PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS: A CASE STUDY OF FIJI

Transcription

1 JOURNAL OF ECONOMIC DEVELOPMENT 17 Volume 31, Number 2, December 26 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS: A CASE STUDY OF FIJI T.K. JAYARAMAN AND CHEE-KEONG CHOONG * The Universiy of he Souh Pacific and Universii Tunku Abdul Rahman Growing public deb of Fiji has been causing concerns all around. As par of counercyclical measures, he Governmen sepped up public expendiure from 21 in response o he adverse consequences of he 2 civilian coup, which winessed a decline in invesor confidence, resuling in a seep fall in privae secor invesmen. Expansionary fiscal policy measures in he annual budges of 21 o 24 as well as unforeseen naural disaser managemen expendiures have pushed he raio of ousanding public deb o naional oupu beyond he level of 5%. This paper seeks o examine he nexus beween deb and growh in Fiji. Keywords: Public Deb, Governmen Expendiure, Growh, Coinergraion, Causaliy JEL classificaion: H5, H6 1. INTRODUCTION Focusing aenion on public deb in indusrial counries and in emerging marke economies, he Inernaional Moneary Fund (IMF) s World Economic Oulook:23, (IMF 23a) noed ha deb levels in emerging economies, which were abou 7% of heir gross domesic produc (GDP), were higher han hose in indusrial counries and observed ha large public deb hampered economic aciviy. Higher axes, which are required o finance he growing deb, exercise upward pressure on real ineres raes crowding ou privae invesmen, hereby affecing long-erm growh. When a governmen finds iself unable o finance is annual fiscal deficis, i is forced o conrac spending or raising revenues, ofen a a ime when fiscal policy needed o be concenraed on sabilizing he economy. If he governmen could no ake hese acions, a deb crisis would ensue. According o IMF (23), he prevailing levels of public deb in emerging economies were no currenly susainable-ha is, coninuaion of pas fiscal policies would no be sufficien o enable he debs o be repaid in he fuure. The susainable level of deb for a ypical emerging economy, based on pas * The auhors are graeful o an anonymous referee for very helpful commens and suggesions.

2 18 T.K. JAYARAMAN AND CHEE-KEONG CHOONG fiscal performance has been esimaed by IMF a 25% of GDP, while esimaes of fiscal policy reacion funcions indicaed ha emerging marke economies as a group have failed o respond in a manner consisen wih ensuring fiscal solvency once he public deb exceeds 5 percen of GDP (IMF 23). Wih excess liquidiy in he sysem in he economy due o poor invesmen climae, domesic borrowing by he Governmen of Fiji during he pas five years for financing is annual fiscal deficis was no found difficul. For 25, governmen deb was esimaed a F$ 2,487 million or 53.4% of GDP. I is expeced o be F$ 2,689 million or 54.2% of GDP in 26. As for he deb porfolio in 25, domesic deb was 93% of oal deb while exernal deb was small a 9%. In recen monhs, here have been monhs increasing references o Fiji s growing public deb by official sudies by inernaional agencies (UNESCAP 26, ADB 26) as well as he counry s cenral bank (Reserve Bank of Fiji 26). Furher, here have been concerns expressed by expers (Prasad (26), Reddy (25), Shah (23), Chand (23)) ha he burden imposed by growing deb in erms of rising annual ineres paymens as well as disorions in he economy creaed by inflaionary endencies would ulimaely hamper growh. There is only one sudy available so far on deb and growh in Fiji (Chandra and Jayaraman (25)), which employed a bivariae model. The objecive of his paper seeks o underake more rigorous a quaniaive analysis wih a mulivariae model, wih a view o asceraining wheher pas public deb has led o economic growh in Fiji. The sudy uilizes he annual daa for a 34-year period from 197 o 23. The remainder of he paper is organized as follows. The second secion presens rends in Fiji s public deb over he las hree decades; he hird secion oulines he mehodology adoped for he sudy; he fourh secion presens he resuls; and he fifh and final secion offers some conclusions wih policy implicaions. 2. TRENDS IN PUBLIC DEBT IN FIJI In Pacific island counries (PICs), exernal aid in erms of grans from bilaeral sources in he pas had been financing capial expendiures of governmens o a subsanial exen. In some PICs, exernal aid had also been financing heir curren budges, by providing suppor for wages and salaries and oher housekeeping expenses. Fiji (Table 1) annually receives exernal grans only o he exen of 2% of GDP, he leas in he region, wih he resul is annual fiscal deficis in recen years have o be financed by increased public borrowing.

