is so as he nex secion delves ino he work by Sephen M Goldfeld in he 1970s. In 1973, Sephen M. Goldfeld examined he issues sysemaically, using quarerly poswar daa up o 1973. Alhough Goldfeld's resuls differed in several imporan ways from hose of he earlier lieraure, which were based mainly on annual daa, his preferred specificaion became he sandard formulaion. The form of he Goldfeld equaion is shown below. The empirical esimaes of he equaion are: ln (M1/P) = α0 + α1 lngnp + α2ln RMS + α3in RSAV + αln( M1-1/P -1) Where, M1 = currency plus checkable deposis; P =he aggregae price level; GNP= real gross naional produc; RMS = a shor-erm marke rae of ineres; RSAV= rae of ineres on savings deposis. In summary, Goldfeld discovered ha he quarerly demand funcion for money was mos sable when: (i) A narrow ransacions definiion of money was used. (ii) A shor-erm marke rae of ineres like he Treasury bill or commercial paper rae was used and when he rae on savings deposis was included. (iii) Measured income (real GNP) was used raher han permanen income or wealh. (iv) Lagged money was included o allow for incomplee adjusmen in he shor run. One of he imporan sabiliy ess ha Goldfeld performed was o examine he abiliy of his equaion o forecas ouside he sample period. I showed no sysemaic endency o drif off in such forecass up o 1973, he year of his original sudy. Goldfeld (1973) hus discovered a single-equaion economeric model expressing he demand for real M1 as a sable funcion of real GNP and nominal ineres raes which did a remarkably good job of
Odularu and Okunrinboye 043 of characerizing quarerly U.S. daa during 1952-1972. This was confirmed boh by he accuracy of is forecass and by he inabiliy of a Chow es o rejec he hypohesis of parameer consancy across subsamples. Bu saring in 1974, forecass from his equaion began o seriously over predic real money balances. These forecass were ou-of-sample dynamic simulaions, which used acual ineres raes and income bu las period's prediced money balances as he lagged dependen variable. Three years laer, again, Goldfeld (1976) found ha by he same crieria of he accuracy of forecass and he resuls of Chow ess, he performance of his money demand equaion deerioraes markedly when he sample period is exended o 1976. In fac, money demand regressions coninue o be plagued by insabiliy when he sample runs hrough he presen day, wih heir forecass sysemaically over predicing acual real MI figures for he lae 1970s and under predicing acual figures for he 1980s. These simulaions showed a cumulaive drif from he firs quarer of 1974 o he second quarer of 1976 of nearly 9 percen. Moreover he error was almos enirely confined o he demand deposi componen of M1, which had an error of over 13%. The moneary equaions in he Federal Reserve Board's FMP Model gave similar resuls. This evidence of sysemaic over predicion of real money balances by he sandard money demand funcion suggesed ha he demand for money had shifed down. This possibiliy was aken o mean ha he demand for money had become "unsable" in he sense ha i had become more difficul o predic ex ane. Re-esimaion of Goldfeld's specificaions over he longer period confirmed ha he failure was more pronounced in he business han in he household secor's equaions. I is no surprising, herefore, ha hose insiuional innovaions, such as negoiable order of wihdrawal (NOW) accouns, money marke muual funds, credi cards, savings deposis of business and sae and local governmens, and checking accouns a muual savings banks, which have been credied wih he insabiliy were becoming more pronounced a hereabou he same ime (Garcia e al., 1979). However, a recurring debae ha ook place in Nigeria in he lieraure on he effeciveness of moneary policy o sabilize he Nigerian economy in erms of price sabiliy and subsequenly simulaing economic growh was on he naure and sabiliy of he demand for money funcion. This debae sared in he early 1970s amongs a group of scholars wihin he Lagos-Ibadan-Ife axis and was popularly called he TATOO debae, an acronym coined from he iniials of he major debaers of hose days. The famous TATOO debae of he 1970s involved five differren people: Tomori (1972), Ajayi (1976), Teriba (1974), Ojo (1974) and Odama (1974). I was Tomori (1972) who firs se ou o examine he facors ha influence he demand for money in he Nigerian economy. He ried o examine wheher here was a sable or unsable demand for money funcion and examined wha consiued a beer definiion of money in he Nigerian conex. He adoped a very simple linear model expressing money as a funcion of nominal/real GDP. Afer applying he OLS echnique, he made he following conclusions. 