Causal Relationship between Agricultural Exports and Exchange Rate: Evidence for India

Transcription

1 Applied Economics and Finance Vol. 4, No. 6; November 2017 ISSN E-ISSN Published by Redfame Publishing URL: hp://aef.redfame.com Causal Relaionship beween Agriculural Expors and Exchange Rae: Evidence for India Dicle Ozdemir Correspondence: Dicle Ozdemir, Assisan Professor of Economics, Mugla Siki Kocman Universiy, Turkey. Tel Received: Sepember 6, 2017 Acceped: Ocober 2, 2017 Available online: Ocober 12, 2017 doi: /aef.v4i URL: hps://doi.org/ /aef.v4i Absrac In his paper we empirically invesigae he causal link beween agriculural expors and real exchange rae in India employing linear and nonlinear causaliy analysis. We carry ou our invesigaion using annual index of he quaniy of agriculural expors in India and real US Dollar o Rupee exchange rae daa which cover he period beween 1961 and We find ha here are no significan changes in he linear and nonlinear causal relaions beween agriculural expors and exchange raes over he sample period under invesigaion. However, our invesigaion does no provide any evidence of bidirecional or unidirecional causaliy beween he agriculural expors o real exchange rae in India. I can be concluded ha one of he key reasons for high levels of he quaniy of agriculural expors in India is long-erm economic growh, raher han he real value of Indian Rupee. Keywords: agriculure expor, exchange rae, causaliy, nonlinear, India 1. Inroducion One of he main poins Schuh (1974) sressed in his paper was ha he exchange rae was admied as an omied variable in economic analysis of he US farm secor for years and argued ha he linkages beween exchange rae and agriculural rade should have been exensively invesigaed by researchers. Schuh argued ha he exchange rae affecs he valuaion of resources wihin a counry, devaluaions of he dollar during he fixed exchange rae scheme led o significan srucural changes in agriculure in he Unied Saes. He concluded ha he over-valuaion and under-valuaion of he dollar had been imporan in explaining he pah of domesic agriculural prices. Susani (2001) esed he effec of Indonesia s currency value on he expors of agriculural producs and found ha boh he oal expors and five expor producs were negaively influenced by is exchange rae movemens. Similar conclusions are repored by Mahew e al. (2006). In his paper, we examine he linear and nonlinear causal dynamics beween he real U.S. exchange rae of Indian Rupee and he amoun of agriculural expors in India. While i is generally acceped ha exchange rae is an imporan economic variable influencing expor, i has no been demonsraed he srengh of his causal reasoning empirically. Since he exchange rae of a currency has an effec on prices of agriculural producs, i has also a grea poenial o play a big role in he compeiiveness of he agriculure indusry of a counry; ha is, here is a negaive relaionship beween he value of a currency and he amoun of expors. Unless domesic producers accep a lower price for heir producs, a srong currency of an exporer counry will make expors more expensive for imporers and make less compeiive. Similarly, a weaker currency would increase expors and make domesic producers more compeiive. This causaliy can also be observed more sricly in he commodiies fuures markes. In some sudies he causaliy is clear beween agriculural expors and exchange rae, while ohers find ha he exchange rae has relaively small effec on he agriculure rade. For insance, Barichello e al. (1998) found ha he commodiies hey examined become more expor compeiive wih he Indonesia s currency depreciaion. Kapombe and Colyer (1999) examined U.S. broiler expors, using a muliple equaion srucural ime series model and concluded ha here is a negaive relaionship beween he exchange rae movemens and he Japanese demand for U.S. broiler expors. Oyejide (1986) used Ordinary Leas Squares (OLS) for he period beween o es he effecs of exchange rae and rade policies on Nigeria s agriculural expor and found ha here is an inverse relaionship beween he real exchange rae and non-oil expors. Akram (2009) used srucural VAR models o show ha a weaker dollar and a reducion in he real ineres raes lead o higher commodiy prices. More recenly, Frank and Garcia (2010), using 36

