University of Wollongong Economics Working Paper Series 2008

Transcription

1 Universiy of Wollongong Economics Working Paper Series 2008 hp:// An Empirical Analysis of Susainabiliy of Trade Defici: Evidence from Sri Lanka Nelson Perera Graduae School of Business Reeu Verma School of Economics Universiy of Wollongong WP May 2008

2 An Empirical Analysis of Susainabiliy of Trade Defici: Evidence from Sri Lanka Nelson Perera Graduae School of Business Reeu Verma School of Economics Phone: Fax: Universiy of Wollongong Norhfields Avenue Wollongong NSW 2500 Ausralia

3 An Empirical Analysis of Susainabiliy of Trade Defici: Evidence from Sri Lanka Nelson Perera and Reeu Verma ABSTRACT In his paper, he long-run relaionship beween Sri Lankan expors and impors during he period 950 o 2006 is examined using uni roo ess and coinegraion echniques ha allow for an endogenously deermined srucural break. The resuls failed o suppor he exisence of a long-run equilibrium beween expors and impors in Sri Lanka. This finding quesions he effeciveness of Sri Lanka s curren long-erm macroeconomic policies and suggess ha Sri Lanka is in violaion of is inernaional budge consrain. Keywords: Trade Defici, Expors, Impors, Uni roo, Srucural Breaks, Coinegraion, Sri Lanka JEL Classificaions: C2, C22, F4, F32 3

4 . Inroducion The exisence of he long-run equilibrium or coinegraion relaionship beween expors and impors has been empirically esed by many researchers. These include Arize (2002); Bahmani-Oskooee (994); Bahmani-Oskooee and Rhee (997); Herzer and Felcias (2006); Irandous and Ericsson (2004); and Narayan and Narayan (2005). Coinegraion beween expors and impors implies ha rade deficis are only a shorerm phenomena and hus susainable in he long-run. This means ha counries are no in violaion of heir inernaional budge consrain, since heir macroeconomic policies have been effecive in bringing expors and impors ino a long-run equilibrium. The objecive of his sudy is o invesigae wheher a long-run equilibrium exiss beween expors and impors in Sri Lanka. Sri Lanka is an ineresing case sudy because i was he firs Souh Asian counry o embark on rade liberalisaion as early as 978 and, despie changes in governmen and inernal conflic; liberal economic policies have been coninually followed for he las 30 years. The approach adoped in he paper is differen o previous sudies in wo ways. Firsly, we use annual daa raher han quarerly daa. Secondly, we use recenly developed uni roo and coinegraion ess ha allow for an endogenously deermined srucural break in individual ime series and he long-run relaionship beween expors and impors. In he nex secion of his paper we provide a brief noe on he behaviour of expors and impors in Sri Lanka during he period 950 o This will be followed by he heoreical raionale for invesigaing wheher expors and impors are coinegraed in secion 3. Following his, we explain he economeric mehodology and discussion on he empirical resuls, in secion 4, and our conclusions are presened in he final secion. 2. Trends in expors and Impors in Sri Lanka In early period of he decade following 950, nearly 90 per cen of expors from Sri Lanka consised of ea, rubber and coconu while rice accouned for 25 per cen of impors. The impor of rice and oher food impors were necessary o mainain he food subsidy scheme promoed by he sae. As a resul of populaion growh, impors of hose iems coninued o grow faser rae han expors. The siuaion was furher worsened by he coninuing decline of erms of rade. The ne resul was impors ino Sri Lanka were 4

5 always higher han expors as can be seen from Figure. As a resul, successive governmens, oped for selecive impor conrols which were inended, in he firs insance, as a proecive measure bu which ended up encouraging and proecing paricular indusries and firms (Cuhberson and Ahukorala: 99: page 328). Even for he period 96 o 977, he governmen aemped o conrol impors simply by adding furher resricions ypically in he form of permis and licenses. Permission o impor was graned o specific agencies and ceilings were placed on goods ha could be impored. In addiion o ariff resricions, resricions on foreign exchange ransacions also played a roll in limiing impors. Given he low commodiy prices in he world marke and lack of indusrial producion for expor marke, expors from Sri Lanka did no grow fas enough o mach he growh of impors. Figure : Expors and Impors in Sri Lanka expors impors $US in millions Source: Cenral Bank of Sri Lanka (2006) 980 Years For a discussion on hose poins, see Kelegama (2006). 5

