Which comes first savings or growth? Time series evidence from ECOWAS countries

Size: px

Start display at page:

Download "Which comes first savings or growth? Time series evidence from ECOWAS countries"

Della Blake
6 years ago
Views:

1 e Theoreical and Applied Economics Volume XXV (208), No. 2(65), Summer, pp Which comes firs savings or growh? Time series evidence from ECOWAS counries Yaya KEHO Ecole Naionale Supérieure de Saisique e d Economie Appliquée (ENSEA), Abidjan, Côe d Ivoire yayakeho@yahoo.fr Absrac. This sudy examines he direcion of causaliy beween domesic savings and economic growh in welve Wes African counries during he period I uses he auoregressive disribued lag (ARDL) approach o coinegraion and he Granger causaliy es. The resuls show ha economic growh causes domesic savings in Guinea-Bissau and Nigeria, while he reverse causaliy running from savings o economic growh holds in Benin, Gambia, Mali, Niger, and Senegal. A wo-way causaliy prevails in Ghana. However, no evidence of causaliy in eiher direcion was recorded for Burkina Faso, Coe d Ivoire, Sierra Leone and Togo. These findings show ha he direcion of causaliy beween savings and growh is counry-specific among he Wes African Counries. Keywords: savings, economic growh, causaliy, ECOWAS. JEL Classificaion: C32, E2, O, O55.

2 248 Yaya Keho. Inroducion The causal relaionship beween savings and economic growh has received a grea deal of research aenion in developmen economics. However he debae wheher economic growh causes savings or savings causes economic growh sill remains an unresolved issue in boh heory and empirics. Wih regard o he heoreical lieraure, here are wo main compeing views of he savings-growh nexus. The famous neoclassical model developed by Solow (956) highlighed he imporance of savings in he process of economic developmen. This model showed ha higher saving raes lead o higher level of per capia oupu and economic growh rae by increasing domesic capial sock in he economy. However, his is only valid in he shor erm. Increases in he saving raes do no affec long-erm growh raes. The endogenous growh models developed by Romer (986) and Lucas (988) also esablished a close connecion beween higher savings and economic growh by emphasizing he convenional belief ha higher savings rae leads o higher economic growh rae. Conrary o his view, he sandard Keynesian model advocaed ha saving depends upon he level of curren income, and he savings rae increases as per capia income increases. All oher hings being equal, economic growh precedes higher savings and no he opposie. There is a grea deal of empirical lieraure ha has esed he savings-growh nexus for many specific counries or counry groups. So far, here is no general consensus on he causal relaionship beween he wo variables. The empirical findings vary cross counries, mehodologies and variables specificaions. Some empirical sudies found ha economic growh leads o higher savings (Sinha and Sinha, 998; Salz, 999; Aanasio e al., 2000; Agrawal, 200; Anoruo and Ahmad, 200; Mohan, 2006; Odhiambo, 2008; Abu, 200), however ohers repored ha savings lead o economic growh hrough capial formaion (Sinha, 999; Alguacil e al., 2004; Masih, and Peers, 200; Bankole and Faai, 203; Para e al., 207; Pickson e al., 207; Siaw e al., 207). Evidence of bidirecional causal link has been repored in he lieraure (Romm, 2005; Lean and Song, 2009; Odhiambo, 2009; Singh, 2009; Tang and Chua, 202; Adeleke, 204; Najarzadeh e al., 204; Tang and Tan, 204). Oher sudies failed o idenify any significan relaionship beween savings and growh (Mavroas and Kelly, 200; Baharumshah e al., 2003; Hundie, 204; Sohan, 204). The analysis of he causal relaion beween savings and economic growh has imporan policy implicaions for developing counries (Alguacil e al., 2004). If saving is a precondiion for economic growh, hen governmens need o design and implemen policies ha promoe savings for higher economic growh. Conversely, if economic growh causes savings, hen effors would be direced owards acceleraing economic growh. Despie hese implicaions here is very scany evidence from Sub-Saharan African counries. The aim of his sudy is o empirically invesigae he direcion of causaliy beween savings and economic growh in he member counries of he Economic Communiy of Wes African Saes (ECOWAS) using ime series daa from The res of he paper is organized as follows. Secion 2 describes he daa and he economeric mehodology employed for esing causaliy. Secion 3 presens he empirical resuls of he sudy. Secion 4 concludes he sudy and provides some policy recommendaions.