1. Income is a significan variable explaining changes in money demand 2. Income is a more imporan variable deermining money demand ha ineres rae. 3. The narrow definiion of money seems o perform beer ha he broad 4. The coefficien of ineres rae is no significan 5. Real income ends o show more significan relaionship ha nominal income. Ojo (1974) quesioned he work of Tomori especially his saisical mehodology. He was concerned mainly wih esablishing ha in a developing counry like Nigeria, characerized by an underdeveloped money marke, and lack of financial asses, he choice facing an individual is more beween money and financial asse. He consequenly specified and esimaed (using he OLS echnique) wo kinds of relaionship beween money and is deerminans. Firs, he specified real money balance as a funcion of curren nominal income and ineres rae. Second, following he insignificance of ineres raes, he specified he real money balance as a funcion of nominal and expeced rae of inflaion. According o Odama (1974), Tomori s model is devoid of any policy use in view of he fac ha he only insrumen (discoun rae) urned ou o be saisically insignifican. He also criicized Tomori in wo aspecs: 1) The formulaion of an alernaive model and he relevance of such a model for policy acions. 2) A modificaion of he saisical resul and conclusion hereof. Teriba (1974) observed ha Tomori s paper suffered several mehodological pifalls and inerpreaional defecs. According o him, reasury bills and ime deposi are he closes subsiues for demand for currency and ha adjusmen lag beween acual and desired cash balances is very close o zero, while income elasiciy of demand for currency is greaer han uniy. On demand for money deposi, he said he closes subsiue is ime deposi while savings is also a beer subsiue han reasury bills o demand deposi and he adjusmen period is fairly fas while ineres elasiciy of demand for deposi is very low and income elasiciy is also low. Ajayi (1976), in addiion o criicizing Tomori s paper (1972), provided answers o quesions like, he sabiliy of he demand funcion, adjusmen mechanism and calculaion of elasiciy for policy decision making. Using he narrow definiion of money (M1) he found ou ha income is abou 80.5% responsible for variaion in money demand bu when he used M2, he found ou ha income even has more impac on money demand which was like
044 Afr. J. Bus. Manage. 85%. When he inroduced he rae of ineres (on reasury bills), he go he wrong sign (posiive) and he value was saisically insignifican. He aribued his o he underdeveloped naure of he counry s money marke. The ineres elasiciy of money was very low so also he adjusing mechanism bu he income elasiciy was high. In his conclusion he suggesed he ineffeciveness of moneary policy in Nigeria. As lively as he TATOO debae was, he issue is sill inconclusive. Two broad evens seem o have dimmed he relevance of he debae carried ou in hose days. The firs is he array of new esimaion echniques (coinegraion) and several es procedures available o researchers since he debae feered in he early 1980s. The second is he developmen in he financial secor since he mid-1980s which may sugges some insabiliy in he demand for money funcion in Nigeria. The firs even has led o he re-examinaion of he naure and sabiliy of he demand for money funcion using error correcion mehods (Teriba, 1992; Nwaobi, 2004) as cied in Busari (2005). Several sudies have been carried ou on he demand for money in Nigeria hough no all made explici aemps a invesigaing he sabiliy