6 The period since 977 can be classified as a more liberal period when all resricions on rade and ransacions were abolished 2. During his period, impors have coninued o grow a a faser rae han expors, and rade balance has always been in defici. However, wih he composiion of expors shifing o more indusrial expors, his has reduced he dependence on he radiional expors of ea, rubber and coconu. 3. Theoreical Foundaions Hused (992) provides a esable model for a small open economy which has he following key feaures: absence of governmen; abiliy o produce and expor a composie good; wih consumers having access o inernaional funds implying a long-run relaionship beween expors and impors. Hused sars wih he individual curren period budge consrain as follows: C () 0 = Y0 + B0 I 0 ( + r) B where C 0, Y0, B0, I 0 and r are he curren consumpion, oupu, inernaional borrowing, invesmen, and a one period ineres rae, respecively; and ( + r ) B is he iniial deb size. Hused makes several assumpions in deriving he following esable model. One of hem is ha ineres rae is saionary wih mean r: X = α + δ + ε (2) M where X is he expors of goods and service; and M is he impors of goods and services. For he necessary and sufficien condiion for he iner-emporal budge consrain of he counry o hold, we require ha δ = and e is saionary. I follows, ha expression (2) provides an idealized framework for deermining he susainabiliy of a curren accoun defici (surplus). In he even his proviso is no me, i would indicae ha he economy is no funcioning properly and fails o saisfy is budge consrain, and herefore, is expeced o defaul on is deb (Hakkio and Rush, 99). Hused (992) ess for coinegraion beween expors and impors in he US plus ineres paymens abroad and finds no evidence of coinegraion for he For he deails on policy measurers implemened during he period, see Jayewardene e al. (987) and Ahukorala and Jayasuriya (994). 6

7 period. An analysis of subsamples and wih srucural break in 983 suppors coinegraion. Founas and Wu (999) es he susainabiliy of he curren accoun for Unied Saes. Using quarerly daa for he period and , hey found ha he series are no coinegraed and curren accouns are no susainable. Apergis e al. (2000) es for he susainabiliy of he Greek curren accoun wih annual daa for he period They found ha he Greek curren accoun defici was susainable. Bahmani-Oskooee (994) used coinegraion echniques o invesigae he long-run relaionship beween expors and impors for Ausralia. He found evidence of coinegraion and concluded ha Ausralian expors and impors will converge in he long-run. Quarerly daa was used by Bahmani-Oskooee and Rhee (997) o model expors and impors for Korea. They found evidence of coinegraion and he coefficien on expors was posiive. Arize (2002) used quarerly daa for he period for 50 OECD and developing counries. Arize found ha for 35 of he 50 counries here was evidence of coinegraion beween expors and impors. He also found ha 3 of he 35 counries had a posiive expor coefficien. Tang (2003) used he bounds esing approach o invesigae he presence of he relaionship beween expors and impors for five ASEAN economies 3. He found ha expors and impors are coinegraed for Malaysia and Singapore only. Narayan and Narayan (2004) found ha a long-run relaionship exiss beween expors and impors for Fiji and PNG using he bounds esing approach. They also found ha coefficien on expors is uniy only in he case of Fiji. On he basis of Johansen s echnique (995), Irandous and Ericsson (2004) found ha here is a coinegraion relaionship beween impors and expors of Germany, Sweden and he USA; bu here is no coinegraion relaionship for he UK. Shiraz and Manap (2005) do no rejec he null of no coinegraion beween expors, impors and real oupu in Sri Lanka. Narayan and Narayan (2005) invesigae wheher here is a long-run relaionship (coinegraion) beween expors and impors for 22 leas developed counries (LDCs) using he bounds esing approach o coinegraion. Their resuls indicae ha expors and impors are coinegraed only for six ou of he 22 counries. Herzer and Felcias (2006) conclude ha here is long-run equilibrium beween expors and impors in Chile. 3 Indonesia, Malaysia, Philippines, Singapore and Thailand. 7