3 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS 19 Table 1. Fiji: General Key Indicaors Surface Area (km 2 ) 18,27 Populaion (25) 835, Toal GDP a curren prices (US $ million) 22 1,75 Per capia GDP in curren price (US $) 2,36 Human Developmen Index (Rank) 81 Aid per capia (US $) Aid per capia (% of GDP) Source: Commonwealh Secrearia (25), ADB (24), UNESCAP (24) Fiji s oal deb, comprising domesic and exernal deb, during he pas wo decades has been low. In fac, he counry made conscious effors during he lae 198s and early 199s o reduce exernal deb by early reiremen of loans from inernaional agencies, including Asian Developmen Bank. Fiji, being one of he counries belonging o lower middle-income group, was no allowed he faciliy of concessional loans a annual 1% rae, which were available o oher PICs. Therefore, wih he hen prevailing comforable inernaional reserves equivalen o six-monhs impors of goods and no facor services, Fiji fel appropriae o reire he exernal deb in advance, raher han pay he loan insallmens and ineres from fuure expor earnings. Wih budge deficis also being close o ye anoher benchmark of 3% of GDP, i was possible o mainain low levels of domesic deb in he 199s around 45% of GDP (Table 2). Table 2. Fiji: Public Deb and Growh (197-2) Toal Domesic Exernal Toal Domesic Exernal GDP Deb Deb Deb Deb Deb Deb Growh Year F$ Mill F$ Mill F$ Mill % Of GDP % Of GDP % Of GDP Rae (%) (Ave) (Ave) (Ave) (es.) NA NA 54.2 NA NA 2.7 Source: Reserve Bank of Fiji (26, 25)

4 11 T.K. JAYARAMAN AND CHEE-KEONG CHOONG From 2 onwards, Fiji sepped up is domesic borrowing o finance is annual fiscal deficis, he major lender being Fiji Naional Providen Fund (FNPF), which is he counry s sauory body collecing conribuions from employers and employees in he formal secor a a legally sipulaed rae from he monhly salaries o be reurned o employees wih ineres afer heir reiremen. While hese funds are readily available for invesmen by FNPF in income yielding projecs in he economy, here has been anoher jusificaion for governmen as well. Invesor confidence was low soon afer he 2 coup and he expecaions ha he privae secor would soon recover did no maerialize. I was herefore considered appropriae by governmen o boos invesmen by incurring fiscal deficis for a while unil privae confidence was resored. Besides borrowing from he public for financing is own budgeary deficis, he governmen guaraneed he borrowing by sae enerprises as well. In 23, Fiji s public deb, including coningen liabiliies, was a lile above 6% of GDP (Narube (24)). A recen World Bank sudy (Gill and Pino (25)) idenified hree reasons why public deb migh be beer han axaion, which were caegorized ino hree: (i) iling; (ii) smoohing; and (iii) sabiliy. Under he firs, i would be more equiable if a counry can finance projecs of long gesaion naure hrough deb, as such projecs benefi fuure (richer) generaions han hrough axing he curren (poorer) generaion. The second reason is raising and lowering axes frequenly migh enail efficiency losses and generae economic uncerainy and hence, deb allows a more efficien manner for conducing couner-cyclical polices and for meeing emergency spending needs. The hird reason is deb ensures sabiliy, since i avoids reliance on prining money. The laer involves high and volaile inflaion, obscuring informaion conen of relaive prices and hereby huring invesmen. However, as Gill and Pino remind us, deb is nohing bu posponed axaion, since i has o be re-paid; and reurns from every infrasrucure projec funded by deb, regardless wheher i is physical or social infrasrucure, have o include no only he user fees bu also higher fuure axes. An IMF sudy (24) on Fiscal Adjusmen in IMF Suppored Programs acknowledged ha misplaced opimism abou privae secor recovery would lead o an undersaemen of he need for a more counercyclical fiscal sance ha is oo igh (Selowsky (24)). Taking he cue from he above, one would conclude ha in rerospec, Fiji s fiscal deficis during would appear o be par of counercyclical measures o compensae he loss of fall in privae invesmen. This raises an imporan quesion abou he naure and composiion of oal public expendiure. Governmen prepares he budge under wo pars: recurren and capial expendiures. While he recurren budge is nohing bu housekeeping expendiure dominaed by wages and salaries, he capial budge relaes o asse creaing invesmen aciviies. The raio of curren expendiure o capial expendiure in PICs, including Fiji has been around 8 o 2. Wheher public expendiure was underaken purely as counercyclical measure o mee he projeced fall in demand or simply as a poliical commimen o sep up growh hrough sae sponsored schemes, he resulan rise in public deb due o fiscal expansion given he revenue posiion, is cerainly a maer of