of he money demand funcion as regards financial innovaion. Asogu and Mordi (1987) as cied in Busari (2005) examine he moneary secor in general o uncover some of he main deerminans of he money demand funcion. Ikhide and Fajingbesi (1998) as cied in Busari (2005) also examine wheher deregulaion of ineres rae in Nigeria under he Srucural Adjusmen Programme (SAP) of 1986 has had any significan impac on he demand for money in Nigeria. Sudies like Essen e al. (1996) as cied in Busari (2005) have dwel exensively on issues relaing o money demand in a liberalizing bu heavily indebed economy using Nigeria as case sudy. The sudy observed ha indebedness could signal o privae economic agens, he direcion of governmen fiscal and moneary policy which in urn influences he demand for money in he domesic economy. Audu (1988) as cied in Busari (2005) represens one of he firs pos-regulaion era effors o examine he sabiliy of money demand funcion. Using seleced Wes African counries, he sudy observed mixed resuls bu was quick o observe a sable money demand relaionship for Nigeria. The sudy by Nwaobi (2002) as cied in Busari (2005) has also made effors o examine he sabiliy of he demand for money in Nigeria. Using a relaively simple model ha specifies a vecor valued auoregressive process (VAR), he money demand funcion was found o be sable and he auhor suggess ha income is he appropriae scale variable in he esimaion of money demand funcion in Nigeria. In anoher sudy, Anoruo (2002) as cied in Busari (2005) explores he sabiliy of he M2 money demand funcion in Nigeria during he Srucural Adjusmen Program (SAP) period. In he sudy i was observed ha he M2 money demand funcion in Nigeria is sable for he sudy period. Furher i was argued ha M2 is a viable moneary policy ool ha could be used o simulae economic aciviy in Nigeria. Theoreical framework and model specificaion The convenional exbook formulaion of he demand for money ypically relaes he demand for real money balances (m = M/P), o he ineres rae, r, and some measure of economic aciviy such as real GNP (y = Y/P), where M = money holdings, P = he price level, and Y = gross naional produc. Thus, m = f ( r, y) Several heories have been pu forward o explain he equaion above. Perhaps he mos saisfying are hose of he ransacions view, in which he demand for money evolves from a lack of synchronizaion beween receips and paymens and he exisence of a ransacions cos in exchanging money for ineres-bearing asses (usually aken o be shor erm (Goldfold,1973). Of relevance o his research projec s model will be a selec few. This will serve as a base for he model o be specified. Keynes formulaed his heory of demand in his well known book, The General Theory of Employmen, Ineres and Money in 1936. According o him, he demand for money arises ou of is liquidiy; liquidiy refers o he converibiliy of an asse ino cash. He hen idenified hree moives for holding money. Transacion moive This arises ou of money s medium of exchange role and arises ou of he need for bridging he gap beween periodic receips and paymens. Keynes recognized boh he income moive for households and business moives for firms. Given he sociey s basic insiuional and echnical cusoms and pracices which govern income receip and he flow of expendiures, he ransacions demand depends on personal income and business urnover. I hus varies in direc proporion o changes in money income. Symbolically i is wrien as: L = k ( Y) Where; L : Transacions demand for money k : The fracion of money income sociey desires o hold as ransacion balances. Y: money income Precauionary moive This arises ou of unforeseen circumsances or expecaions regarding he uncerain fuure by economic agens. Keynes posied ha households someimes keep money for unexpeced coningencies such as medical