8 4. Economeric Mehodology and Empirical Tesing Uni Roo Tess Mos exising sudies examine he saionary properies of variables by using he Augmened Dickey Fuller (ADF) (979, 98) or Philip-Perron (988) uni roo ess. A problem wih hese ess is ha neiher es allows for he possibiliy of a srucural break. Perron (989) argued ha in he presence of a srucural break, he sandard ADF ess are biased owards he non-rejecion of he null hypohesis. I should be noed ha in Perron (989) procedure daing of he poenial break is assumed o be known a priori in accordance wih he underlying asympoic disribuion heory. Tes saisics are hen consruced by adding dummy variables represening differen inerceps and slopes, hereby exending he sandard Dickey-Fuller procedure. However, Chrisiano (992) criicized Perron s known assumpion of he break dae as daa mining 4. Chrisiano argued ha he daa based procedures are ypically used o deermine he mos likely locaion of he break and his approach invalidaes he disribuion heory underlying convenional esing. Since hen, several sudies have been developed using differen mehodologies for endogenising he break dae. Some of hese include Banerjee e al. (992), Zivo and Andrews (992), Perron (997) and Lumsdaine and Papell (998). These sudies have shown ha bias in he usual uni roo ess can be reduced by endogenously deermining he ime of he srucural break. Perron (997) proposed a class of es saisics which allows for wo differen forms of srucural break: he Innovaional Oulier (IO) model where changes are assumed o ake place gradually; and he Addiive Oulier (AO) model, which allows for he srucural change o ake place insananeously. Perron s (997) procedure is he more recen and updaed echnique and is he mos widely used procedure o deermine a single break. 4 The Perron (989) mehod is an exogenous break es as opposed o he mehods developed by Chrisiano (992) and Zivo and Andrews (992) which are endogenous break ess. The criicism of daa mining is someimes also leveled a he endogenous break ess. 8

9 The IO model allows for a gradual change in he inercep (IO) and gradual changes in boh he inercep and he slope of he rend funcion (IO2) such ha: IO: x = µ + θ DU + β + δ D( T ) + α x + c x + e (3) b - i - i i= IO2 : x = µ + θ DU + β + γ DT + δ D( T ) + αx + c x + e (4) b i i i= where T b denoes he ime of break ( < T b < T) which is unknown, DU = if > T b and zero oherwise, DT = T if > T b and zero elsewhere, D(T b ) = if = T b + and zero oherwise, x is any general ARMA process and e is he residual erm assumed whie noise. The null hypohesis of a uni roo is rejeced if he absolue value of he - saisic for esing α =l is greaer han he corresponding criical value. Perron (997) suggess ha he ime of srucural break (T b ) can be deermined by wo mehods. The K K firs approach is ha equaions (3) or (4) are sequenially esimaed assuming differen T b wih T b chosen o minimize he -raio for α =. In he second approach, T b is chosen from among all oher possible break poin values o minimize he -raio on he esimaed slope coefficien (γ ). The runcaion lag parameer (k) is deermined using he daa-dependen mehod proposed by Perron (997). The choice of k in his mehod depends upon wheher he - raio on he coefficien associaed wih he las lag in he esimaed auoregression is significan. The opimum k (or k*) is seleced such ha he coefficien on he las lag in an auoregression of order k* is significan and ha he las coefficien in an auoregression of order greaer han k* is insignifican, up o a maximum order k (Perron, 997). The hird model is he Addiive Oulier (AO) model. In conras o he gradual change in he IO model, he AO model assumes srucural changes ake place insananeously; ha is i allows for a sudden and rapid change in he rend funcion. When considering he AO model for esing a uni roo, a wo-sep procedure is used. Firs he series is derended using he following regression: 9