2 weekly daa from 1998 o 2008 divided ino wo sub-periods (i.e and ), developed a new empirical approach and concluded ha here is a significanly causal effec from he crude oil prices o agriculural commodiy prices for he second sub-period raher han he firs sub-period. 2. Theoreical Backround Schuh argued ha any change in exchange rae policy plays a prominen role in agriculural expors because of is relaive expense in oher counries. The overvalued dollar led o depressed prices and lower farm profis and a higher rae of oupu supply. Schuh assered his belief ha an imporan share of he rise in agriculural prices in mid-1973 is a resul of moneary phenomena which induced an expor boom in an economy ha was already responding o expansive moneary policies, and in he case of agriculure, increased he foreign demand for U.S. oupu a he same ime ha his demand was already rising from emporary bad weaher condiions in oher counries and a emporary decline in he Peruvian fishmeal indusry" (Schuh, 1974, p. 12). Knowing he effecs of he value of he dollar on U.S. agriculural expors, Schuh s view was ha he overvalued dollar had srongly affeced he demand for aggregae U.S. agriculure expors. Obviously, hese effecs play a vial role in undersanding he impac of policies direced oward economic developmen. In order o undersand he role of exchange rae movemens in he dynamics of agriculural expors/impors, we should ake a closer look a Figure 1, which shows monhly movemens of he quaniy of Indian agriculural expors from 1961 hrough Augus 2013 using ime series provided by he FAOSTAT, where RER and EXPQUANT are he real exchange rae of US dollar and he index of agriculural expor quaniy of India. As is apparen from he figure ha shows he degree of co-movemen, as measured by urning poins common o he exchange rae of dollar and agriculural expor quaniy index, if here is depreciaion in Indian Ruppee, dollar price will rise and his will evenually lead o an increase in expor quaniy. Ghosh (1999, 2008, 2013) lieraure has also sudied he srucural changes and growh in India and repored ha his esimaes show significan srucural breaks in Indian agriculure in and and concluded ha bes growh performance in agriculure secor was during 1980s. RER EXPQUANT Figure 1. Real exchange rae and agriculural expor quaniy index Agriculural expor has been he backbone of India s expor effors because conribues abou 12.3 percen of Gross Domesic Produc (GDP) and provides employmen o abou 58 percen of oal populaion. Moreover, India is also among he world's larges producers and consumers of a range of crop and livesock commodiies, and ranked 17h as an agriculural imporer and 13h as an exporer during (Landes, 2016). The share of expors in GDP increased from 24 percen in o 24.8 percen in , while he share of impors declined from 30.7 percen o 28.4 percen, resuling in improvemen in ne expors by 3.1 percen poins of he GDP (KPMG, 2014). Finally, exchange rae reforms iniiaed in 1991 has conribued o he significan increases in expors of radable agriculure and foreign exchange reserve posiions. 3. Economeric Mehodology Granger (1969) defines causaliy beween wo saionary series x and y in erms of predicabiliy. Le y and x represen wo saionary and ergodic ime series processes. We are ineresed in exploring wheher here is a Granger-causal relaionship beween he wo series wihou resricing he analysis o a specific funcional form. For he null hypohesis variable x canno Granger cause y (Granger, 1988). m Y Y X i i j j i 1 j 1 n (1) 37