5 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS 111 concern. The concern is abou he way he growing annual ineres burden would be borne by he governmen. Ineres paymens have o be effeced ou of he primary balance defined as surplus of curren revenues over curren expendiure, excluding ineres paymens. Sahay (24) in her sudy on he island counries in he Caribbean region, wih which PICs have been sharing many commonaliies (Fairbairn and Worrell (1995)), has shown ha during , all he highly indebed counries in he Caribbean, excep Jamaica, were failing o generae primary surpluses. The resul has been furher deerioraion in heir overall fiscal balances, leading o furher rise in deb levels, as ineres paymens have o be financed by recourse o addiional public borrowing. If he public borrowing has been resored o for financing economic growh hrough income generaing invesmen projecs as well as infrasrucure projecs supporing privae secor invesmen, he counry s revenue poenial would also rise, as growh in ax revenues is posiively relaed o economic growh (Jayaraman (26)). Only hen, rise in public deb could be defended on he ground ha ineres burden can be borne wihou any addiional effor for raising ax revenues. The key poin of our examinaion in his paper is, herefore, wheher public deb has conribued o economic growh in Fiji. 3. DATA AND METHODOLOGY For his sudy, we use annual real gross domesic produc (RGDP) and sock of real public deb (DEBT) covering a 34-year period (197-23) o examine he deb-growh nexus in Fiji. Since a bivariae model is likely o suffer from deficiencies arising from he omission of oher relevan explanaory variables, we propose o include wo addiional explanaory variables, which we consider essenial for explaining he growh in deb. One is real Treasury bill rae (RTB) in percen, represening he cos of borrowing and he oher is raio of wages and salaries o oal expendiure (WSTE), represening he allocaional naure of oal budge funds. We expec ha lower cos of borrowing as represened by a fall in RTB, which reflecs moneary policy, should encourage governmens o borrow more. Governmens are generally relucan o resoring o rise in ax raes and fees and user charges, as addiional ax effor hurs re-elecion chances. Addiionally, ax cus and oher discreionary ax exempion measures by minisers and bureaucras lead o eiher reducion in or sagnaion of revenues. On he oher hand, rise in recurren coss including wages and salaries and oher consumpion expendiures along wih minimum criical capial expendiures ofen lead governmens o explore avenues of financing fiscal deficis. Deb financing is he easy way ou. I is, herefore, hypohesized ha he higher he proporion of recurren expendiures cos in he oal budgeed expendiure, which is proxied here by WSTE, higher would be he deb. Several sudies have shown ha rise in consumpion expendiures of governmen such as wages and salaries has negaive impac on privae invesmen, hereby adversely affecing economic growh