3 Which comes firs savings or growh? Time series evidence from ECOWAS counries Daa, model and mehodology The sudy uses annual ime series daa for a sample of welve member counries of he Economic Communiy of Wes African Saes (ECOWAS), namely: Benin, Burkina Faso, Coe d Ivoire, Gambia, Ghana, Mali, Niger, Nigeria, Senegal, Sierra Leone and Togo. The counries were seleced based on daa availabiliy. The variables considered are real gross domesic produc (hereafer GDP) as a proxy for economic growh and real gross domesic savings (hereafer GDS). The daa se comes from he World Developmen Indicaors of he World Bank and spans he period of All he daa used are in consan local currency unis and convered ino naural logarihms in he economeric analysis. We use gross domesic savings as our focus is on he oal amoun of resources available in he economy for invesmen. The economeric model used in his sudy heoreically follows he Solow and he Keynesian hypoheses. According o he Solow model, higher savings precede economic growh. As such, he empirical model specifies economic growh as a funcion of savings: lngdp 0 lngds () The coefficien θ indicaes he long-run impac of domesic savings on economic growh. A large and significan coefficien suggess ha he financial marke works efficienly so ha savings lead o invesmen and economic growh. On he oher hand, he Keynesian model saes ha savings are a funcion of income. Thus, he empirical model o be esimaed is specified as follows: lngds 0 lngdp 2 (2) where he coefficien ϕ indicaes he long-run impac of economic growh on domesic savings. We es for he presence of a long-run relaionship beween domesic savings and economic growh using he Auoregressive Disribued Lag (ARDL) bounds es developed by Pesaran e al. (200). This approach has many advanages over oher radiional alernaive mehods. I has been shown o have superior properies in small sample size. The bounds esing approach o coinegraion is based on he following equaion: lngdp n i0 2i m 0 lngdp 2 lngds i i i lngdp lngds e (3) The presence of a long-run relaionship is esed by resricing coefficiens of lagged level variables equal o zero. Tha is, he null hypohesis of no long-run relaionship is H 0 : 2 0. This hypohesis is esed by he mean of an F-es. The asympoic criical values are provided by Pesaran e al. (200). The bounds esing procedure is sensiive o i

4 250 Yaya Keho he selecion of he lag srucure (m, n). In his sudy, he lag srucure was seleced using he Akaike Informaion Crierion (AIC) wih maximum lag lengh on each variable se o five. Coinegraion es indicaes only wheher or no a long-run relaionship exiss beween savings and growh. I does no indicae he direcion of he causal relaionship among hem. Engle and Granger (987) argued ha as long as variables are coinegraed, causaliy mus exis a leas in one direcion. Following heir mehodology he direcion of causaliy beween savings and economic growh can be deeced by esimaing he following Error Correcion Models: ln ln p p GDP i lngdp i i lngdsi ecm e i i (4) p p GDS 2 2ilnGDP i 2ilnGDSi 2ecm e2 i i (5) where ecm - is he lagged residuals of he long-run relaionship beween savings and GDP. Here again, he lag lengh p is deermined using he Akaike Informaion Crierion (AIC). The significance of he differenced explanaory variables indicaes he shor-run causaliy, whereas he significance of ecm - confirms he long-run causal relaionship. For example, savings do no cause GDP growh in he shor-run if φ =φ 2 = = φ p =0. Similarly, GDP growh does no cause savings in he shor-run if none of γ 2i is saisically differen from zero. 3. Empirical resuls Before we proceed wih he esimaion of he empirical models, we need o invesigae he ime-series properies of he variables using uni roo ess. We apply he PP es of Phillips and Perron (988) and he KPSS es of Kwiakowski e al. (992). These ess have been performed under he models wih consan and rend for he level series and wih consan for series in firs difference. This sep is necessary because he bounds es requires he dependen variable o be inegraed of order one (I()) and he explanaory variables o be I(0) or I(). The resuls presened in Table reveal ha boh lngdp and lngds series have uni roos in heir level bu become saionary afer aking heir firs difference. This suggess ha he variables under sudy are inegraed of order one. Table. Uni roo es resuls Counry PP es KPSS es lngdp lngds ΔlnGDP ΔlnGDS lngdp lngds ΔlnGDP ΔlnGDS Benin * * * 0.39 ** Burkina * * 0.9 * * Coe d Ivoire * * Gambia * * * * Ghana * * 0.84 * 0.3 ** Guinea-Bissau * *