Odularu and Okunrinboye 045 emergencies or evens while firms held balances above ransacionary balances based on expecaions abou he economy e.g. a boom or depression. Keynes held ha he level of precauionary balances varied wih income and no ineres rae changes. Symbolically: L ( ) p = k p Y Where; L : Precauionary demand for money p k p : The fracion of money income sociey desires o hold as precauionary balances. Keynes usually lumped boh moives ogeher as hey were boh affeced by he same insiuional facors which he assumed given and fairly sable in he shor run adding o he fac ha hey were boh ineres inelasic. Mahemaically, L1 = L + Lp = k ( Y ) + k p ( Y ) = k( Y ) Where; L : Demand for acive balances 1 Speculaive moive This falls under he idle balances held by economic agens according o Keynes. He posied ha people hold or hoard money above heir acive balances for he purpose of being able o earn some form of gains by speculaing on bond prices. Since individuals knew ha an inverse relaionship exiss beween bond prices and ineres rae, hey held money for he opporuniy o parake in such speculaive aciviies so as o earn some form of ineres. According o Keynes, here hus exised an inverse relaionship beween speculaive demand for money and ineres raes. Funcionally, his is expressed as: L2 = f ( i) Where; L 2 : Speculaive demand for money i : ineres rae Keynes concluded by posiing ha he oal demand for money consiss of demand for acive balances ( L 1 ) and ha of idle balances ( L 2 ). Thus, L = L1 + L2 L = k( Y ) + f ( i) However, Keynes demand for money heory has been criicized for unnecessarily bifurcaing aggregae demand for money ino ransacions and speculaive demand. The ransacions demand for money depended on income le- vel (bu Keynes had assumed a consan relaion beween money holdings and income). His speculaive demand was based on porfolio approach which considered he yields of asses viz-a-viz heir compeiion wih money held in individuals porfolio. Again, he furher limied his analysis o wo asses; money and bonds. The combinaion of demand moives wih wo differen approaches is inconsisen (Paul, 2004). Furhermore on he heory of he demand for money, Baumol-Tobin Porfolio Formulaion of he Demand for Money is perhaps mos widely augh demand for money heory which seeks o explain he demand for money as a funcion of income and ineres raes. I arose as a defence by Keynesians o he inconsisencies of Keynes liquidiy heory. Is simples version is he so-called square roo of money holdings and i was pu forward by wo economiss. Tobin (February 1958), looked a he demand for money from he risk angle in his "Liquidiy Preference as Behavior owards Risk" paper while Baumol (1952) in his Transacions Demand For Cash: An invenory Theoreic Approach. His equaion is: M = ky 2 r 1 2 This implies ha nominal money holdings for cos minimising individuals vary direcly wih he square roo of planned nominal expendiures and inversely wih he square roo of marke ineres rae. I could also be expressed in real erms by deflaing each nominal variable above wih he price index. Mos empirical validaions of he above heory use he narrow money sock (currency plus demand deposis, M1) as he dependen variable ofen deflaed by he implici GNP deflaor. Income is defined as real GNP or GDP and he ineres rae is usually measured in wo ways: by he rae on commercial paper and by he rae on ime deposis. Several auhors regression specificaions base heir regressions using his syle. E.g. (Hafer and Hein, 1984) (Judd and Scadding, 1982) ec. Their explici specificaion usually is: M = β + β y + β r + β r + U c d 0 1 2 4 Where; Y: income c r : Rae of commercial paper (variable used as a measure of financial innovaion.) d r : Rae on ime deposis. M: moneary aggregae. : ime Usually he growh rae of money supply is used; alernaive specificaions use a lagged value of money supply as one of he regressors which necessiaes he use of