6 112 T.K. JAYARAMAN AND CHEE-KEONG CHOONG (Barram and Ward (1993), Alesina e al. (1999), and Giannaros e al. (1999)). In addiion o hese wo explanaory variables, RTB and WSTE, a dummy variable represening poliical insabiliy was also added. I akes he value of zero for years prior o and afer 1987, one for 1987, he year of wo coups and 2, he year of civilian coup. Accordingly, he equaion o be esimaed is: RGDP = β + β1debt + β2rtb + β3wste + β4dum + ε, (1) ε where is he Gausian error erm and all variables are expressed ino he naural logarihm, and DEBT represens real public deb. For examining he long-erm relaionship beween economic growh and public deb, we resor o he auoregressive disribued lag (ARDL) model proposed by Pesaran, e al. (21). The ARDL procedure has become increasingly popular in recen years for several reasons: Firs, he echnique is more appropriae o be used in esing he long run relaionship beween variables when he daa are of a small sample size (Pesaran, e al. (21)). Second, here is no resricion imposed on he order of inegraion of each variable under sudy. This implies ha he es allows esing for he exisence of a coinegraing relaionship beween variables in levels irrespecive of wheher he underlying regressors are I () or I (1). This is differen from he general bivariae and mulivariae coinegraion frameworks, which require ha ime series in he sysem should be non-saionary in heir levels and ha all ime series in he coinegraing equaion should have he same order of inegraion. Following Pesaran, e al. (21), we consruced he vecor auoregression (VAR) of order p (VAR(p)) for deb-growh model: p Z = μ + βiz i + ε, (2) i= 1 where Z is he vecor of boh X and Y, where Y is he dependen variable (RGDP) and X is he vecor marix represens a se of explanaory variables (DEBT, [ ] RTB, WSTE and DUM). μ = μ Y, μ X, is a ime or rend variable, and β i is a marix of VAR parameers for lag i. According o Pesaran, e al. (21), he dependen variable mus be I (1) variable, bu he regressors, or explanaory variables can be eiher I () or I (1). We can furher develop a Vecor Error Correcion Model (VECM) as follows: p 1 i= 1 i i p 1 ΔZ = μ + α + λz 1 + γ Y + γ X + ε, (3) i= i i where Δ = 1 L and α = α Y, α ]. We pariion he long-run muliplier marix as [ X

7 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS 113 follows: λyy λyx λ =. λxy λxx The diagonal elemens of he marix are unresriced, so he seleced series can be eiher I() or I(1). If λ =, hen Y is I(1). In conras, if <, hen Y is I(). YY The VECM procedures described above are imporan in esing of a mos, one coinegraing vecor beween dependen variable ( Y ) and a se of regressors ( X ). Furher, following he assumpions made (unresriced inerceps and no rends) and resricions imposed ( λ XY =, μ and α = ) by Pesaran, e al. (21) in Case III, herefore, we re-formulae Equaion (3) o derive he following Unresriced Error Correcion Model (UECM) o examine he long run relaionship beween real GDP and deb. λ YY ΔRGDP = β + β RGDP + + p i= 1 p i= β ΔRGDP β ΔWSTE 1 i i + β DEBT + p i= 2 + u, 7 1 β ΔDEBT + β RTB i 3 + p i= β WSTE 4 β ΔRTB i 1 + β DUM 5 (4) where u is he whie noise error erm; Δ is he firs difference operaor; and p is lag srucure, which deermined by Akaike s informaion crierion. There are wo seps in esing he coinegraion relaionship beween economic growh and is explanaory variables. Firs, we esimae Equaion (4) by ordinary leas square (OLS) echnique. Second, we examine he long run relaionship by imposing he resricion ha all esimaed coefficiens of lagged one level variables equal o zero. Tha is, he null hypohesis is β1 = β2 = β3 = β4 = agains is alernaive hypohesis. In order o es he above hypoheses, following Pesaran, e al. (21), we will apply eiher sandard Wald es or F-saisic, which has a non-sandard disribuion ha depends on few facors such as sample size, he inclusion of inercep and rend variable in he esimaion, and number of regressors. If he F-saisic obained from he resricion is less han lower bound criical value, we do no rejec he null hypohesis of no long run relaionship. In conras, if he compued F-saisic is greaer han upper bound criical value, hen we rejec he null hypohesis and conclude ha here appears seady sae long run equilibrium beween he variables under sudy. However, if he F-saisic falls wihin lower and upper bound criical values, hen he resuls are inconclusive and he saionariy of he series mus be examined and invesigaed. Narayan (25) argues ha he use of Pesaran, e al. s (21) criical values for small sample sudy may produce misleading resuls because he criical values calculaed are

