Stock markets and economic growth in oil exporting countries: evidence from Kuwait.

Transcription

1 MPRA Munich Personal RePEc Archive Sock markes and economic growh in oil exporing counries: evidence from Kuwai. El Mosafa Benour 15. May 2014 Online a hp://mpra.ub.uni-muenchen.de/55997/ MPRA Paper No , posed 26. May :17 UTC

2 Sock markes and economic growh in oil exporing counries: evidence from Kuwai. El Mosafa Benour, he Arab Planning Insiue, Kuwai ABSTRACT: This paper empirically invesigaes he role played by he Kuwai sock marke on he real secor. I uses useful seps and echniques o se up a regression based on he Mankiw-Romer-Weil model o answer wheher here is an evenually posiive effec of he sock marke developmens on he real economy. The resuls show a posiive impac of he marke capialisaion on he Gross Domesic Produc. The elasiciy of he marke capializaion o GDP is around This impac is also confirmed by an auoregressive vecor model via esimaion and impulse response funcions on boh oal and non oil GDP. Key Words: Non oil GDP, Marke Capializaion, Cobb-Douglas, Human Capial. 1

3 INTRODUCTION The heoreical lieraure on he role of he financial secor on economic growh and is implicaions in he differen sages of developmen has been widely discussed in recen years 1. However, i is ineresing o noe ha he endency of his lieraure has no remained consisen over ime. Thus, empirical invesigaion is becoming he main ool o assess for each counry or a group of counries, he role of financial secor on he real economic aciviy. Recenly, sock markes, as par of he whole financial sysem, are growing fas o compee he role of he banking inermediaion o spread savings among invesors. This benefi role could be offse by he insabiliy characerizing he sock marke and leading o negaive repercussions on he real economy (Winkler, 1998). Sock marke variables are very volaile inducing an uncerainy around he invesors expecaions and increasing he risk. A connexion beween he real secor and sock exchange marke ransmis he marke developmens o he real economic aciviy. Daa on Kuwai Sock Exchange shows ha marke capializaion is 16 imes volaile han GDP 2. This paper presen a simple mehodology based on Cobb-Douglas producion funcion o invesigae he role played by he Kuwai sock marke on he real secor. I uses useful seps and echniques o se up a regression based on he Mankiw-Romer-Weil model o answer wheher here is an evenually posiive effec of he sock marke developmens on he real economy. The resuls show ha he sock marke has a posiive impac on he economic aciviy. In wha follows, we firs describe he model. Second, we presen daa mehodology consrucion. In he hird secion we presen and discuss resuls. The fourh secion is for a VAR model esimaion o suppor he Mankiw-Romer-Weil model resuls and hen we conclude. 1 See, for example, Levine (2005) and Beck (2009) for a survey of he heoreical and empirical lieraure. 2 The Variance of marke capializaion over he variance of GDP is for he sample

4 1. DESCRIBING THE MODEL The mehodology of he paper follows he same one as in Arusha (2010) applied o Kuwai sock marke series daa. I examines he influence of he sock marke on he real economy by exending he Mankiw-Romer-Weil model (1992) o incorporae a sock marke variable on i. The model sars from a Cobb-Douglas producion funcion form, linking he oupu level Y o he physical capial K and labour L as inpus: Y A. K. L (1) Where A is he oal facor produciviy erm and; and represen he shares of physical capial and labour respecively. Augmening he model by inroducing wo oher inpus, human capial H and he sock exchange marke variable S, he model becomes: Y A. K H S. L (2) Where and are he shares of he human capial and marke capializaion. Scaling he model according o labour (dividing by L) and rearranging, hen inroducing he naural logarihm, we have he following equaion: Log ( Y / L ) c. Log ( K / L ). Log ( H / L ). Log ( S / L ). Log ( L ) (3) Where c Log(A), is a consan erm and 1capures he deviaion from consan reurns o scale. Puing small capials o design scaled variables we rewrie he preceden equaion as: Log y ) c. Log ( k ). Log ( h ). Log ( s ). Log ( L ) (4) ( This form allows inerpreing direcly he parameers as elasiciies and measures he impac of he inpus on he oupu. Our objecive is o assess, for he Kuwai, he relaionship beween he sock marke variable and he level of he economic aciviy; he Gross Domesic Produc. 3