046 Afr. J. Bus. Manage. auo-correlaion correcive echniques. ln M = β + β ln y + β ln r + β ln r + β ln M + U c d 0 1 2 4 5 1 The resuling inference from heir heory is ha he demand for money is posiively relaed o income and inversely relaed o ineres rae. Model specificaion To successfully examine he impac of financial innovaion on he demand for money in Nigeria, he following model will be used for our empirical es. M = f ( Y, RTD, RTΒ, DSΑ P, CPI, M u)... (i) Where; 1, M: a moneary aggregae (in he case of his sudy M2) (Miller (1991) finds ha he naural logarihms of M2 and Income (proxied by real GNP) are coinegraed. However Trehan (1984) as cied in Miller (1991) Found ha in Wes Germany, real M1 and M2 were no coinegraed raher M3 was more appropriae. However in Nigeria, as M2 is more in line wih official moneary conduc, i has been adoped as my moneary aggregae, Anoruo (2002) as cied in Busari (2005)) Y: Income as capured by Gross Domesic Produc RGDP which seems o be mos appropriae proxy variable for capuring he level of ransacion (Alhough, some auhors conend ha wealh is a beer measure of capuring he level of ransacions. E.g. Laidler (1993), Melzer (1963) and Brunner and Melzer (1963) as cied in Goldfeld (1973), here appears o be a common ground in lieraure ha income could sill be used). RTD: Nominal Rae of ineres on ime deposis kep in commercial banks. Ineres rae measures he opporuniy cos of holding money ha is, he reward for paring wih liquidiy. I reflecs he degree of subsiuabiliy beween money and bonds or oher forms of financial asses. This is appropriae for our use of M2. RTB: Nominal Rae on Treasury Bills (The 4-6 monh Commercial Paper rae is ofen used as an indicaor for financial innovaion, due o difficuly in accessing i. We have decided o use a proxy as pu forward by Goldfeld (1973) ha is, he Rae on reasury bills. Alhough some oher auhors such as Miller (1991) propose he use of he dividend- price raio as a proxy; his is no readily available for he enire period under our scope). (A Proxy variable for he 4-6 Monh Commercial paper rae). DSAP: dummy variable o capure he financial innovaions ha have aken place since he sweeping reforms of he Srucural Adjusmen Programme (SAP) embarked upon by Nigeria in 1986 which led o changes in he financial sysem (Busari (2005) used a dummy variable o capure changes in he Nigerian financial secor since 1986 upwards ha is, pos SAP). CPI: Consumer Price level (Miller (1991) included price level in esimaing he demand for money as he found i highly significan). M -1 : one period lag of M : Time period u : Sochasic random erm. In a more explici and economeric form, M l = β 0 +β 1Y +β 2RTD +β3rtβ +β4dsα P +β 5CPI +β 6 M + U..... (ii) Represening he above equaion in a log-linear form, log M =β +β log Y +β log RTD +β log RTΒ +β DSΑ P+β log CPI +β log M + U... (iii) 0 1 2 3 4 5 6 1 A model of demand for money should esablish a sable relaionship beween demand for money and he facors influencing i. Theoreically, he demand for money is hypohesized o be an increasing funcion of some measure of income or wealh. The coefficien of real income β ) should be posiive since real income demanded ( 1 rises wih he level or value of ransacions. The coefficiens β2 and β 3 of he wo raes RTB and RTD respecively are expeced o be negaive. This is because of he inverse relaionship ha exiss beween ineres raes and real cash balances. The esimaion echnique o be used in he above model is he coinegraion echnique which is an improvemen on he classical Ordinary Leas Squares echnique. One reason for he choice of his echnique is ha, firs, i is generally argued ha mos economic series are nonsaionary ha is, have a srong rend over ime. By nonsaionary, we mean ha he variables do no have a mean which is consan over ime and as such direc applicaion of leas squares echnique could give spurious resuls. This causes he resuls of mos OLS regressions o be saisically invalid and difficul o inerpre in a heoreical conex (Melnick, 1995). Coinegraion, error-correcion modeling involves four seps. Though in a hin line separaes seps wo and hree which necessiaes heir merging. Firs, deermine he orders of inegraion for each of he variables under consideraion; ha is, difference each series successively unil saionary series emerge. Second, esimae coinegraion regressions wih ordinary leas squares, using variables wih he same order of inegraion. Third, es for saionary residuals of he coinegraion regressions. Finally, consruc he error-correcion models. (Miller, 1991) These seps are furher explained as follows.