3 where Y and X are he firs difference of Y and X, respecively, is a consan, m and n are he lag lenghs enough o make disurbance erm whie noise, and is ime. The firs sep in his analysis concerns he saionary of he series. Since he Granger (1969) approach requires he variables in he sysem o be saionary, he firs sep should be o deermine wheher or no he variables conain uni roo. We apply he augmened Dickey Fuller (ADF) es in our sudy and selec he lag order by he Akaike Informaion Crierion and Schwarz Informaion Crierion. Furhermore, if wo series are no coinegraed, hen we can conclude ha a long-erm economic relaionship exiss beween hem and hey move ogeher in he long run. As a resul, we should employ he causaliy es based on a VAR specificaion raher han a VECM. On he oher hand, one of he mos prominen of he nonlinear causaliy ess is he modified version of he Baek and Brock (1992) es which was developed by Hiemsra and Jones (1994) who allow for he exisence of shor-erm auocorrelaions in ime series and laely modified by Diks and Panchenko (2006). While he DP es has he abiliy o deec nonlinear causal relaionships, i does no provide any informaion on he source of nonlinear dependence. 4. Daa and Empirical Resuls We examine empirically invesigae he causal link beween he real exchange rae and agriculural expors in India employing a linear and nonlinear causaliy analysis. We carry ou our invesigaion using annual agriculural expor quaniy in India and he real Indian Rupee-US Dollar exchange rae daa ime series daa from 1961 o The daa required was colleced from Food and Agriculure Organizaion of he Unied Naions (FAOSTAT) and Reserve Bank of India. Fig.2 shows he illusraions of evoluions of exchange rae and he quaniy of agriculural expor in India. Noe ha here appears o be srong co-movemen beween expors and exchange rae. However, wheher his co-movemen relaes o causaion will be examined using he relevan ess. Prior o invesigaing Granger causaliy, Table 1, he resuls of uni roo ess based on Dickey and Fuller (1979) (ADF) and Phillips and Perron (1988) (PP) repors ha he saisics significanly confirm ha he level values of boh series are non-saionary; ha is, boh variables are I(1). Table 1. Uni Roo Tes RER EXQ ΔRER ΔEXQ Inercep only ADF *** *** PP *** *** Trend and inercep ADF *** *** PP *** *** Noe: The opimal lag lenghs are chosen based on Schwarz informaion crierion (SIC). * rejecion of he null hypohesis a 10% significance level. **rejecion of he null hypohesis a 5% significance level. *** rejecion of he null hypohesis a 1% significance level. The nex sep is o es wheher a long-run relaionship exiss beween he variables. As Engle and Granger (1987) saes ha, since our series are co-inegraed, he causaliy es needs o be applied on a vecor error correcion model (VECM) raher han an unresriced vecor auoregressive model (VAR). Therefore, we should keep in mind ha, before examining causaliy relaionship, we should firs check he coinegraion relaionship beween he series. Thus, he Johansen es of coinegraion has been conduced o es he long-run coinegraing relaion of agriculural expor wih exchange rae wih appropriae assumpions on rends and lags. Table 2 shows ha exchange rae is no coinegraed wih agriculural expor; ha is, here is no long-run relaionship beween exchange rae and agriculural expor. 38

4 Table 2. Johansen Coinegrain Tes Null Hypohesis Trace Saisics Max-eigen saisic None A mos Based on he firs differences of he wo series, i can be seen from he Table 3 ha he resuls are no supporive of a Granger-causal relaionship beween real exchange rae and agriculural expor quaniy. Tha is, unidirecional/bidirecional causaliy relaionship does no exis beween real exchange rae and agriculural expor in India. Table 3 also repors ha linear causaliy behavior beween agriculural expor and exchange rae does no exis afer he procedure of VAR filering. Table 3. Resuls of Linear Causaliy Analysis Null Hypohesis F-Saisic 5% Criical Value Noe: Poenial heeroscedasiciy of he error erm is correced by using Whie robus sandard errors. The lag lengh selecion is based on Schwarz Informaion Crierion (SIC). To confirm nonlinear linkages beween he series, firsly we employ a BDS es and show he resuls in Table 4. The BDS approach essenially ess for deviaions from idenically and independenly disribued (i.i.d.) behavior in ime series. We choose he lags I x =I y =1, he consan C for he bandwidh is se a 1.5 and le he embedding dimension (m) vary from 2 o 6 1. Because he p values are below 0.05, he exisence of non-linear dependencies has been confirmed, he residuals being no normally disribued. Tha