5 2. DATA CONSTRUCTION To esimae he previous equaion, we should have on our possession daa for GDP, capial sock, human capial, Marke capializaion as a proxy for he sock variable and labour. We use World Developmen Indicaors (WDI) o have GDP, Labour and marke capializaion. However, he capial sock and human capial are no observed. We consruced he capial sock according o he equaion: K ( 1 ) K I (5) 1 I simply means ha, he curren sock is generaed as he pas one K 1 depreciaed by some amoun K 1 and augmened by he curren flow of he invesmen I. δ is a depreciaion rae assumed equal 6% and he firs value K 0 2.5* GDP. Marke capializaion is observed since 1984, he official dae where he Kuwai Sock Exchange becomes reorganized in an official body. The flow of invesmen is Gross Capial Formaion (GCF) from WDI daabase. As we do no have sufficien long ime series for real GCF, we use he GDP deflaor o have real GCF from nominal values. For missing values for , we use simple linear exrapolaion o fill he gap. Human capial is approached as he average years of schooling derived from he gross enrolmen school as percen of populaion aged 15 and over for primary, secondary and eriary schooling. Barro and Lee have consruced a large daabase for 146 counries, including Kuwai, from 1950 ill 2010, by sex and by 5 years inervals ( We use linear exrapolaion o fill in he gap in order o display a complee ime series of human capial. 3. ESTIMATION The model is implemened in EViews. The marke capializaion used o es he relaionship beween GDP and he KSE marke is available only from 1988 ill 2010, wih wo missing values for 1990 and We handled he probelem by using backward exrapolaion for he period and forward exrapolaion for he wo preceden years. 4

6 The esimaion is in log level and no in differences alhough he variables are no saionnary. This approach is deffended for he fondamenal following reason. The relaionship beween finance and growh is raher a long run relaionship han shor run. In economerics and saiscis, he long run is seen o be he common economic rend of he daa in level, while he shor run is he volailiy of he growh rae series. However, running regressions in general is governed by a lo of probabiliy laws and hypohesis o check. Saisical ess are used o assess he qualiy of he esimaed model. Table 1 shows long run (levels) and shor run (growh raes) correlaion marices. There is an imporan posiive correlaions beween oal GDP, non oil GDP and marke capializaion in levels ( abou 80%). In growh raes, he correlaions are sill posiive bu weaker. However, he correlaion is only an indicaion of saisical linkages and does no mean causaliy. The causaliy issue is discussed in he fourh secion. Table 1: Varibales marice Correlaions in levels and growh raes. Sample Correlaions in levels Toal GDP Marke Capializaion Non Oil GDP Toal GDP Marke Capializaion Non Oil GDP Sample Correlaions in growh raes Toal GDP Marke Capializaion Non Oil GDP Toal GDP Marke Capializaion Non Oil GDP The economeric model o esimae is he equaion (4) in he preceden secion augmened by an error erm: Log ( y ) c. Log ( k ). Log ( h ). Log ( s ). Log ( L ) (5) 5

7 2 Where N(0, 0 ) is a processus supposed o have a gaussien processus properies where he mean equal 0, he variance independen of ime (normal and idenically disribued), or also wha is called whie noise processus. I run wo regressions: he unresriced specificaion which is equaion (5) and in case of he nulliy of he parameer, he equaion exibi a consan reurns o scale. Under his assumpion; 1 0, we have 1. We re-esimae he new specificaion and call i resriced specificaion. The wo specificaions are esimaed for he wo samples: and and he resuls are shown in he able 2. The residuals of he esimaions sugges o correc he effec of he observaions around 1990 and For his purpose, we consruc a dummy variable ha we used o offse he effec of he war in his period over he variables for he sample The esimaion resuls confirm ha marke capializaion have a posive impac on he real GDP level for he wo specificaions and over he whole and reduced samples. The elasiciy is highly acceped and is, 25.8% and 27.6% for he whole sample ( ) and respecively for he wo specificaions. Considering he reduced sample where all he daa are observed, he coefficien is arround 17%. The impac of he human capial is rejeced; he p-values associaed wih are superiors o 5%, which means ha here is no effec of human capial on he economic aciviy. This resuls should be considered wih precauion regarding he mehod of daa exrapolaion of human capial. The coeffiecien μ is also rejeced in he unresriced specificaion. A wald hypohesis es of he nulliy of boh and confirm he nulliy of hese coefficiens. F-Saisic probabiliies are respecively 0.14 and 0.22 for he samples and The second column of he able presens a resriced specificaion ( 0) which mean ha he Cobb-Douglass producion funcion exibi a consan reurns o scale confirmed by he nulliy of his coefficien as shwon in he firs column. The hird column omi he human capial rejeced in he firs and second esimaion and he overall properies are unchanged over boh samples. For all esimaions, Durbin-Wason is around 1.60 over and 1.86 over

8 indicaing he absence of auocorelaions especially on he shorer sample where daa are no compleed by exrapolaion. Table 2: Summary of esimaion resuls Unresriced Specificaion Resriced Specificaion Resriced Specificaion wihou human capial Sample P- P- P- P- P- P- Coef. Coef. Coef. Coef. Coef. Coef. Value Value Value Value Value Value α β NA NA NA NA γ µ ** ** ** ** ** ** ** ** Adj. R SE DW NA: no applicable since he human capial is omied from he equaion. ** In he resriced specificaion, is null and VAR MODEL ASSESSMENT A Vecor Auo-Regressive (VAR) model, iniiaed by Sims (1980) is a sysem in which each componen of he vecor explains he dependan variable by is proper pas values and he pas values of he oher variables presen in he sysem. To esimae a VAR model, we firs deermine he opimal lag o inroduce in he model. All he selecion crieria concorde abou he firs lag 3. Then, a granger causaliy 4 es is conduced and show ha causaliy run ou from marke capializaion o boh oal GDP and non oil GDP (able 3). The opposie direcion is rejeced a he 5% probabiliy hreshold. 3 Eviews displays five crieria for his purpose: LR es, Final predicor error, Akaike informaion, Schwarz informaion and Hannan-Quinn informaion crierion. 4 Granger says ha X causes Y if he pas values of he X help predic Y. 7