Odularu and Okunrinboye 047 Deermining he order of inegraion The mos popular approach is o use wha are called augmened Dickey-Fuller, or ADF, ess. They were proposed originally by Dickey and Fuller (1979) under he assumpion ha he error erms follow an Auoregressive process of known order. Basically wha his sep seeks o do is esablish wheher a paricular ime series is saionary or non-saionary. If non-saionary hen is has o be differrenced eiher once or wice. To carry ou his es, we es he null hypohesis of a difference saionary agains he alernaive hypohesis of a rend saionary. Tha is: 0 ( ) ( ) H :Y ~ I 1 1 H :Y ~ I 0 The es saisics of he esimaed coefficien of y is hen used o es he null hypohesis ha he series is non saionary. If he absolue value of he es saisics is higher han he absolue value of he criical T value (which could be a 1, 5 or 10%), hen he series is said o be saionary Therefore, we rejec he null hypohesis. If he null hypohesis canno be rejeced hen y canno be saionary ha is, y is non saionary. I may be of order one hais, I (1) or order wo ha is, (2) or have an even higher order of inegraion. This will be revealed by differrencing y ill i becomes saionary. Co inegraion regression The second sage proceeds o obain he coinegraion (error correcion) vecor in he regression equaion using OLS. Tes for saionary residuals of he co inegraion regressions Here, we es if he residuals ( u ) are saionary. This involves examining he esimaed residuals from he regression direcly by performing a uni roo es of he ADF ype. Once i is discovered ha he residuals here are saionary, hen i is possible ha our variables are coinegraed in he long run. Consruc he error-correcion models (ECM) The final sage in he model building process requires he consrucion of error consrucion models. This involves regressing he firs difference of each variable in he co inegraion equaion ono lagged values of he firsdifferences of all of he variables plus he lagged value of he error-correcion erms (ha is, he error erm from he co inegraion regression). (Miller, 1991) The ECM incurporaes he full (shor-run) dynamics of he saed model. A his sage, all he convenional saisical ess of significance are considered o be appropriae. The purpose of he ECM is o swich o a shor run model. The ECM indicaes he speed of adjusmen from shor run equilibrium o he long run equilibrium sae. The greaer he co-efficien of he parameer, he higher he speed of adjusmen of he model from he shor run o he long run. Daa analysis and Inerpreaion This secion presens he resuls obained in he sudy. Table 1 shows he uni roo es of he variables a levels while Table 2 shows he uni roo es of he variables a he firs difference. Table 3 shows he uni roo es of he residual obained from he ordinary leas square regression while Table 4 shows he error correcion model. RESULTS DISCUSSION Here, a series saionariy ess was carried ou on all variables. This es is paramoun due o he non- saionariy feaure of mos annual ime series daa. This was carried ou using he Augmened Dickey Fuller (ADF) es saisics. Table 1 above showed ha all he variables were no saionary in levels. This can be seen by comparing he observed values (in absolue erms) of he ADF es saisics wih he criical value (also in absolue erms) of he es saisics a he 5 and 10% level of significance es. Therefore, he null hypohesis is acceped and i is sufficien o conclude ha here is he presence of uni roo in he variables a he 5 and 10% levels of significance. Following from he resuls obained above, all he variables were differenced once he ADF es was conduced on hem. Table 2 shows he resuls obained. A close look a he able reveals ha all variables are saionary a he 5% level of significance excep M2, which was significan a he 10% level of significance. Also, CPI was no saionary when a rend was applied o i in is firs difference form. Thus, on he basis of he resuls in Table 2, he null hypohesis is rejeced and i is safe o conclude ha he variables are saionary. This implies ha he variables are I (1) series, ha is, inegraed of order 1. Co-inegraion es Here, wo seps ake place. Firsly, an ordinary leas squares regression was carried ou using he variables in our model specified wih he exclusion of DSAP. This hus convers he form of our model o: Y Log M =β +β log +β log RTD +β log RTΒ +β log CPI +β log M + U 0 1 P 2 3 5 6 1