5 Which comes firs savings or growh? Time series evidence from ECOWAS counries 25 Counry PP es KPSS es lngdp lngds ΔlnGDP ΔlnGDS lngdp lngds ΔlnGDP ΔlnGDS Mali * * * * 0.64 * 0.2 * Niger * * 0.2 * 0.20 * * Nigeria * * * 0.96 * 0.75 * Senegal * * 0.87 * Sierra Leone ** * * 0.97 * ** Togo ** * * ** Noe: * (**) denoes he rejecion of he null hypohesis a he 5% (0%) level. The resuls of he ARDL bounds es are displayed in Table 2. From his able we see ha eigh counries show significan exising long-run relaionship beween lngdp and lngds. Table 2. ARDL bounds es for coinegraion Counry FGDP FGDS Coinegraion? Benin * 3.27 Yes Burkina No Coe d Ivoire No Gambia * 5.0 ** Yes Ghana ** Yes Guinea-Bissau ** Yes Mali * 5.95 ** Yes Niger * Yes Nigeria * Yes Senegal 3.05 * * Yes Sierra Leone No Togo No * Noe: * ( ** ) denoes he rejecion of he null hypohesis a he 5% (0%) level. The resuls of he Granger-causaliy ess are presened in Table 3. The esimaes of he error correcion erms show ha savings Granger-cause GDP in he long-run in Benin, Gambia, Ghana, Mali and Senegal. This suggess he need for hese counries o encourage savings o promoe susainable growh. On he oher hand, he evidence from Guinea-Bissau, Niger and Nigeria is consisen wih he Keynesian model ha posulaed causaliy from economic growh o savings. In he shor run, he radiional view ha higher savings leads o higher economic growh holds in Benin, Mali, and Niger. The reverse causaliy running from economic growh rae o growh rae of savings holds only in Guinea-Bissau. A bi-direcional causaliy beween growh rae of savings and economic growh rae was found in Ghana, suggesing ha savings are boh a cause and a consequence of economic growh. Finally, here is no causal relaionship beween he wo variables in Burkina Faso, Coe d Ivoire, Sierra Leone and Togo. Our findings for Nigeria is consisen wih Abu (200) bu conradic wih Adeleke (204) and Bankole and Faai (203). Our findings are also in agreemen wih Pickson e al. (207) for Ghana. In he case of Coe d Ivoire, our resuls conradic wih Anorou and Ahmad (200) who found bi-direcional causaliy beween savings and economic growh. The absence of causaliy for Burkina Faso, Coe d Ivoire, and Togo is also in line wih Konya (2005) bu his auhor repored no causaliy beween savings rae and per capia income in Benin, Ghana, Nigeria, and Senegal.