8 114 T.K. JAYARAMAN AND CHEE-KEONG CHOONG generally lower han hose generaed by Narayan who used similar GAUSS code used by Pesaran, e al. (21). Narayan (25) has generaed a new se of criical values ranging from 3 o 8 observaions. Since he sample size in our sudy is small (ha is, 34 observaions) and as he criical values provided by Pesaran, e al. (21) are calculaed on he basis of large sample sizes of 5 and 1 observaions and 2 and 4 replicaions respecively, we propose o use he criical values provided by Narayan (25). Once he variables are found o be coinegraed, hen he nex sep is o use he error-correcion model o esimae he shor-run dynamic causaliy relaionship. Equaion (3) can now be consruced ino a vecor error-correcion model (VECM) in order o capure boh shor- and long -run impac of he vecor. Defining Z as he vecor of he poenially endogenous variables, we can model Z as an unresriced 1 vecor auoregression (VAR) model wih lag-lengh up o 3: Z where = σ Z 2 A Z + A Z + A Z U where U ~ IN(, ), (5) is (4 x 1) vecor consiss of RGDP, DEBT, WSTE and RTB. Each of he is (4 x 4) marix of parameers. The 4-VAR model as saed in Equaion (5) will be used if here is no long run relaionship o be found in he bound esing approach. However, if here is a coinegraion relaionship, hen he following vecor error correcion will be applied o examine he long- and shor -run causaliy beween variables. Ai ΔZ = Γ ΔZ + Γ ΔZ + ΠZ + U , (6) where ΔZ = [RGDP, DEBT, WSTE and RTB], Γ1 = ( I A1 ), Γ2 = ( I A1 A2 ) and Π = ( I A1 A2 A3). Γi measures he shor-run effec of he changes in Z. The (4 x 4) marix of Π ( = αβ ') conains boh speed of adjusmen o disequilibrium (α ) and he long-run informaion ( β ) such ha he erm β ' Z 3 embedded in Equaion (6) represens he (n-1) coinegraing vecor in he model. Accordingly, we can re-sae he Equaion (6) as follows: 1 The maximum lag lengh up o 3 is suggesed as he frequency of he daa is annual and here are 34 observaions in he sudy.

9 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS 115 ΔRGDP ΔRGDP ΔDEBT ΔDEBT = Γ1 ΔWSTE ΔWSTE ΔRTB ΔRTB ΔRGDP ΔDEBT + Γ2 ΔWSTE ΔRTB 2 RGDP β11 β21 β31 β41 DEBT β12 β22 β32 β42 WSTE β 13 β23 β33 β43 RTB α 11 α12 α13 α 21α 22 α 23 + α 31α32 α33 α 41 α 42 α 43 (7) There are wo seps involved in he esimaion of error-correcion model (ECM). Firs, we idenify he unique long-run relaionship based on heory ha represens he economic relaionship underlying he long run model among real GDP, real deb, he raio of wages and salaries o oal expendiure and real Treasury bill rae. Secondly, we esimae he shor-run model wihin he VECM o find ou he shor run causal relaionship. The shor run model is of ineres since we can sudy he behaviour of each variable in he esimaed sysem in response o he residual from he coinegraing equaion (error-correcion erm - ECT). The ECT measures he speed of adjusmen of each variable in response o a deviaion from he seady sae equilibrium relaionship. Since he objecive of he sudy is o examine he causaliy relaionship beween economic growh and deb, he wo equaions are derived from Equaion (7) as follows: Δ Δ k k k RGDP = 1 ECT 1 + π jδrgdp j + τ jδdebt j + λ jδy j + u1 j=1 j=1 j=1 β, (8) k k k DEBT = 2 ECT 1 + φ jδrgdp j + δ jδdebt j + η jδy j + u2 j=1 j=1 j=1 β, (9) where ECT 1 is he one-period lagged error correcion erm, Y is he vecor comprising WSTE and RTB, and u 1 and u 2 are whie noise error erms. In hese wo equaions, he series real GDP and deb are coinegraed when a leas one of he coefficiens β1 or β2 is no zero. In ha case, wo series will display long-run relaionship. If β1 and β 2 =, we conclude ha deb Granger causes RGDP in he long run. On he oher hand, if β2 and β 1 =, RGDP will Granger cause deb. If boh β1 and β2 are nonzero, he conclusion hen is ha here exiss a feedback relaionship beween economic growh and deb. The shor-run relaionships beween growh and deb are signified by he coefficiens τ j ' s and φ' s. If τ j ' s are no all zero, movemens in deb will cause growh in he shor-run. If φ i ' s are no all zero, movemens in growh will cause deb in he shor-run. The shor-run as well as long-run dynamic causaliy relaionships beween growh and