is, he agriculural expor and exchange rae display significan nonlinear properies. Table 4. Lineariy Tes Resuls Dimension BDS Saisic Sd. Error z-saisic Prob RER EXQ Noe: The 10%, 5%, and 1% criical values are 1.645, 1.960, and 2.575, respecively. See Brock e al. (1995) The sudy now proceeds wih he invesigaion of he nonlinear causal connecion beween he series. Following Bekiros and Diks (2008), i is possible o examine he nonlinear srucure of he series in wo seps; as firs sep, he DP nonparameric Granger causaliy es is employed, hen, as second sep, o deec sric nonlinear causaliy in naure, he es is reapplied o he filered VAR residuals. As Hiemsra and Jones (1994) menioned, afer removing linear causaliy wih a VAR model, any causal linkage beween he residuals of VAR model can be considered as nonlinear predicive power. Considering he relaively small sample size of our daa, we se he value of he bandwidh equal o 1.5, based on he suggesion of Diks and Panchenko (2006) and he resuls are discussed for one lag (lx=ly=1). Table 4 provides he lineariy es resuls in order o examine wheher any remaining causaliy relaionships are sricly nonlinear. The resuls, repored in he corresponding panels of Table 5, show ha nonlinear causaliy relaionships do no exis beween exchange rae and agriculural expor. Our resuls of linear and non-linear causaliy behaviors do no suppor he saemen ha currency shocks are exogenous o he agriculural expors in India. The depreciaion or appreciaion of Rupee has no srong influence on India s quaniy of agriculural expors. Therefore, we can say ha increases in he 1 The opimal lag lenghs (m) were auomaically deermined wihin he model. 39

5 quaniy of agriculural expors in India are mainly caused by economic growh, raher han he flucuaions in he exchange rae of Indian Rupee. Table 5. Resuls of Nonlinear Causaliy Analysis. Null Hypohesis T-Saisic 5% Criical Value Noe: The common lag lengh is l x =l y =1 5. Conclusion This paper has examined he Granger causaliy beween he exchange rae (Indian Rupee versus U.S. dollar) and agriculural expors in India. Because of he srong effec of real exchange rae on he agriculural expors, a policy ha dampens he real exchange rae can enhance agriculural expors. Since macroeconomic condiions and policies have a grea role in he deerminaion of domesic agriculural secor developmen, exchange rae policies have also imporan repercussions on inernaional agriculural rade. The main goal of his work was o invesigae abou he link beween he quaniy of he agriculural expors and exchange rae in India during To give empirical evidence o his hypohesis he ime series analysis was performed and co-inegraion and causaliy nexus were explored using linear and non linear Granger causaliy approaches. Alhough here is no direc link beween Granger causaliy and causaion in economic sense, our findings sugges ha he real exchange rae and he quaniy of agriculural expors are no co-inegraed; ha is, here is no long run relaionship beween exchange raes and agriculural expors in India. The resuls of boh he linear and nonlinear Granger causaliy analysis sugges o accep he presence of neuraliy hypohesis in he Indian markes which means ha he real U.S. exchange rae and he agriculural commodiies do no cause each oher in a sricly linear and nonlinear sense. Tha is, our invesigaion does no provide any evidence of bidirecional or unidirecional causaliy beween he agriculural expors o real exchange rae in India. In oher words,, he real exchange rae does no have predicive power for agriculural expors. The drop in he predicive power of exchange rae in explaining he increase or decrease in he quaniy of agriculural expors in India may be aribued o he produciviy and efficiency in he agriculural secor as he economy expands over ime. While i is difficul o quanify he ne effecs of exchange rae policies on agriculural rade, i is unrealisic o seperae he exchange rae policies and srenghening compeiiveness of agriculure expors. I is obvious ha one of he mos damaging consequences of overvaluaion of a counry s currency is he undesirable impac on agriculural rade of ha counry. From he