6 252 Yaya Keho Table 3. Resuls of Granger causaliy ess Shor-run causaliy Long-run causaliy ECT-=0 Counry GDS does no cause GDP GDP does no cause GDS GDP GDS Benin * (0.000) (0.70) -0.6 * (-3.286) (.425) Burkina Faso 0.44 (0.59) 0.00 (0.967) - - Coe d Ivoire.348(0.245) (0.942) - - Gambia (0.58) (0.856) * (-4.5) (-0.048) Ghana * (0.000) * (0.023) * (-3.77) (.809) Guinea-Bissau (0.568) 3.20 ** (0.073) (0.433) * (-2.694) Mali 8.77 * (0.033) (0.882) * (-2.622) 0.76 (.498) Niger * (0.040) (0.578) (0.05) * (-2.343) Nigeria (0.344) 0.0 (0.94) (-0.708) * (-3.497) Senegal 0.7 (0.678) (0.465) * (-3.775) 0.84 (0.876) Sierra Leone (0.638) (0.928) - - Togo (0.469) (0.40) - - Noe: GDP denoes Gross Domesic Produc, GDS is Gross Domesic Savings. Saisics for Shor-run causaliy are Chi-square saisics wih p-values in parenheses. Saisics for long-run causaliy are coefficiens on ECT- wih -saisics in brackes. The aserisks * and ** denoe saisical significance a he 5% and 0% levels, respecively. 4. Conclusion and policy recommendaions The objecive of his sudy was o invesigae he causal relaionship beween domesic savings and economic growh for welve ECOWAS counries. Using ime series annual daa from 98 o 204, he ARDL bounds es for coinegraion was firs applied o discover he exisence of a long-run relaionship beween he wo variables. The resuls reveal ha here is a posiive long run relaionship beween he wo variables in eigh counries. Granger causaliy ess were hen performed o depic he direcion of causaliy beween savings and economic growh. The resuls provide suppor o he Solow s model and he Keynesian predicion. I has been found ha economic growh causes savings in Guinea-Bissau and Nigeria while he reverse causaliy running from savings o economic growh holds in Benin, Gambia, Mali, Niger, and Senegal. For Burkina Faso, Coe d Ivoire, Sierra Leone and Togo here is no evidence of causaliy beween savings and economic growh in eiher direcion. Finally, a wo-way causaliy prevailed in Ghana. I is clear from hese resuls ha policy makers in Benin, Gambia, Ghana, Mali, Niger, and Senegal are required o implemen policies o simulae savings in order o accelerae economic growh. For he remaining counries, effors should be done o creae appropriae condiions for beer allocaion of domesic resources o growh-led secors. References Abu, N., 200. Saving-economic growh nexus in Nigeria, : Granger causaliy and co-inegraion analyses. Review of Economic and Business Sudies, 3(), pp Adeleke, A.M., 204. Saving-growh nexus in an oil-rich exporing counry: A Case of Nigeria. Managemen. Science and Engineering, 8(3), pp Agrawal, P., 200. The relaion beween savings and growh: Coinegraion and causaliy evidence from Asia. Applied Economics, 33(4), pp