048 Afr. J. Bus. Manage. Table 1. Uni roo es a levels. VARIABLE ADF (UNTRENDED) ADF(TRENDED) LOGM2-0.830393-2.504330 LOGRTD -1.378753-0.424075 LOGRTB -1.116305-1.615467 LOGCPI 0.667872-1.843997 LOGRGDP 0.138303-1.594776 NOTE: ADF represens Augmened Dickey Fuller. Table 2. Uni roo es a firs difference. VARIABLE ADF (UNTRENDED) ADF (TRENDED) DLOGM2-3.200968* -3.493219** DLOGRTD -3.373649* -4.143771* DLOGRTB -5.233463* -5.168919* DLOGCPI -3.184444* -3.143145 DLOGRGDP -3.678475* -3.646344* Noe: *Saionary a 5 percen; ** Table 3. Uni roo es of residuals. Saionary a 10 percen. VARIABLE ADF TRENDED ADF UNTRENDED RESID -4.971036-4.845470 Table 4. Shor run error correcion model. VARIABLE COEFFICIENT STD. ERROR T.STAT C 0.118694 0.047476 2.500101 DLOGCPI 0.301801 0.080615 3.743719 DLOGRGDP(-3) 1.424761 0.316660 4.499345 DLOGRTD 0.114115 0.0773840 1.474653 DLOGRTD(-3) -0.190934 0.086969-2.195435 DLOGRTB -0.044140 0.075042-0.588204 DLOGM2(-1) 0.459597 0.187357 2.453051 DLOGM2(-3) -0.280104 0.113231-2.473728 DSAP -0.034131 0.031105-1.097300 EC(-1) -0.601865 0.252520-2.383429 R-squared: 0.797900; Adjused R-squared: 0.711285; D-W saisic: 2.190635; F-saisic: 9.212085. Our resuls are hus presened in he appendix. The residuals from he above regression are hen saved and esed for saionariy (using he ADF mehod) o prove if he variables are coinegraed in he long run be-fore an error correcion model can be pu forward. Given ha he residuals from he co-inegraing regression are saionary, hen i is possible for coinegraion o ake place among our variables. The resul of he uni roo es of he residuals is presened in Table 3 below. From he able above, he residual was saionary a 5% level of significance. As a resul of his, one can righly say ha here is a long run relaionship beween all he variables used in he demand for money funcion. Given his resul, i is now possible o proceed o esimae an error correcion model, o reconcile he shor-run behavior of he variables in he specified model wih heir long-run behavior. The criical ADF es saisic a levels for he residual is (-2.957110 and -2.617434) Unrended and (- 3.557759 and -3.212361) rended for (5 and 10%, respecively). Error correcion presenaion This is he las sage in he coinegraion process and involves esimaing our previous equaion however his ime wih our error correcion facor as a dependen variable. This involves regressing he firs difference of each variable in he coinegraion equaion ono lagged values of he firs-differences of all of he variables plus he lagged value of he error-correcion erm. The resul obained is presened below. A close inspecion of he able above indicaes ha he error correcion model has a high coefficien of deerminaion. This can be seen from R-squared of 79% and he adjused R- squared of abou 71%. The R-squared shows he percenage of variaion in he dependen variable ha was accouned for by variaions in he explanaory variables. The finess of every regression resul is based on is R-squared. The F-saisic value of 10.84832 shows ha he overall model is saisically significan a 1 and 5% levels of significance. This is because i is greaer han he criical values of 2.57 and 3.79 a 1 and 5% respecively. This means ha all he explanaory variables simulaneously explain he variaions in he real demand for money. Also, all our variables are saisically significan a 95% confidence inerval wih he excepion of DSAP, RTD and RTD. Furhermore, he DW saisic, which is a measure of auo correlaion, shows ha he error correcion model is free from he problem of serial correlaion due o is value (2.19). As a resul of his, an error correcion model esimaed can be confidenly relied upon for making inferences on role of financial innovaion on he demand for money. The EC, which is he error correcing erm in he model, indicaes he speed of adjusmen from shor run equilibrium o he long run equilibrium sae. The greaer he co-efficien of he parameer, he higher he speed of adjusmen of he model from he shor run o he long run. In he model, one would noice ha he ECM (EC above) is saisically significan a 5%. This shows ha here is a dynamic adjusmen from shor run o log run. The coefficien of he ECM is 0.60. This indicaes ha 60% of he errors in he shor run are correced in he long run. As regards he behaviour of our explanaory variables wih respec o he regress and, a posiive relaionship exiss beween he hird lag of RGDP and M2 confirming