10 * y µ β DT y = + + γ + % (5) where * y% is he derended series and DT ( T ) = b if Tb > and zero oherwise. This assumes ha a srucural break only impacs on he slope coefficien. Thus, he es is hen performed using he -saisic for α = in he regression: y% = y% + c y + e κ α i i i= The uni roo null hypohesis is rejeced in favour of he alernaive if he -saisic for α is significan and greaer han he criical values abulaed by Perron (997). We used annual daa for he period from 950 o The expors and impors daa was obained from Trend in Key Economic Variables in Special Saisical Appendix from Cenral Bank of Sri Lanka (2006). In our analysis, wo series of expors and impors were used: in one series he value of expors and impors are measured in curren Sri Lankan rupee values and in he second series, he value of expors and impors are measured in US dollars. Table provides uni roo ess using he Perron (997) mehod. The resuls obained indicae ha boh he expors and impors, measured eiher in Sri Lankan rupees or US dollars, are non-saionary under srucural change a a five per cen significance level for all cases. 0

11 Table : Perron s (997) Uni Roo Tess: Addiive Oulier Model (AO); and he Innovaional Model (IO2) Par : Series in US Dollars Series Model Break Poin T b Lag k Tes Saisic αˆ Criical Values a 5% Resul X AO Uni Roo M AO Uni Roo X IO Uni Roo M IO Uni Roo Par 2: Series in Sri Lankan Rupees Series Model Break Poin T b Lag k Tes Saisic αˆ Criical Values a 5% Resul X AO Uni Roo M AO Uni Roo X IO Uni Roo M IO Uni Roo Coinegraion Once he order of inegraion of each variable is deermined, we esed for coinegraion. As menioned earlier, ignoring he issue of poenial srucural breaks can render invalid saisical resuls no only for uni roos ess bu also in erms of coinegraion ess. Kuniomo (996) argued ha in he presence of srucural change, radiional coinegraion ess which do no allow for a srucural break may produce spurious coinegraion resuls. The effec of poenial srucural breaks on he resul of

12 he ADF es for coinegraion was also recognized by Gregory and Hansen (996). They show ha ADF es ends o under-rejec he null hypohesis of no coinegraion in he presence of a srucural break. Considering he imporance of he effecs of a poenial srucural break, we applied he Gregory and Hansen (996) coinegraion procedure ha allows for an endogenously deermined srucural break. The problem of esimaing coinegraion relaionships in he presence of poenial srucural break is addressed by Gregory and Hansen by inroducing a residual based echnique. The echnique is o es he null hypohesis of no coinegraion agains he alernaive of coinegraion wih a srucural break. Here, he break poin is unknown and is deermined by finding he minimum values for he ADF -saisic. The Gregory and Hansen procedure akes ino accoun he exisence of a poenial unknown and an endogenously deermined single break, allowing for srucural shifs in eiher he inercep alone, in boh rend and level shif or a full break. Tha is, Gregory and Hansen presen hree models for esing coinegraion where hey allow for he exisence of srucural break in he coingeraing vecor. The firs model is known as a level shif model (Model C). This model conains an inercep and a level shif dummy as follows: y u u ϕ τ y + e =,, n. (6) T = α 2 The second model (C/T) conains an inercep and a rend wih a level shif dummy: y u u ϕ τ y + e =,, n. (7) T = β + α 2 The hird model is he full break model called a regime shif (C/S), allowing for change in boh inercep and slope as follows: T T = u + u2ϕ τ + α y2 + α 2 y2ϕ τ e =,, n. (8) y + Model C/S includes wo dummy variables, one for he inercep and one for he slope. In he conex of our analysis, y and y 2 are he expors and impors; u and α are he inercep and slope coefficiens before he shif; u 2 and α 2 denoe he changes o he inercep and slope coefficiens a he ime of he shif. The dummy variable is denoed by ϕ τ and is defined by: ϕ = 0, if [ ητ ] τ and τ ϕ =, if > [ ητ ] 2