7 Which comes firs savings or growh? Time series evidence from ECOWAS counries 253 Alguacil, M., Cuadros, A. and Ors, V., Does saving really maer for growh? Mexico ( ). Journal of Inernaional Developmen, 6(2), pp Anoruo, E. and Ahmad, Y., 200. Causal relaionship beween domesic savings and economic growh: Evidence from seven African counries. African Developmen Review, 3(2), pp Aanasio, O.P., Picci, L. and Scorcu, A.E., Saving, growh, and invesmen: a macroeconomic analysis using a panel of counries. The Review of Economics and Saisics, 82(2), pp Baharumshah, A.Z., Thanoon, M.A. and Rashid, S., Savings dynamic in Asian counries. Journal of Asian Economics, 3(6), pp Bankole, A.S. and Faai, B.O., 203. Relaionship beween savings and economic growh in Nigeria. The Social Sciences, 8(3), pp Engle, R.F. and Granger, C.W.J., 987. Coinegraion and error correcion: represenaion, esimaion, and esing. Economerica, 55(2), pp Hundie, S.K., 204. Savings, invesmen and economic growh in Ehiopia: Evidence from ARDL approach o co-inegraion and TYDL Granger-causaliy ess. Journal of Economics and Inernaional Finance, 6(0), pp Konya, L., Saving and growh: Granger causaliy analysis wih boosrapping on panels of counries. Journal of Economic Research, 0(2), pp Kwiakowski, D., Phillips, P.C.B., Schmid, P. and Shin, Y., 992. Tesing he null hypohesis of saionariy agains he alernaive of a uni roo. Journal of Economerics 54 (-3), pp Lean, H.H. and Song, Y., The domesic savings and economic growh relaionship in China. Journal of Chinese Economic and Foreign Trade Sudies, 2(), pp Lucas, R. E., 988. On he mechanics of economic developmen. Journal of Moneary Economics, 2(), pp Masih, L. and Peers, M., 200. A revisiaion of he savings growh nexus in Mexico. Economics Leers, 07, pp Mavroas, G. and Kelly, R., 200. Old wine in new boles: esing causaliy beween savings and growh. The Mancheser School, 69(), pp Mohan, R., Causal relaionship beween savings and economic growh in counries wih differen income levels. Economics Bullein, 5(3), pp. -2. Najarzadeh, R., Reed, M. and Tasan, M., 204. Relaionship beween savings and economic growh: he case for Iran. Journal of Inernaional Business and Economics, 2(4), pp Odhiambo, N. M., Financial deph, savings and economic growh in Kenya: a dynamic causal linkage. Economic Modelling, 25(4), pp Odhiambo, N.M., Savings and economic growh in Souh Africa: a mulivariae causaliy es. Journal of Policy Modelling, 3(5), pp Para, S. K., Murhy, D.S., Kuruva, M. B. and Mohany, A., 207. Revisiing he causal nexus beween savings and economic growh in India: an empirical analysis. Economia, 8(3), pp Pesaran, M.H., Shin, Y. and Smih, R., 200. Bounds esing approach o he analysis of level relaionships. Journal of Applied Economerics, 6(3), pp Phillips, P.C.B. and Perron, P., 988. Tesing for a uni roo in a ime series regression. Biomerika, 75(2), pp Pickson, R.B., Enning, K.D. and Siaw, A., 207. Savings-growh nexus in Ghana: coinegraion and causal relaionship analyses. Theoreical Economics Leers, 7, pp

8 254 Yaya Keho Romer, P., 986. Increasing reurns and long-run growh. Journal of Poliical Economy, 94(5), pp Romm, A.T., The relaionship beween savings and growh in Souh Africa: A ime series analysis. Souh African Journal of Economics, 73(2), pp Salz, I.S., 999. An examinaion of he causal relaionship beween savings and growh in he hird world. Journal of Economics and Finance, 23(), pp Siaw, A., Enning, K. and Pickson, R., 207. Revisiing domesic savings and economic growh analysis in Ghana. Theoreical Economics Leers, 7, pp Sinha, D., 999. Saving and economic growh in Sri Lanka. Indian Journal of Applied Economics, 8(3), pp Singh, T., Does domesic saving cause economic growh? A ime-series evidence from India, Journal of Policy Modeling, 32(2), pp Sinha, D. and Sinha, T., 998. Car before he horse? The savings-growh nexus in Mexico, Economics Leers, 6(), pp Solow, R.M., 956. A Conribuion o he heory of economic growh. Quarerly Journal of Economics, 70(), pp Sohan, S., 204. Causal Relaionship beween domesic saving and economic growh: evidence from Cambodia. Inernaional Journal of Economics and Finance, 6(9), pp Tang, C.F. and Chua, Y., 202. The savings growh nexus for he Malaysian economy: a view hrough rolling samples. Applied Economics, 44(32), pp Tang, C.F. and Tan, B.W., 204. A revalidaion of he savings growh nexus in Pakisan. Economic Modelling, 36, pp

The Relationship between Money Demand and Interest Rates: An Empirical Investigation in Sri Lanka

The Relationship between Money Demand and Interest Rates: An Empirical Investigation in Sri Lanka The Relaionship beween Money Demand and Ineres Raes: An Empirical Invesigaion in Sri Lanka R. C. P. Padmasiri 1 and O. G. Dayarana Banda 2 1 Economic Research Uni, Deparmen of Expor Agriculure 2 Deparmen