10 116 T.K. JAYARAMAN AND CHEE-KEONG CHOONG deb can be assessed by forming hypoheses and esing hem on he esimaed coefficiens in he Equaions (8) and (9). In general, six possible esable hypoheses concerning he shor-run and long-run influence of deb on growh ( DEBT RGDP ) and growh on deb ( RGDP DEBT ) can be formulaed. These can be described and summarized in Table 3. Table 3. Six Possible Tesable Hypoheses beween Deb on Growh Granger Causaliy Tes Tesable Hypoheses Descripion H ST DEBT RGDP (No ST linkage) τ j = DEBT does no Granger Cause (j = 1,...,k) DCGDP in he shor-erm H LT DEBT RGDP (No LT linkage) β = 1 DEBT does no Granger Cause DCGDP in he long-erm H DEBT NO RGDP (No ST or LT linkages) β 1= and τ j = DEBT does no Granger Cause (j = 1,...,k) DCGDP in he shor-erm and long erm H ST RGDP DEBT (No ST linkage) φi = DCGDP does no Granger Cause (i = 1,...,k) DEBT in he shor-erm H LT RGDP DEBT (No LT linkage) β = 2 DCGDP does no Granger Cause DEBT in he long-erm H RGDP NO DEBT (No ST or LT linkages) β 2 = and φi = DCGDP does no Granger Cause (i = 1,...,k) DEBT in he shor-erm and long -erm These individual hypoheses can be esed using sandard F-ess on he esimaed coefficiens of he error-correcion model. The six hypoheses are used o examine he lead-lag and feedback relaionships beween deb and economic growh as well as oher variables. 4. EMPIRICAL RESULTS We sar by esing he long run relaionship among real GDP (RGDP), real oal deb (RTD), real reasury bill rae (RTB) and wages and salaries o oal expendiure raio (WSTE) by esimaing Equaion (4) by ordinary leas squares (OLS) mehod. Four equaions, each wih four differen variables, namely RGDP, RTD, RTB and WSTE as dependen variable along wih oher variables as explanaory variables are hus esimaed and he calculaed F-saisics are repored in Table 4. The level of significance chosen for our sudy is 1%. Among four equaions, only in regard o he equaion wih RGDP as dependen variable, we find he null hypohesis of no long-run relaionship sands

11 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS 117 rejeced as he compued F-saisic (8.25) is greaer han he upper bound value eiher from Pesaran, e al. (21) [5.6] or Narayan (25) [6.37]. In he oher hree cases, we find he null hypohesis canno be rejeced. Table 4. Bound Tes for Coinegraion Analysis Based on Equaion (4) Dependen Variable Compued F-saisic RGDP 8.25 *** RTD 3.62 RTB 1.67 WSTE 2.35 Pesaran e al. (21) a Narayan (25) b Criical Lower Upper Lower Upper Value bound value bound value bound value bound value 1 percen percen percen Sources: a Criical values are obained from Pesaran e al. (21), Table CI (iii) Case III: unresriced inercep and no rend, p. 3. b Criical values are obained from Narayan (25), Table case III: unresriced inercep and no rend, p *** indicaes significance a 1% level. Having found he exisence of a long-run relaionship beween real GDP, real deb, real reasury bill rae and he raio of recurren expendiures o oal expendiure when real GDP is he dependen variable, we esimae he long run elasiciies. The esimaed Equaion (1), wih figures in parenheses represening calculaed values is as follows: RGDP = RTD.11RTB WSTE +.2DUM. (4.57) (5.26) ( 4.73) (5.5) (.84) (1) Thus, we find ha coefficiens of he explanaory variables namely RTD, RTB and WSTE no only have he expeced signs in accordance wih heoreical expecaions bu are also found saisically significan. While esimaed coefficiens of oal deb and WSTE are posiive, he coefficien of real Treasury bill rae is negaive. Deb posiively influences economic growh. So oo WSTE: higher recurren expendiure, which conribues o consumpion, posiively affecs GDP. The lower he ineres rae, which is represened by RTB, higher is he economic growh. Dummy variable (DUM) for poliical insabiliy appears o have had no effec on economic growh, as he coefficien was no saisically significan. Equaion (1) is adequae as indicaed in he diagnosic checking oupu, i.e., he