13 where he unknown parameer τ ε (0,) denoes he relaive iming of he change poin. The hree models (expressions 6-8) are esimaed o deermine he ime of he srucural break and also o es for he exisence of coinegraion relaionship beween expors and impors. The resuls and he criical values are repored in Table 2. Firs par of Table 2 repors resuls for expors and impors series measured in US dollar erms. These resuls sugges ha models C and C/S do no rejec he null of no coinegraion a he five per cen significance level. However, model C/T rejecs he null of no coinegraion a he five per cen significance level. The break daes of 979, 985 and 99 deeced by he Gregory and Hansen procedure correspond wih he liberalizaion of he Sri Lanka economy in 978, he civil war inensifying in 983 and he escalaion of violence beween he Sri Lankan army and he separaiss in 990. The second par of Table 2 repors resuls for expors and impos measure in Sri Lankan Rupee erms. These resuls sugges he models C and C/T do no rejec he null of no coinegraion a he five per cen significance level. However, model C/S rejecs he null of no coinegraion a he five per cen significance level. The break daes of 995 and 998 deeced by Gregory and Hansen procedure are differen from he previous series. Table 2: Gregory and Hansen (996) Coinegraion Tess wih Srucural Break Par : Series in US Dollars Model Break Poin ADF Criical Value Resul* 5% C Do No Rejec he Null Hypohesis C/T Rejec he Null Hypohesis C/S Do No Rejec he Null Hypohesis 3

14 Par 2: Series in Sri Lankan Rupees Model Break Poin ADF Criical Value Resul* 5% C Do No Rejec he Null Hypohesis C/T Do No Rejec he Null Hypohesis C/S Rejec he Null Hypohesis * The null hypohesis being no coinegraion beween expors and impors. Criical values are provided by Gregory and Hansen (996). The resuls from he above analysis are inconclusive: in four cases, here is no coinegraion beween expors and impors bu in wo cases suggess ha here is a coinegraion beween expors and impors. Wha does coinegraion or lack of coinegraion beween impors and expors ell us abou he sae of he economy? According o Hused (992), coinegraion is o be expeced under he hypohesis ha he economy is working properly, and ha breaking inernaional budge consrains can lead o a lack of coinegraion. This implies ha susained exernal imbalances are he oucome of disored markes, or bad policy. Irandous and Ericsson (2004) argue ha he lack of coinegraion suggess fundamenal policy problems, unless here are permanen produciviy shocks ha lead o a non-saionary impor expor relaionship. In a well funcioning economy, wihou permanen one-sided produciviy shocks, coinegraion is o be expeced. However, given he curren global environmen, he exernal balance is deermined no only by rade balance, bu also by he balance in he services and paymens secor. This is more relevan o counries like Sri Lanka, where services expors and privae remiances are very significan par of he curren accoun. Generally speaking, he conclusion is ha a lack of coinegraion suggess fundamenal policy problems which could be challenged in he curren globalize economic environmen. Therefore, fuure research ino his issue should be direced o include hose elemens of he curren accoun. 4

15 5. Conclusion The purpose of his paper was o invesigae he susainabiliy of curren accoun of Sri Lanka by employing he Hused (992) esing procedure. The procedure used here is o esimae coinegraion beween expors and impors. In our analysis, wo series of expors and impors were used: in one series he value of expors and impors are measured in US dollars and in he second series, he value of expors and impors are measured in curren Sri Lankan rupee values. The period covered was 950 o The resuls from he above analysis are inconclusive. In four cases, here is no coinegraion beween expors and impors; however, in wo cases i is suggesed ha here is coinegraion beween expors and impors. In summary, a coinegraion es based on he Gregory and Hansen approach does no suppor he exisence of long-run equilibrium beween expors and impors. The empirical findings sugges ha he curren accoun of Sri Lanka is no susainable (and his violaes is ineremporal budge consrain) in he long-run. However, given he changes in he world rade sysem furher research is needed o include services and remiances of he curren accoun. 5