Odularu and Okunrinboye 049 economic heory (Keynes e al) as regards he relaionship beween income and he demand for cash balances. Secondly, ineres rae also conforms our apriori expecaion in ha, he sign of is coefficien is negaive implying an inverse relaionship beween he demand for cash balances and he rae of ineres. The hird variable in our model CPI also aligns wih heory in ha i has a posiive sign. The variable ofen used o capure financial innovaion in mos empirical lieraure is he 4-6 monh commercial paper (being proxied by he reasury bill rae in our model) he co-efficien of i in our model is negaive (- 0.044140) which confirms wha heory says. However, i is no saisically significan. I was no dropped as his affeced our Akaike informaion crierion; raising is value. This could be raced o he poor developmen of he money marke where he reasury bill rae rules. This hus leads o a conclusion ha financial innovaion has had an impac hough no significan impac on he demand for money in Nigeria under he period of our scope. The innovaions ha have occurred given he massive financial secor reforms ha characerized he SAP era have had an impac on he demand for money hough his is no significan hence, our resul allies wih ha of Busari (2005). Worhy of noe is ha hough a presen an appreciable level of innovaion seems o be aking place a presen, i is pos consolidaion which is ouside he scope of his research projec. H 0 : SAP era financial secor liberalizaion policies have had no impac on he Demand for money. H 1 : SAP era financial secor liberalizaion policies have had an impac on he Demand for money. In order o invesigae wheher he financial secor liberalizaion during SAP in he Nigerian economy has affeced he real demand for money, a dummy variable was included in he error correcion model. The dummy variable was no significan a he 5% level however is exclusion raised he value of our Akaike Informaion Crieria and affeced he values of some of our regressors. Inspie of his, is co-efficien ook on a negaive sign. This means ha Srucural Adjusmen Programme which saw o sweeping changes in financial secor has no led o some financial innovaions which indirecly or direcly affec he demand for money. Conclusions and Recommendaions This research projec has looked a he demand for money and how i has been affeced by financial innovaions in he financial secor of Nigeria arising ou of he Srucural Adjusmen Programme (SAP) of 1986. The erm financial innovaion refers o anyhing which ensures greaer access o informaion, quicker means of carrying ou ransacions and greaer ease of liquidiy wih lower risk. I needs no be a echnological innovaion as Solans (2003) poined ou even he euro is a financial innovaion, I has boh reduced ransacion coss and eliminaed exchange rae risks, and has also aced as a caalys for a number of improvemens in various areas ha have helped o creae a more efficien financial sysem in he euro area as a whole. However, is effec on he demand for money is wha aroused so much ineres o i among economic scholars. Of paricular ineres has been is effec on he sabiliy of he demand for money, in ha if is impac on he demand for money is significanly large, hen he effeciveness of moneary policy may be seriously hreaened. In order o order o ascerain his impac, a model was specified and esimaed using he coinegraion echnique mehod. Daa for he analysis was aken from 1970-2004. Main findings and heir implicaions Afer carrying ou appropriae analysis using our model i was discovered ha on he basis of individual ess of significance, all he slope coefficiens were individually saisically significanly differen from zero wih he excepion of DSAP, RTD and RTB which failed he es of significance a he 5% level. Hence our major findings are as follows: 1) Lagged Ineres on ime deposis is negaively relaed o he demand for money. 2) Lagged Treasury bill rae is negaively relaed o he demand for money. 3) Real income is posiively relaed o he demand for money. 4) Price level is posiively relaed o he demand for money. 5) Srucural Adjusmen Programme has had no indirec effec on he demand for money via financial innovaion. In view of he above findings, he following are possible implicaions arising: 1) The low ineres elasiciy of our demand for money is indicaive of underdeveloped naure of he money marke in Nigeria. The money marke paricularly he reasury bills are dominaed by governmen (he Cenral Bank) wih he end resul being ha he marke lacks he deph and flexibiliy ha i migh have had wih he presence of a diversified paricipan base. Hence in a model, expressing he demand for money as a funcion of Treasury bill rae, i is definiely going o be significan. This is also indicaive of he ill developed naure of our financial sysem. Keynesian docrine holds ha for he smooh funcioning of his liquidiy preference heory he money marke mus be well developed. 1) Income level is a primary deerminan of demand for money by economic agens in Nigeria. 2) The analysis also shows ha he amosphere is no conducive for he effecive use of moneary policies, how-