12 118 T.K. JAYARAMAN AND CHEE-KEONG CHOONG model has he desired properies of OLS mehod such as serially uncorrelaed, consan variance or homoscedasiciy of residuals and has a correc funcional form. Moreover, he CUSUM and CUSUM of Squares plo 2 show ha he parameers of he model are sable over ime. 3 Since we found he exisence of a coinegraing relaionship among RGDP, RTD, RTB and WSTE, here should be Granger causaliy a leas in one direcion. For esing he direcion of emporal causaliy beween he variables, we resor o esing Granger causaliy wihin a vecor error correcion model (VECM). The es resuls are repored in Table 5. Table 5. Granger Causaliy Tess Based on Vecor Error Correcion Model Dependen F-saisic ECT -1 Variable ΔRGDP ΔRTD ΔRTB ΔWSTE ( saisic) ΔRGDP *** *** ΔRTD ΔRTB ΔWSTE Noe: *** significance a 1% level. The resuls show ha ECT in all he four equaions has he correc negaive sign. However, only he ECT in he RGDP equaion is found saisically significan a 1 per cen level, which confirms he resuls we obained from he bounds es of coinegraion. This implies ha in he long run he causaliy runs from deb, Treasury bill rae and he raio of recurren expendiures o oal expendiures, o GDP and ha change in GDP are a funcion of disequilibrium in he coinegraing relaionship. The ECT coefficien of.92 indicaes ha adjusmen owards he long run equilibrium is abou 92% per annum, suggesing any deviaion from he long run equilibrium is correced subsanially in he following year. Turning o shor-run causal effecs, we find ha shor-run causaliy runs only from deb o growh. Thus, here is only unidirecional relaionship, running from deb o 2 The use of boh CUSUM and CUSUM squares ess are aims o show he sabiliy of he esimaions. Neverheless, i is well known ha in he esimaion of lagged dependen variables, boh ess are less accurae as a guide. In addiion, given he large number of parameers being esimaed and he small sample size, hese ess are more likely o have low power. Therefore, hese ess are used as complemenary diagnosic checking o our esimaed models. The validiy of boh models (Tables 4 and 5), noneheless, is reliable as demonsraed by oher diagnosic ess such as serial correlaion, heeroscedasiciy and mis-specificaion ess. 3 The diagnosic checking resuls are available upon reques.

13 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS 119 growh, boh in he long run and shor run. 5. SUMMARY AND CONCLUSIONS The paper employed he bounds esing approach o examine he relaionship beween Fiji s economic growh, public deb, real ineres rae and raio of governmen recurren expendiure o is oal expendiure during a 34-year period (197-23). Empirical analysis revealed ha economic growh had a long-run relaionship wih public deb, real ineres rae and raio of governmen recurren expendiures o oal expendiure. Second, vecor error correcion modeling procedure esablished ha in he long run erm causaliy was only unidirecional and ha i was from deb, ineres rae and raio of governmen recurren expendiures o oal expendiures, o GDP. In he shor run, he causaliy ran from deb o GDP. The oher wo variables, ineres rae and raio of governmen recurren expendiure o is oal expendiure in he shor run had no effec on real GDP. Thus, we have he resul ha deb influenced economic growh, boh in he long and shor runs. REFERENCES Alesina, A., Ardagna, S., Peroi, R., and Schianarelli, F. (22), Fiscal Policy, Profis and Invesmen, American Economic Review, 92, Asian Developmen Bank (ADB) (24), Key Indicaors of Asian and Pacific Developing Counries 24, ADB Manila: ADB. (26), Asian Developmen Oulook 26, Manila: ADB. Barram, E., and B, Ward. (1993), The Exernaliy Effec of Governmen Expendiure on Invesmen in OECD Counries, Applied Economics, 25, Chand, S. (23), Economic Trends in he Pacific Islands, Pacific Economic Bullein,18(1), Chandra, S.A., and T.K. Jayaraman (25), Managemen of Public Deb and Economic Growh in Fiji, Paper presened a he Conference on Developmen Sudies, Suva, Fiji Islands, December. Commonwealh Secrearia (25), Small Saes: Economic Review and Basic Saisics, 1, London: Commonwealh Secrearia. Fairbairn, T.I.J., and D, Worrell. (1996), Souh Pacific and Caribbean Island Economies: A Comparaive Sudy, Brisbane: Foundaion for Developmen Cooperaion. Giannaros, D., Kolluri, B., and M, Panik. (1999), An Empirical Analysis of he Effecs of Governmen Spending on Capial Invesmen: Evidence from O.E.C.D.