emirical findings in his aricle, i can be concluded ha alhough exchange rae variabiliy canno be blamed for he loss of compeiiveness of agriculural secor in India, agriculural expors value can be improved furher if he governemn enhances expor compeiiveness and encourage farmers wih financial incenives o use highsophisicaed echnologies. References Akram, Q. F. (2009). Commodiy prices, ineres raes and he dollar. Energy Economics, 31, hps://doi.org/ /j.eneco Baek, E. G., & Brock, A. W. (1992). A General Tes for Non-Linear Granger Causaliy: Bivariae Model. Technical Repor, Korean Developmen Insiue and Universiy of Wisconsin-Madison. Barichello, R., Pearson, S., & Selim, M. (1998). The Impac of he Indonesian Macroeconomic Crisis on Agriculural Profiabilies. Paper Prepared for USAID Mission, Jakara, Indonesia. Bekiros, S. D., & Diks, C. G. H. (2008). The nonlinear dynamic relaionship of exchange raes: Parameric and nonparameric causaliy esing. Journal of Macroeconomics, 30, hps://doi.org/ /j.jmacro Brock, W., Decher, W. D., Lebaron, B., & Scheinkman, J. (1995). A Tes for Independence Based on he Correlaion Dimension, Working papers, Wisconsin Madison - Social Sysems. Dickey, D., & Fuller, W. (1979). Disribuion of he Esimao rs for Auoregressive Time Series wih a Uni Roo. Journal of he American Saisical Associaion, 74,

6 Diks, C., & Panchenko, V. (2006). A new saisic and pracical guidelines for nonparameric Granger causaliy esing. Journal of Economic Dynamics and Conrol, 30, hps://doi.org/ /j.jedc Engle, R. F., & Granger, C. W. J. (1987). Coinegraion and Error Correcion: Represenaion, Esimaion and Tesing. Economerica, 55, hps://doi.org/ / Frank, J., & Garcia, P. (2010). How srong are he linkages among agriculural, oil, and exchange rae markes?. Proceedings of he NCCC-134 Conference on Applied Commodiy Price Analysis, Forecasing, and Marke Risk Managemen, S. Louis, US. Ghosh, M. (1999). Srucural Break and Uni Roo in Macroeconomic Time Series: Evidence from a Developing Economy. Sankhya (The Indian Journal of Saisics), 61, Par 2, Series B, Ghosh, M. (2008). Economic Reforms and Indian Economic Developmen Seleced Essays, Bookwell, New Delhi. Ghosh, M. (2013). Srucural Breaks and Performance in Agriculure. In: Liberalizaion, Growh and Regional Dispariies in India. India Sudies in Business and Economics. Springer, India. hps://doi.org/ / _6 Granger, C. (1969). Invesigaing Causal Relaions by Economeric Models and Cross-specral Mehods. Economerica, 37(3), hps://doi.org/ / Granger, C. (1988). Some Recen Developmens in A Concep of Causaliy. Journal of Economerics, 39, hps://doi.org/ / (88) Hiemsra, C., & Jones, J. D. (1994). Tesing for linear and nonlinear Granger causaliy in he sock price-volume relaion. Journal of Finance, 49, Indian Economic Survey Key Highlighs, KPMG in India. (July, 2014). Rerieved from Kapombe, C. M., & Colyer, D. (1999). A Srucural Time Series Analysis of U.S. Broiler Expors. American Journal of Agriculural Economics, 21, hps://doi.org/ /s (99) Landes, M. (2016). USDA s Economic Research Service. Rerieved from hp:// Mahew, S., Terry, R., & Agapi, S. (2006). Exchange raes, foreign income, and US agriculure. Bullein Number 06-1, Economic Developmen Cener, Deparmen of Applied Economics, Universiy of Minnesoa, Minneapolis. Oyejide, T. A. (1986). The Effecs of Trade and Exchange rae Policies on Agriculure in Nigeria. IFPRI Research Repor, 55. Phillip, P., & Perron, P. (1988). Tesing for a Uni Roo in Time Series Regression. Biomerika, 75(2), hps://doi.org/ /biome/ Schuh, G. E. (1974). The exchange rae and US agriculure. American Journal of Agriculural Economics, 56(1), hps://doi.org/ / Susani, Y. F. (2001). The effec of exchange rae on Indonesian agriculural expors. PhD disseraion, Oklahoma Sae Universiy, Sillwaer, OK. Copyrighs Copyrigh for his aricle is reained by he auhor(s), wih firs publicaion righs graned o he journal. This is an open-access aricle disribued under he erms and condiions of he Creaive Commons Aribuion license which permis unresriced use, disribuion, and reproducion in any medium, provided he original work is properly cied. 41