9 Table 3: Pairwise Granger Causaliy Tess for Marke capializaion and GDP. Lags = 1 Sample: Null Hypohesis: Obs F-Saisic Prob. LGDP_R does no Granger Cause LS_R LS_R does no Granger Cause LGDP_R LNOILGDP_R does no Granger Cause LS_R LS_R does no Granger Cause LNOILGDP_R LGDP_R sands for logarihm real GDP, LS_R for real marke capializaion and LNOILGDP_R for logarihm real non oil GDP. Johansen ess rejec any relaionship of coinegraion beween GDP (for boh oal and non oil) and marke capializaion. Table 4 presens he oupu es for boh GDP and non oil GDP and marke capializaion. I shows ha race es indicaes no coinegraion a he 5% level according o MacKinnon-Haug-Michelis (1999) p-values. We conclude ha he final sysem beween GDP and marke capializaion is a sandard VAR wih one lag. Table 4: Johansen Coinegraion ess Sample (adjused): Trend assumpion: Linear deerminisic rend Marke Capializaion, Toal GDP Unresriced Coinegraion Rank Tes (Trace) Hypohesized No. of CE(s) Eigenvalue Trace Saisic 5% Criical Value Prob.** None A mos Marke Capializaion, non oil GDP Unresriced Coinegraion Rank Tes (Trace) Hypohesized No. of CE(s) Eigenvalue Trace Saisic 5% Criical Value Prob.** None A mos Trace es indicaes no coinegraion a he 0.05 level * denoes rejecion of he hypohesis a he 0.05 level **MacKinnon-Haug-Michelis (1999) p-values 8

10 The Cholesky impulse response funcion repored in he figures 1 and 2 race he responses generaed by he VAR model for each variable o he oher variable impulsed by one uni of is sandard deviaion. Figures show boh posiive effecs. There is no immediae response of he marke capializaion o he real GDP. The effec is null in he firs year and become more imporan in he second year hen reaches is maximum in he hird and fourh years. While, he response of marke capializaion o boh oal GDP and non oil GDP is non null from he firs year: here is an immediae effec and his effec is maximal in he second year. Response of GDP o Marke Capializaion Response of Marke Capializaion o GDP Figure 1: Response Funcions of he oal GDP and Marke Capializaion Response Of Non Oil GDP o Marke Capializaion Response Of Marke Capializaion o Non Oil GDP Figure 2: Response Funcions of he non Oil GDP and Marke Capializaion 5. CONCLUSION We esed in his paper he impac of marke capializaion on he economic aciviy level in he long run. We used for his purpose a model derived from heoreical economic approach and afer several ess and esimaions, we conclude ha here is effecively a posiive effec of he sock marke exchange over he GDP in Kuwai. The 9

11 elasiciy is on average around 20% meaning ha an increase in he sock marke capializaion by 1% implies he GDP increase by 0.2%. A sandard VAR model also confirm his effec. However, he resuls may suffer from he shorcoming of he daa and shor samples. While he paper presen an individual case of an oil exporing counry, he work could be exended o panel daa over he group of he GCC counries or oil exporing counries. REFERENCES Arusha, C. (2010): "Do sock markes lead o economic growh?" Journal of Policy Modelling, 32(4), pp Demirguc-Kun, A. and Levine, R. (1996): Sock Markes, Corporae finance and Economic Growh: Overview, World Bank Economic Review, 10(2), pp Beck, T. (2009): The Economerics of Finance and Growh. In Palgrave Handbook of Economerics, 2, ed. Terence Mills and Kerry Paerson, Hounds mill: Palgrave Macmillan, pp Demeriades, P. and Hussein, K. A. (1996): "Does Financial Developmen Cause Economic Growh? Time-series Evidence from 16 Counries", Journal of Developmen Economics, 51, pp Levine, R. (2005): Finance and Growh: Theory and Evidence. In Handbook of Economic Growh, ed. Philippe Aghion and Seven N. Durlauf, Amserdam: Elsevier, pp

12 Mankiw, N. G., Romer, D. and Weil, D. (1992): A conribuion o he empirics of economic growh, Quarerly Journal of Economics, 2, Winkler, A. (1998): The Dual Role of Financial Markes in Economic Developmen: Engine of Growh and Source of Insabiliy. A Survey of Economic Theory wih Reflecions on heeas Asian Financial Crisis, Inernaionale Projek Consul working paper, 18, Frankfur. 11