14 12 T.K. JAYARAMAN AND CHEE-KEONG CHOONG Counries, Inernaional Economic Journal, 12, Gill, I., and B, Pino. (25), Public Deb in Developing Counries: Has he Marke Based Model Worked? Working Paper 3674, Washingon D.C.: World Bank. Inernaional Moneary Fund (IMF) (23), World Economic Oulook 23, Washingon D.C., Inernaional Moneary Fund. (24). Fiscal Adjusmen in IMF Suppored Programs, Washingon D.C., Inernaional Moneary Fund. Jayaraman, T.K. (26), Macroeconomic Reform and Resilience Building in Small Saes, in L. Briguglio, G. Cordina, and E.Kisanga ed., Building he Economic Resilience of Small Saes, Mala: Islands and Small Saes Insiue and London: Commonwealh Secrearia. Narayan, P.K. (25), The Saving and Invesmen Nexus for China: Evidence from Coinegraion ess, Applied Economics, 37, Narube, S. (24), Fiji Economic Updae: Can We Susain Our Economic Performance? Reserve Bank of Fiji Symposium on Towards Prosperiy for All, Suva: Reserve Bank of Fiji, Pesaran, M.H., Shin, Y., and R, Smih. (21), Bounds esing approaches o he analysis of level relaionships, Journal of Applied Economerics, 16(3), Prasad, B. (26), Warning Agains Runaway Deb, Wansolwara, 11(2), 1. Reddy, M. (25), Public Finance in Fiji, Lauoka, Fiji Islands: Fiji Insiue of Applied Sudies, Reserve Bank of Fiji (RBF) (25), Quarerly Review December 25, Suva, Fiji Islands, Reserve Bank of Fiji. (26), Moneary Policy Saemen: June 26, Suva, Fiji Islands, Reserve Bank of Fiji. Sahay, R. (24), Sabilizaion, Deb and Fiscal Policy in he Caribbean, Paper presened a he Seminar on Developmenal Challenges Facing he Caribbean, cosponsored by he Cenral Bank of Trinidad and IMF, Por of Spain, June. Shah, S. (23), An Unpayable Burden of Deb, Sunday Pos, 3, 4. Selowsky, M. (24), IMF Book Forum: Fiscal Adjusmen in IMF-Suppored Program, quoed in IMF Survey, June 28, 188. Unied Naions Economic and Social Commission for Asia and he Pacific (UNESCAP) (24), Economic and Social Survey 24, Bangkok, UNESCAP. (26), Economic and Social Survey 26, Bangkok, UNESCAP

15 PUBLIC DEBT AND ECONOMIC GROWTH IN THE SOUTH PACIFIC ISLANDS 121 Mailing Address: T.K. Jayaraman;School of Economics Faculy of Business and Economics, The Universiy of he Souh Pacific Suva, Fiji Islands. Chee-Keong Choong; Deparmen of Economics Faculy of Accounancy and Managemen Universii Tunku Abdul Rahman, Malaysia Selangor D.E., Malaysia. Manuscrip received June 26; final revision received November 26.