16 References Apergis, N., K.P. Karakilidis, and N.M. Tabakis (2000), Curren accoun defici susainabiliy: The case of Greece, Applied Economic Leers, 7: Arize, A. (2002), Impors and Expors in 50 Counries: Tess for Coinegraion and Srucural Break, Inernaional Review of Economics and Finance, Vol. :0-5. Ahukoralan P. and S. Jayasuriya (994), Macroeconomic Policies, Prices and Growh in Sri Lanka. Washingon DC, : World Bank. Bahmani-Oskooee, M.(994) Are Impors and Expors of Ausralia Coinegraed?, Journal of Empirical Inegraion, Vol. 9 No. 4: Bahmani-Oskooee, M. and Rhee, H.-J. (997) Are expors and impors of Korea Coinegraed? Inernaional Economic Journal, Vol. : Banerjee, A., R. L.Lumsdaine and J.H. Sock (992), Recursive and Sequenial Tess of he Uni Roo and Trend-Break Hypohesis: Theory and Inernaional Evidence, Journal of Business and Economic Saisics, Vol. 0: Annual Repor (2006), Trends in Key Economic Variables in Special Saisical Appendix. Cenral Bank of Sri Lanka Chrisiano, L.J. (992), Searching for a Break in GNP, Journal of Business and Economic Saisics Vol. 0: Cuhberson, A.G and Ahukorala, P (99), Sri Lanka in D. Papageougiou, M Michaely and A. M. Choksi (eds) Liberalizing Foreign Trade, Vol. 5 Blackwell Publishers. Founas, S., Wu, J-L. (999), Are The U.S. Curren Accoun Deficis Really Susainable? Inernaional Economic Journal, 3(3), Auumn: Gregory, A. W. and B. E. Hansen, (996) Tess for coinegraion in models wih regime and rend shifs, Oxford bullein of Economics and Saisics, Vol. 58: Hakkio, C. S. and Rush, M. (99) Is he budge defici oo large?, Economic Inquiry, Vol. 29: Herzer, D. and Felcias, N. L. D. (2006) Is here a Long-run Relaionship beween Expors and Impors in Chile? Applied Economic Leers, Vol. 3:

17 Hused S. (992), The Emerging U.S. Curren Accoun Defici in he 980s: A Coinegraion Analysis, The Review of Economics and Saisics, February: Irandous, M and Ericsson, J. (2004), Are Impors and Expors Coinegraion? An Inernaional Comparison, Meroeconomica, Vol. 55(), pp Jayawardena, L., Maasland, A and Radhakrishnan, P. N., (987, Sabilizaion and Adjusmen Policies and Programs: Sri Lanka SWIDER, Counry Sudy No. 5, Helsinki. Kelegama, S., (2006), Developmen under Sress: Sri Lankan Economy in Transiion Sage Publicaions India: New Delhi. Kuniomo, N. (996), Tess of Uni Roos and Coinegraion Hypoheses in Economeric Models, Japanese Economic Review, 47(), pp Lumsdaine, R. L., and D. H. Papell (997), Muliple Trend Breaks and he Uni Roo Hypohesis, Review of Economics and Saisics, Vol. 79, No. 2, pp Narayan, P.K. and S. Narayan (2005), Are expors and impors coinegraed? Evidence from 22 leas developed counries, Applied Economic Leers, 2, pp Narayan, P.K. and S. Narayan (2004), Is here a long-run relaionship beween expors and impors? Evidence from wo pacific island counries, Economic Papers, Vol.23, No.2: Perron, P. (989), The Grea Crash, The Oil Price Shock, and The Uni Roo Hypohesis, Economerica, Vol. 57, pp Perron, P. (997), Furher Evidence on Breaking Trend Funcions in Macroeconomic Variables, Journal of Economerics, Vol. 80: Shiraz, S.S., T.A. Manap (2005), Expor-Led Growh Hypohesis: Furher Economeric Evidence from Souh Asia, The Developing Economies, December: Tang, T.C. (2003), Are Expors and Impors of he Five ASEAN Economies Coinegraed? An Empirical Sudy, Inernaional Journal of Managemen, Vol. 20, No.: Zivo, E. and K. Andrews (992), Furher Evidence on The Grea Crash, The Oil Price Shock, and The Uni Roo Hypohesis, Journal of Business and Economic Saisics, Vol. 0, No. 0: