Inernaional Journal of Buine, Humaniie and Technology Vol. 3 No. 5; May 13 The Deerminan of Supply of Kenya Major Agriculural Crop Expor from 1963 o 1 Leniy Kananu Maugu Lecurer in Economic, Chuka Univeriy P.O Box 19-6 Chuka Kenya Rael Mwirigi Lecurer in Markeing, Chuka Univeriy P.O Box 19-6 Chuka Kenya John Njoroge Maara Lecurer in Economic, Chuka Univeriy P.O Box 19-6 Chuka Kenya Neba Galo Lecurer in Finance, Chuka Univeriy P.O Box 19-6 Chuka Kenya Abrac Kenya ha been experiencing low expor growh rae in general and agriculural expor in paricular and ye an increae in agriculural crop expor can conribue ignificanly o economic growh and improve he ciizen welfare. Thi udy inveigaed he deerminan of agriculural crop expor upply for Kenya over he period 1963-1. Annual ime erie daa were colleced from Kenya Saiical Abrac and he IMF Inernaional Financial Saiic (IFS). A diequilibrium model of agriculural crop expor wa ued. The regreion reul howed ha he real exchange rae wa a ignifican deerminan of ea, pyrehrum and horiculure expor bu no of coffee expor. Producive capaciy a proxied by GDP wa found o be a ignifican in deermining coffee, ea and aggregae expor. El-Nino rainfall, a capured by a dummy, wa ignifican for coffee expor, while rade liberalizaion, alo capured by a dummy wa only ignifican in deermining pyrehrum expor. Key word: Agriculural, Expor, Real exchange rae, Kenya Inroducion Like in many Sub Saharan African counrie, agriculure ha an imporan role in Kenya overall economic developmen. Kenya agriculural ecor make everal pecific conribuion o he economy. Fir, agriculural expor conribue over 6 percen o oal annual foreign exchange earning (Republic of Kenya. 1996). Thi i imporan becaue a noed by Tom and Faick (1967), many le developed counrie depend heavily upon agriculural expor for foreign exchange earning o finance impor of capial good. Secondly he ecor aborb a large proporion of he labour force in rural area ha i, abou 8 percen, and mo of hem depend on he agriculural ecor for heir livelihood (Republic of Kenya, 199). Moreover, agriculure i he ource of food for he counry, wih more han 8 percen of he food conumed coming from local producion (Republic of Kenya, 199). Finally, agriculure i a ource of raw maerial for agro-baed indurie. Agriculural expor in Sub-Saharan Africa have declined ignificanly in he la wo decade. The region hare of global agriculural expor ha declined from 8.% o 3.6% in. 5
Cener for Promoing Idea, USA www.ijbhne.com For he very poor counrie boh manufacured and agriculural expor have declined (Tonia and John, 5). Like in many Sub Saharan counrie, agriculural expor in Kenya have been declining. The annual volume of Kenya agriculural expor for major crop i hown below for he year 1963-1. Toal agriculural expor (on) 9 8 7 6 5 3 1 1963 1966 1969 197 1975 1978 1981 198 1987 199 1993 1996 1999 5 8 year Fig 1.1 Graph of Kenya agriculural expor for major crop, 1963-1 (in on). In he fir wo decade afer independence, agriculural expor were hi by boh negaive and poiive exernal hock in he form of wo oil crie (1973 and 1978) and he 1976/77 coffee boom, repecively. Oil crie led o decline in GDP, which i one of he deerminan of expor. In 197, here wa an increae in agriculural expor which coincided wih he rade liberalizaion of 197 (Republic of Kenya, 198). The increae in 1976 wa a a reul of inroducion of expor promoion aciviie. For example, he Expor Compenaion Scheme (ECS) in 1976 and eablihmen of he Kenya Exernal Trade Auhoriy (KETA), which renghened and reorganized expor promoion. There wa a decline in he agriculural expor beween 198 and 1985, which wa a a reul of he removal of bonu rae of 15 percen o hoe exporer who had increaed heir expor he previou year and he impac of 198 drough. Agriculural expor marked an increae in 1993, when hey roe harply o reach he highe peak. Thi coincided wih he removal of foreign exchange conrol, rade barrier and progreive implemenaion of reform. The poor performance in 1996-98 wa accompanied by ignifican reducion in quaniie and price of mo expor iem (Republic of Kenya, 1999). Thi perhap wa due o bad wheher condiion, ha i, abnormal rain in 1997/98. The increae in 1999 wa a a reul of El-Nino rainfall. The governmen of Kenya recognize ha increae in agriculural crop expor can conribue ignificanly o economic growh and welfare (Republic of Kenya, 1986). Indeed, ome of he economic reform purued by he governmen were aimed a expor promoion hrough expor crop uch a coffee, ea and horiculure and diverificaion of expor. However, Kenya ha experienced low expor growh rae in general, and agriculural expor in paricular (Republic of Kenya, 1). Thi i depie he meaure aken by he governmen o boo expor uch a expor promoion cheme. In heir udie of expor in Kenya, Okore (1987) and Ng eno (1991), focued on manufacured expor and aggregae expor repecively, on he bai of daa running up o mid 198. Thi udy focued on agriculural expor upply. The ample period encompae pre and po liberalizaion period o ake ino conideraion he likely impac of rade liberalizaion on he upply of agriculural expor, which ha no been done in Kenya. In addiion he udy employed recen advance in ime erie economeric modeling o boo i predicive power. Maerial and mehod Modeling agriculural expor funcion The upply of agriculural expor i aumed o be influenced by relaive price beween raded and non raded good i.e. real exchange rae, domeic capaciy variable proxied by GDP, rainfall and rade liberalizaion, he wo laer one being capured by dummie. 55
Inernaional Journal of Buine, Humaniie and Technology Vol. 3 No. 5; May 13 In hi udy he modeling of agriculural crop expor follow Khan (197) and Goldein and Khan (1985) procedure. Two verion of he baic model of he deerminan of he volume of expor are conidered. The fir i an equilibrium model ha aume ha here are no lag in he repone of expor o change in i deerminan. Thi ugge ha he adjumen of he acual level of expor o mee deired level of expor i inananeou. Given he profi funcion ( p), he ne upply funcion can be derived from i uing Hoelling lemma (Varian, 199). The profi funcion i differeniaed wih repec o price o give he upply funcion a hown below. Le ( p) be he firm ne upply funcion for good i. where p i he price. y i Then ( p) = л(p)/ p i.3.1 y i For i=1,,n Auming ha he derivaive exi and ha p i > The fir derivaive of he profi funcion wih repec o price a ha price: y ( p ) = л ( p )/ p p mu equal he maximizing facor upply a In he following model real exchange rae repreen he price and he upply funcion for expor a pecified by Goldein and Khan (1985) i a follow; ln X 1 ln RER ln CAP 3DR DT e..3. Where; ln X : The log of volume of agriculural crop expor upplied ln RER : The log of real exchange rae ln CAP : The log of producive capaciy DR: Dummy variable for rainfall DT: Dummy variable for rade liberalizaion are elaiciie 1 3, The diequilibrium model i employed o accoun for he poibiliy ha adjumen of acual equilibrium volume may ake place wih ome delay. Thi model aume he exience of ime lag in he repone of expor upply o change in i deerminan. Goldein and Khan uggeed he parial adjumen mechanim given below. ln X ln X X ) Where; 1 (ln X ln 1 X : Acual level of expor X : Deired level of expor : Coefficien of adjumen. I lie beween and 1 The adjumen funcion aume ha expor adju only parially o he difference beween expor upply in period and he acual expor in he previou period -1. Subiuing 3. ino 3.1 yield he following equaion: ln X ln RER ln CAP DR DT ln X 1 3 Where lnx i he log of volume of agriculural expor in year and X 1 i he log of lagged volume of agriculural expor. All he oher variable are a defined earlier. 1 1, 3 3 5 1 5 1 e 56
Cener for Promoing Idea, USA www.ijbhne.com.since i poiive and given ha 1 3,, and 5 are poiive, i follow ha 1 3, and 5 are poiive Eimaion echnique Time erie daa for he ample period 1963-1 were ued. For Ordinary Lea Square (OLS) mehod o be ued o eimae he model pecified, aionariy e for variable wa performed. Thi i becaue if he daa were non aionary he reul would have been puriou, ha i, wih a large R-quared and ignifican -aiic being obained ye would have no economic meaning (Gujarai, 1995). a) Graphical inpecion mehod-the rend of all variable included in he model were inpeced for aionariy, bu baing a deciion on wheher or no a erie i non-aionary by caual inpecion may be mileading. Therefore, Augmened Dickey Fuller (ADF) and Philip Perron (PP) e for aionariy were conduced. b) Te of he uni roo hypohei-the preence of uni roo wa performed uing he Augmened Dickey Fuller (ADF) and Phillip Perron (PP) e baed on he following model (Gujarai, 1995). y y a a y 1 e1 y 1 a e1 Random walk wih a drif....3. Random walk wih a drif and a rend....3.5 The ADF e aumed ha he daa generaing proce wa auogreive of he fir order. Thi wa done o ha auocorrelaion in he error erm doe no bia he e. The ADF include fir-difference lag in uch a way ha he error erm i diribued a a whie noie. The e i baed on he regreion model of he form; y a T y y.3.6 1 j j e1 j1 Where y i he variable eed for uni roo, j i lag lengh and T i ime rend. Uni roo e implied eing he null hypohei ha =1 (non aionary).the abolue value of convenionally compued aiic known a (au) compared wih he criical value. If he abolue value of au aiic exceeded he criical value of ADF, he null hypohei ha he given erie wa non aionary wa rejeced. I wa concluded ha he variable wa aionary. The number of ime he original erie wa differenced before i became aionary wa he order of inegraion. The diribuion heory governing he ADF e aume ha he error erm are aiically independen and have a conan variance. The PP e i an improvemen of ADF ha allow he mild aumpion concerning he diribuion of error. The PP e on i par addree he problem of he unknown rucure of daa generaing proce under he null hypohei by adjuing he -aiic for he poenial omied variable bia reul (Gujarai, 1995). The PP e i baed on he regreion; y 3.7 y 1 1 To e for a uni roo, equaion 3.7 wa eimaed by OLS and he -aiic of wa correced for erial correlaion. If he reul of he uni roo e howed ha he variable were no aionary in heir level, hen a coinegraion analyi wa performed. Daa To achieve he objecive of he udy, annual ime erie daa for period 1963 o 9 were colleced. Daa wa gahered from variou econdary ource namely; Inernaional Financial Saiic (IFS) and Kenya Saiical Abrac. Reul and Dicuion a) Graph inpecion mehod-an inpecion of graph for he variable howed ha flucuaion and volailiy characerized heir behaviour. Thi ugge ha all he variable exhibi non- conan mean and variance. However baing a deciion on wheher or no a erie i non-aionary by caual inpecion may be mileading. Appropriae mehod of eing for aionariy i Augmened Dickey Fuller (ADF) and Phillip Perron (PP) e. b) Uni roo e -Uni roo e indicaed ha he erie for he volume of expor of coffee, ea, pyrehrum, horiculure and real exchange rae are inegraed of order zero, ha i, I(), meaning ha hey are aionary a level. 57
Inernaional Journal of Buine, Humaniie and Technology Vol. 3 No. 5; May 13 Producive capaciy i inegraed of order one, ha i, I(1) meaning ha i i aionary a i fir difference. Thee finding indicae ha he lef hand ide of he model i an I() which implie ha he regreion uing Ordinary Lea Square (OLS) i meaningful, hence no valuable long-erm informaion wa o be lo if hey were o be differenced. Since all variable apar from producive capaciy in he model were found o be aionary, coinegraion e wa no neceary. Model elecion crieria Before eimaing he model, a deciion wa made regarding he number of lag o be included. The Akaike Informaion Crierion (AIC) and he Schwarz Bayeian Crierion (SBC) were ued for hi purpoe. From he reul AIC and SBC poined o one (1) a he appropriae lag lengh for ea expor and he aggregae model (volume of ea, coffee, horiculure and pyrehrum). For coffee and horiculure expor model, AIC wa minimal when he number of lag equal one, while SBC when number of lag equal wo. For boh model, he appropriae number of lag choen wa wo ince SBC i a uperior crieria han AIC. For pyrehrum expor model, he be lag lengh wa wo ince i wa he lowe value of he informaion crieria. Diagnoic e for regreion reidual Before reporing he regreion reul for he model eimaed, diagnoic e reul were preened and dicued fir. The J-B normaliy e wa applied o e wheher he erie i normally diribued or no. The null hypohei ae ha he regreion reidual are normally diribued. A mall probabiliy value (p-value) for J-B coefficien lead o he rejecion of hi hypohei. The reul a hown by he able below how ha he aggregae of he eleced crop, individual crop; ea, horiculure, pyrehrum and coffee expor model give J-B e aiic of.5,.1, 3.5,.7 and.6 wih probabiliy value of.57,.67,.55,.66 and.56 repecively. The null hypohee were acceped ince probabiliie are greaer han.5, which indicae ha he reidual are normally diribued. To e wheher reidual were erially correlaed LM e wa applied. The null hypohei of he LM e i ha here i no erial correlaion up o lag order p, where p i a pre-pecified ineger (Godfrey, 1988). A mall probabiliy value of le han.5 lead o he rejecion of hi hypohei. The reul are hown in he able1 below. Thee reul how ha he probabiliy value are.1,.35,.7,.3 and.3 for ea, horiculure, pyrehrum and coffee expor model repecively. Thi lead o accepance of he null hypohei. Hence he reidual were no erially correlaed. Furher, o e wheher here i Auoregreive Condiional Heerokedaiciy (ARCH) in he reidual he ARCH e wa applied. The null hypohei i ha here i no auoregreive condiional heerokedaiciy in he reidual. To accep he null hypohei, he probabiliy mu be greaer han.5. The e reul preened in he able below how ha he probabiliie of all he expor model are greaer han.5 hence here are no ARCH preen in he reidual of he model. Whie Heerokedaiciy e wa applied o e for heerokedaiciy in he reidual from a lea quare regreion. Whie e i a e of he null hypohei of no heerokedaiciy again heerokedaiciy of ome unknown general form (Whie, 198). A hown in he able1 below he probabiliie of hi e for aggregae of eleced crop, ea, pyrehrum, horiculure and coffee expor model are.58,.55,.8,.7 and.56 repecively. The reul how ha here i no heerokedaiciy preen in he reidual ince he probabiliie of hi e for all expor model are more han.5. Ree which and for Regreion Specificaion Error Te wa o e for omied variable, incorrec funcional form of a model and correlaion beween he independen variable and he error erm. The null hypohei ae ha here i no mipecificaion in he model. A mall probabiliy of hi e lead o rejecion of he null hypohei. The reul of all model rejeced he null hypohei ince hey all had probabiliy value greaer han.5. Therefore, all he model were well pecified and parameer were no omied. Chow foreca e wa ued o e for rucural change. The null hypohei of hi e i ha here i no rucural change. In all model he null hypohee are acceped a 5% ince hey have probabiliie greaer han.5. Hence here were no rucural change in he erie and he model could afely be ued for forecaing and predicion. 58
Cener for Promoing Idea, USA www.ijbhne.com Table 1: Diagnoic e reul Te ype Aggregae Tea Pyrehrum Horiculure Coffee Jarque-Bera normaliy e Te aiic.5.1.7 3.5.6 probabiliy.57.67.55.66.56 Whie Heerokedaiciy e Te aiic.3 1.5.11.1 1.56 probabiliy.58.55.8.7.56 Serial correlaion LM e Te aiic.1 1.5 3.57 3..11 probabiliy.1.35.7.3.3 Ramey RESET e Te aiic 1.1.1.16 5.5.3 probabiliy.88.5.8.1. ARCH LM e Te aiic.8 1.5 3.67.11.1 probabiliy.68.5.8.73.69 Chow foreca e Te aiic.6 3.5 3.11.9.17 probabiliy.67.78.7.53.6 Eimaed model and reul The regreion reul for he coffee for he coffee expor are preened in he able below. Table : Regreion reul for he coffee expor model Independen Variable Coefficien -Saiic Conan 9.. Fir difference of log GDP.8 3. Log real exchange rae -.99-1. Fir lag of log coffee expor.35-1. Fir lag of log GDP.5.13 Fir lag of log real exchange rae.69.98 Second lag of log coffee expor -.8-1.5 El-Nino 1998 D=1, oherwie.8 1.91 Coffee boom 1977 D=1, oherwie.71.66 Adjued R-quared Durbin-Waon a F-aiic Prob (F-aiic) -.63. 1.33.87.1 The Durbin Waon aiic (1.33) how he preence of erial correlaion beween he variable in he coffee expor model. However, baing a deciion on wheher or no he variable are erially correlaed uing D-W aiic may be mileading, becaue he D-W aiic e for Auoregreive (1) error. The appropriae e i LM e for correlaion, which e for higher order ARMA error a dicued earlier. The value of adjued R- quared i.63 indicaing ha he overall fi of he regreion i 63 percen, ha i, 63 percen of he variaion in coffee expor i explained by he variable ha were included in he model. Thi leave 37% of he variaion in coffee expor o be explained by exogenou variable. The probabiliy of he F-aiic i.1 herefore he null hypohei ha he coefficien are equal o zero i rejeced.the reul for he coffee expor model how ha coffee expor upply elaiciy wih repec o producive capaciy and fir lag of producive capaciy wa poiive and aiically ignifican. The repone of coffee o real exchange rae wa fairly elaic, negaive and inignifican. However, coffee expor were affeced poiively by lagged real exchange rae bu i wa alo a inignifican deerminan of agriculural crop expor. The lagged coffee expor coefficien wa alo found o be inignifican bu very elaic. To capure he exreme weaher condiion, a dummy variable wa inroduced o capure he El-Nino rainfall in 1998 and wa found ignifican. Tha i, El-Nino rainfall hifed he coffee expor upply curve upward. Coffee boom in 1977 had ignifican impac on coffee expor; by hifing he upply curve upward. 59
Inernaional Journal of Buine, Humaniie and Technology Vol. 3 No. 5; May 13 Hence all he variable a a group explain he coffee expor model. For ea expor upply model, he reul are preened in he able 3 below. Table 3: Regreion reul for ea expor model Independen Variable Coefficien -Saiic Conan.79.79 Fir difference of log GDP 1.7.67 Log real exchange rae.3.1 Fir lag of log ea expor.63 3.5 Fir lag of log GDP -.9 -.78 Fir lag of log real exchange rae -.1 -.7 Second lag of log ea expor.31 1.86 Adjued R-quared Durbin-Waon a F-aiic Prob (F-aiic).95.1 1.6 1.51. The model i well fied becaue he R-quared i very large, ha i, 95% of he variaion in ea expor are caued by change in he variable included in he model. The probabiliy of he F-aiic i., which lead o rejecion of he null hypohei ha he coefficien are no joinly equal o zero.the reul how ha fir difference of log GDP, log real exchange rae, fir and econd lag of ea expor had ignifican impac on upply of ea expor. Thi i conien wih he reul preened by Yang (1978). Berlow and Sene (1995) alo found imilar reul. The auhor concluded ha real exchange rae wa he ingle mo imporan facor explaining growh in he volume of agriculural expor. On he oher hand fir lag of GDP and fir lag of real exchange rae were aiically inignifican and heir elaiciie were very low. Thu, change in hee wo deerminan would no caue ubanial change in ea expor. The model ha follow i for horiculure expor upply. I reul are given in he able below. Table : Regreion reul for horiculure expor model Independen Variable Coefficien -Saiic Conan -1.3 -.9 Fir difference of log GDP.11 1.3 Log real exchange rae 1..7 Fir lag of log horiculure expor 1.11 5.8 Fir lag of log GDP.3.9 Fir lag of log real exchange rae.11 -.1 Second lag of log horiculure expor.7 -.39 Trade liberalizaion of 1973 D=1, oherwie.71.66 Adjued R-quared Durbin-Waon a F-aiic Prob (F-aiic).88.5 1.65 6.98. The model wa well fied wih R-quared of.88. Moreover, here wa erial correlaion beween he variable in horiculure expor model. Thi i evidenced by he DW aiic of 1.65. However, LM e wa ued. The probabiliy of F-aiic i. hence he coefficien are no equal o zero and he relaionhip beween dependen and independen variable i meaningful. From he regreion reul preened in he able, upply elaiciie of horiculure expor wih repec o log real exchange rae and i fir lagged expor volume were high and aiically ignifican. All he variable had expeced ign. Oher udie for example, Ng eno (1991) found imilar reul. 6
Cener for Promoing Idea, USA www.ijbhne.com For inance, real exchange rae wa found being ignifican in mo of he udie. The coefficien of fir difference of log GDP, fir lag of log GDP, fir lag of log real exchange rae and econd lag of log horiculure expor were inelaic and inignifican, bu hey had correc ign. Therefore change in hee variable would no caue ubanial change in expor of horiculure. Trade liberalizaion in 1973 had inignifican impac on horiculure expor. Table 5 how he regreion reul of pyrehrum expor upply. Table 5: Regreion reul for pyrehrum expor upply model Independen Variable Coefficien -Saiic Conan 6.8 1.76 Fir difference of log GDP 1.1.5 Log real exchange rae -1.8 -.5 Fir lag of log pyrehrum expor.5 3.3 Fir lag of log GDP -1.7 -.7 Fir lag of log real exchange rae.6.11 Second lag of log pyrehrum expor.1.96 Second lag of log GDP.13 -.9 Second lag of log real exchange rae..9 Third lag of log pyrehrum expor.76 1.89 Trade liberalizaion of 1973 D=1, oherwie.3.5 Adjued R-quared Durbin-Waon a F-aiic Prob (F-aiic).55.1 1.67 1.77. The reul for he pyrehrum expor model how ha expor upply elaiciie wih repec o fir difference of log GDP, log real exchange rae and fir lag of log GDP were high, bu only log real exchange rae wa elaic and aiically ignifican. Thu, pyrehrum expor reponded ubanially o change in real exchange rae. In conra o finding by Amin (1996), real exchange rae had negaive ign conrary o expecaion. The fir, econd and hird lag of log pyrehrum expor wa found ignifican deerminan bu hey were fairly inelaic. Thi mean ha, upply of pyrehrum expor depended on expor of previou year. All he oher variable had expeced ign apar from he fir lag of GDP. Trade liberalizaion in 1993 had ignifican impac on pyrehrum expor, by hifing he upply curve upward. The able 6 how he regreion reul of he aggregae model. Table 6: Regreion reul of aggregae of coffee, ea, horiculure and pyrehrum Independen Variable Coefficien -Saiic Conan 1.1. Fir difference of log GDP 1.7. Log real exchange rae -1.75 -. Fir lag of log aggregae expor.35.5 Fir lag of log GDP.5.13 Fir lag of log real exchange rae.69.97 Second lag of log aggregae expor.8 -. Adjued R-quared Durbin-Waon a F-aiic Prob (F-aiic) -.53. 1.39.87. From he reul, all variable had he expeced ign apar from log real exchange rae. However, real exchange rae i he mo ignifican variable influencing agriculural expor upply and i fairly elaic. Thi confirm he reul found by Amin (1996) ha a mall change in exchange rae will caue expor o change ignificanly. Fir difference of log GDP i alo aiically ignifican variable and fairly elaic. Thu, change in GDP would caue ubanial expor change. Ng eno (1991) alo found imilar reul. On he overall, he lagged expor variable coefficien were no aiically ignifican a 5% level, alhough hey had he expeced ign. Therefore, a Amin (1996) noed, he volume of expor upply wa no affeced by pa expor. 61
Inernaional Journal of Buine, Humaniie and Technology Vol. 3 No. 5; May 13 The Durbin Waon aiic (1.39) how he preence of erial correlaion beween he variable in he model. However, LM e for erial correlaion wa ued due o reaon given earlier. The value of he adjued R-quared i.53 indicaing ha he overall fi of he regreion i 53%, ha i, 53% of he variaion in oal agriculural crop expor i explained by he variable ha were included in he model. Concluion and recommendaion Since real exchange rae wa found ignifican in deerminaion of volume of expor for mo of he crop, he Governmen of Kenya hould conider an examinaion of he queion of depreciaion of he real exchange rae. The Cenral Bank of Kenya hould only inervene in foreign marke o limi undeirable flucuaion caued by mimache beween he upply and demand of foreign currencie. Thi will help o increae he volume of agriculural expor. Alo, flexible exchange rae will have expanionary effec by wiching demand away from impor and making expor more compeiive. The governmen hould develop policie ha focu on how o improve on he producive capaciy of hi counry in he hor erm ince, a mall change in GDP caue expor o change ignificanly. I hould pu meaure ha will increae rainfall for inance, emphai on planing of more ree o arac rain and renghen he policy ha no unauhorized cuing of ree. Acknowledgemen We are graeful o he following iniuion for all he uppor: Chuka Univeriy, Kenyaa Univeriy and Univeriy of Nairobi. We are alo graeful o African Economic Reearch Conorium (AERC) for heir uppor. Reference Amin, A. (1996). The Effec of Exchange Rae Policy on Cameroon Agriculural Compeiivene. AERC Reearch Paper No. Nairobi. AERC Berlow, R., & Sene, F. (1995). The Turkih Expor Boom; Ju Reward or Ju Lucky? Journal of Developmen Economic. Vol. 8: pp 111-133. Godfrey, L.G. (1988). Specificaion Te in Economeric, Cambridge Univeriy pre. Goldein, M., & Khan, S (1985). The Supply and Demand for Expor; A Simulaneou Approach. Review of Economic and Saiic, 6. pp 111-119. Gujarai, D.N. (1995). Baic Economic. Third Ediion, Singapore McGraw-Hill Inc.Khan, M. (197). Impor and Expor Demand in Developing Counrie. IMF Saff Paper, 1. pp 678-93. Ng eno, N.K. (1991). Kenya Expor Perfomance. Trade and Developmen in SSA. Mancheer Univeriy Pre UK. Vol 6 No. 31 pp98-11. Okore, J.O. (1987). Deerminan of Kenya Manufacured Expor; An Emperical Analyi. M.A Thei, Univeriy of Nairobi. Republic of Kenya, (198): Naional Developmen Plan 198-1985. Nairobi: Governmen Priner.Republic of Kenya, (1986): Budge Speech for he Year 1986/1987. Nairobi: Governmen Priner.Republic of Kenya, (1): Budge Speech for he Year 1/. Nairobi: Governmen Priner.Republic of Kenya. (1999): Naional Developmen Plan 1999-. Nairobi: Governmen Priner.Republic of Kenya, Saiical Abrac (1963-1), Nairobi: Governmen Priner.Tom, E., and Faick, J. (1967). World Trade in Seleced Agriculural Commodiie. The Philipine Economic Journal, Vol, pp. 159-17. Tonia, K., and John, R (5). Agriculural Expor: imporan Iue for SSA. Economic Journal. Vol., pp19-17. Whie, H. (198). A Heerokedaiciy-Conience Marix and a Direc Te for Heerokedaiciy. Economeric, 8, 817-838. Yang, Y. (1978), Manufacured Expor in Developing Counrie. World Developmen Vol.18, No. 3, pp 38-63. 6