Afrcan Journal of Agrcultural Research Vol. 5(9), pp. 805-812, 4 May 2010 Avalable onlne at http://www.academcjournals.org/ajar ISSN 1991-637X 2010 Academc Journals Full Length Research Paper Determnants of the place of sell and prce of kale for Kambu, Kenya Beatrce Salasya 1 and Kees Burger 2 1 Kenya Agrcultural Research Insttute, P.O. Box 169-50100, Kakamega, Kenya. 2 Development Economcs Group, Wagenngen Unversty and Research Centre (WUR). Accepted 20 Aprl, 2010 Kale s a major source of cash for many households n Kenya. A study of households n Kambu dstrct revealed that kale made the hghest contrbuton to household ncome among the crops. The farmers of Kambu sell ther kale ether n Narob, at farm gate, or at the local market and fetch dfferent prces. The farm gate prce s hghest when kale s sold n Narob, but only a small fracton of the farmers sell kale n Narob. Ths paper endeavored to know what nfluenced the decson of where to sell the kale, and why the prces were so varable between destnatons. The results showed that the prce of kale was nfluenced by the place of producton and dstance to a local market. The results further showed that the decson to sell kale n Narob where the prce was hghest depended on labor avalablty, the household s wealth status and transport access to the market. Poorer households were more lkely to seek hgher prces n the cty than the wealther households. Improvng marketng of kale could be a way of targetng the poorer households and mprovng ther welfare. Key words: Kale, market outlets, prce, Kambu. INTRODUCTION The lvelhood of many households the world over depends on the sale of agrcultural commodtes. Hortcultural crops such as kale are partcularly benefcal to smallholders because the benefts can be rpped wthn a short tme. Indeed the role of hortcultural crops n mprovng the welfare of rural households has receved a lot of emphass n Kenya and else where (Swernberg, 1995; Kmenye, 1995; Mnot and Ngg, 2002). Kale, a popular green vegetable consumed by almost every household n Kenya has been shown to be a major source of cash for many households. For example, results of a study n Kambu dstrct, whch neghbors Narob cty, revealed that kale made the hghest contrbuton to household ncome among the crops and also had the hghest returns to varable nputs among the crops (Salasya, 2005). However, even though kale made the hghest contrbuton to household ncome, not all the farmers sold kale n the cty where prces are hghest. Neven and Reardon (2005) found that smallholder farmers sell to brokers and get a prce that just lets them *Correspondng author. E-mal: beatsakwa@yahoo.com. Tel: +254 56 30031. break-even at best. The farmers n Kambu and other areas that surround Narob cty can utlze ther proxmty to the urban market to ncrease ther ncome from kale by ncreasng the sales made drectly to the cty. It has been noted that smallholders need dynamc markets to enable them escape from poverty, and urban markets are at present the most dynamc food markets due to ncreasng urban populaton and ncomes (Weatherspoon and Reardon, 2003). But as Neven and Reardon, 2004 observed, smallholders are squeezed out as ndrect supplers and are left to operate n the tradtonal system, whch fetches lower prces. In the dataset used n ths study there s a large varaton n the prce of kale, wth the prce rangng from less than 2 Kenya shllngs (ksh) per klo to 15 ksh per klo. The farmers sold ther kale ether n Narob, at farm gate, or at the local market and receved dfferent prces, wth some farmers gettng better prces than others. The farm gate prce was hghest when kale was sold n Narob, and lowest when kale s sold at the local market. A dsturbng queston then s why not all the farmers n Kambu sell ther kale n Narob where prces are hghest. Are there barrers preventng some farmers from accessng the Narob market? If as Fafchamps and Hll (2005) found, poor farmers receve lower prces because they cannot access markets wth
806 Afr. J. Agrc. Res. Table 1. Descrptve statstcs for varables used n the analyss. Varable descrpton (n=296) Mean Standard devaton Maxmum Mnmum Kale output prce (ksh./kg) 7.11 3.77 1.30 15.0 Kale output (kg) 3038 3092 25000 15 Farm sze n hectares (ha) 0.88 0.93 0.05 8.10 Educaton of household head (years n school) 8.68 5.08 0.00 16.0 Age of household head (years) 53.5 15.0 18.0 100 Number of adult males 1.93 1.37 0.00 9.00 Number of adult females 2.02 1.62 0.00 6.00 Number of chldren below 14 years 2.03 0.17 0.00 10.0 Cattle n tropcal lvestock unts (TLU) 1.29 0.12 0.00 13.7 Dstance to the nearest market (km) 2.11 1.63 0.01 10 Area under tea and/or coffee (hectares) 0.09 0.23 0.00 2.41 Dependency rato (chldren to adults) 0.62 0.65 0.00 3.33 Gender of the household head (% female headed) 22.3% Membershp of organzatons (% members) 87.1% Percentage ownng a bcycle 30.4% Percentage ownng a donkey cart 19.3% Percentage ownng a car 13.9% Man occupaton of household head (% farmng) 64.9% better prces, then ther welfare can be rased by offerng nsttutonal alternatves that enable them to access markets wth better prces. The paper nvestgates the factors nfluencng the farmers decsons on where to sell kale, and why dfferent farmers sell n dfferent market outlets and fetch dfferent prces. The paper thus follows the lnes of Fafchamps and Hll (2005) to nvestgate the relatonshp between the prce of kale and the market condtons facng dfferent households. Unlke coffee, whch Fafchamps and Hll nvestgated, kale s hghly pershable and hence the produce needs to reach the market wthn a short tme after harvestng. The specfc objectves of the paper are: to nvestgate what nfluences the farmers choce of whch market outlet to sell the kale to; examne why there s a large prce varaton n the prce offered at the dfferent market outlets and; to dentfy what characterstcs of kale nfluenced ts prce. Some of the questons to be answered are: whch households are more lkely to sell kale n Narob where prces are hghest and what needs to be done for most households to sell the kale n Narob? The remanng part of the paper s organzed as follows: Methodology whch descrbes the study area, data collecton methods and analytcal procedures. The results are presented and dscussed and the paper ends wth concluson. METHODOLOGY The study area The study s carred out n Kambu dstrct, one of the dstrcts n central provnce of Kenya, located just north of Narob Cty. Kambu was selected because t s located close to Narob cty and hence t was expected that farmers could easly carry ther produce to the cty wthout fear of spolage, consderng that kale s a pershable commodty, and wthout ncurrng hgh transacton costs. Average annual ranfall n Kambu vares wth alttude between 500 mm n the lower parts up to 2000 mm n the hgher forest zones (Jaetzold and Schmdt, 1983). The populaton densty s hgh wth an average of 562 persons per square klometer (CBS, 2001) but also very varable wth some areas havng populaton denstes exceedng 1500 persons/km 2. The average household has 3.9 persons lvng on 0.7 ha of land creatng a very hgh dependency on agrculture (ROK, 2005), whch s the man economc base. Data and data collecton Data collecton was through household ntervews, conducted n the local language by traned enumerators usng a pre-tested structured questonnare between October and December 2003. Sub-locatons (the smallest admnstratve unt n Kenya) fallng wthn hgh agrcultural potental areas of Kambu formed the samplng frame. The study was lmted to the hgh potental areas because those are the areas sutable for kale producton, and also t s possble to produce kale through out the year and hence can supply the urban market contnuously. Nne sub-locatons located n Gthungur, Lmuru and Lar dvsons were randomly selected. Two pars of major landmarks (permanent features such as a school, a church, a tradng centre) n each of the selected sub-locatons were randomly selected on a map and transect lnes drawn jonng each par. Samplng was thereafter done followng as closely as possble the marked transects by a traned enumerator. Every ffth household frst on the left and then on the rght was selected alternately and n ths way, a random sample from all sub-locatons was obtaned. The number selected from each sub-locaton was proportonal to the populaton densty of that sub-locaton. In total 297 households from the three dvsons (115 from Gthungur; 95 from Lmuru and 86 from Lar) were ntervewed. Data collected from each household
Salasya and Kees Burger 807 Table 2. The man destnatons and the prce of kale for the sample households. Number of households sellng Average quantty produced per household (kg) Average quantty sold per household (kg) Average retal prce (Ksh.) Average farm gate prce (ksh.) Total 156 3485 3191 7.21 7.09 Mddleman 118 3631 7.06 6.98 Open market 38 1823 7.78 7.42 Narob (all) 33 4070 3646 11.3 10.8 - To mddleman 15 4897 10.3 9.82 - In open market 18 2602 12.2 11.6 Local market (all) 29 2660 2488 4.36 4.23 - To mddleman 9 5525 5.61 5.46 - In open market 20 1121 3.79 3.69 Farm gate 94 3534 3248 6.68 6.68 Source: Calculatons from household survey data; NB: The quanttes and prces for one year. were: household data ncludng age, gender, educaton level and occupaton of the household head, household sze dsaggregated by age and gender, farm sze and off-farm ncome sources. Also collected were detaled producton data at ndvdual actvty level, ncludng use of labor. Insttutonal data on dstance to nput and output markets, prces of nputs and outputs, access to credt and extenson servces and organzatonal membershp were also collected. Descrptve statstcs of the data used n the estmatons are shown n Table 1. Analytcal tools Two analytcal approaches were followed. Frst the determnants of the prce of kale were examned by applyng the hedonc hypothess, and then the factors nfluencng the choce of where to sell the kale was nvestgated usng a probt model. Determnants of the prce of kale The Rosen s hedonc prcng model, was employed whch s based on the hedonc hypothess that each good s characterzed by the set of all ts characterstcs x = ( x1, x2,... xk ). It s assumed that the preferences of economc actors wth respect to any good are solely determned by the correspondng characterstcs vector of that good and that for any good, there s a functonal relatonshp between the prce p and ts characterstc vector x (Rosen, 1974) so that; p = f (x). (1) Ths functon specfes the hedonc relatonshp or the hedonc regresson typcal for any good. Hedonc models have been wdely used to evaluate the mplct prces of many agrcultural commodtes, e.g. (Baley and Peterson, 1991 and Lansford et al., 1998). Based on the functonal relatonshp hedonc prces can be calculated by takng the partal dervatves of the hedonc functon (Rosen, 1974). Hedonc regressons have also been appled on durable goods such as automobles (Couton et al., 1996) where varous techncal characterstcs of the car are ncluded n the hedonc prce equaton. It has also been appled on non durable goods wth dentfable characterstcs such as wne e.g. (Combrs et al., 1997). However n ths study we estmate the hedonc functon to dentfy how the dfferent characterstcs of kale, that are not necessarly nherent n the product tself, nfluence ts prce (sgnfcance and drecton of effect). In the estmaton we assume the ordnary lnear approach whch takes the form: p = k β 0 + β k x k (2) k = 1 Where p s the prce of kale and β k are coeffcents to be estmated and reflect the absolute prce of x k, and xk are the characterstcs of kale hypotheszed to nfluence ts prce. The characterstcs hypotheszed to nfluence the prce nclude, the place of sale, dstance to the nearest market, to whom kale s sold (mddleman or open market), and the locaton where t s produced. Results of the hedonc estmaton are presented n Table 4. The probt model Bnary choce models also called unvarate dchotomous models are the most commonly used to analyze decsons for or aganst a partcular practce (Verbeek, 2003). These models essentally descrbe the probablty that y = 1 drectly, although they are often derved from an underlyng latent varable model. The general form of the model may be stated as follows: p ( y 1 \ x ) = G( x, β ) = (3) Ths equaton says that the probablty of havng y = 1 depends on the vector x contanng ndvdual characterstcs e.g. ncome,
808 Afr. J. Agrc. Res. Table 3. Determnants of the prce of kale results of the hedonc equaton estmaton. Varable Coeffcent Std. error Constant 9.38*** 0.90 Where sold dummy (1 = Narob; 0 = elsewhere) 3.24*** 0.71 Where sold dummy (1= Local market; 0 = elsewhere) -2.48*** 0.95 To whom sold (1 = mddleman; 0 = open market) -1.19 0.78 Dstance to the nearest market (km) -0.51*** 0.18 Dvson dummy (1= Gthungur; 0 = Lar and Lmuru) -2.10** 0.84 Dvson dummy (1= Lmuru; 0 = Gthungur and Lar) -1.25** 0.55 Adjusted R-squared 0.45 N 152 Dependent varable = Farm gate prce of kale n ksh./kg. *, **, *** ndcate sgnfcance levels of 0.1, 0.05, and 0.001 respectvely. educaton level, age and martal status. The common models that emerge are ether the probt or the logt dependng on the dstrbuton functon chosen for the stochastc term. The two dstrbuton functons are smlar so that f one corrects for the dfference n scalng, the logt and probt models typcally yeld very smlar results n emprcal work. In ths paper we use the probt model to analyze farmers decsons to sell kale at the dfferent destnatons. The farmers could ether sell kale at the local (nearest market), at farm gate, or n the cty (Narob). Followng Verbeek (2003), the model s specfed as: * ' y = x β + ε, ε NID 0,1) ( (4) * y Is unobserved and s referred to as a latent varable. The assumpton s that an ndvdual farmer chooses to sell at a partcular destnaton when the utlty dfference of sellng there and not sellng there exceeds a certan threshold, zero n ths case, so that y = 1 (Sell at destnaton ) f and only f y * > 0 y = 0 f y * 0 The choce to sell at destnaton s affected by the varables whose coeffcent vector β, are the subject of estmaton; as usual the error termε s assumed to be ndependent of all x. The parameters are estmated by the method of maxmum lkelhood. The ndependent varables n the emprcal model ( x s) consst of dfferent sets of data hypotheszed to nfluence the farmer s decson to sell at destnaton ( y ) descrbed n Appendx 1. RESULTS AND DISCUSSION Determnants of the prce of kale ' x. These varables are Table 2 summarzes the quantty of kale produced and sold at the dfferent destnatons and the prces receved. The three man destnatons where kale s sold are Narob cty, the local market and farm gate. When the destnaton s Narob and/or the local market, kale s ether sold to a mddle man, or n the open market, whereas when the destnaton s farm gate t s always to a mddleman. It s obvous from Table 2 that the prce of kale vares manly based on where t was sold. The hghest prce s pad when kale s sold n Narob, whereas the lowest prce s pad when kale s sold at the local market. A t test comparng prces at one destnaton wth prces at the other two destnatons combned confrm that prces n Narob are sgnfcantly hgher and prces at the local market sgnfcantly lower both at (P<0.001) and the prces when sold at farm gate are sgnfcantly lower at (P<0.1). Results of the hedonc estmaton n Table 3 confrm that the prce of kale s sgnfcantly hgher n Narob and s sgnfcantly lower at the local market. The results further reveal that dstance to the nearest market has a sgnfcant negatve nfluence on the farm gate prce, the nearer the market the hgher the prce. Although kale s not necessarly sold at the nearest market, dstance to the nearest market s an ndcaton of accessblty e.g. to where publc transport can be avalable and thus has an nfluence on the transacton costs nvolved. There s no sgnfcant prce dfference between sellng n the open market and sellng to a mddleman. The results of the hedonc regresson (Table 3) also show that the prce of kale s sgnfcantly hgher for households located n Lar dvson than for those located n Gthungur or Lmuru dvsons. From the data, households n Lar manly sell ther kale ether at farm gate or n Narob whereas those n Gthungur sell manly at the local market and surprsngly none sells n Narob. Those n Lmuru sell manly at farm gate wth a few sellng n Narob and at the local market. Why then ths observed scenaro, that for example the household n Lar dvson should sell n Narob and fetch hgher prces and those n Gthungur sell at the local market and fetch lower prces? Results of the probt model n Table 3 shade more lght. The probt model results n Table 4
Salasya and Kees Burger 809 Table 4. Probt results on factors nfluencng the choce of where to sell kale. Varable Narob Local market Farm gate Constant -0.20 (1.07) 1.40 (1.35) -1.43 (0.92) Number of adult males 0.29** (0.14) -0.20 (0.26) -0.17 (0.12) Number of adult females 0.36*** (0.14) -0.05 (0.19) -0.22* (0.12) Dependency rato (chldren to adults) 0.44* (0.25) -0.17 (0.30) -0.14 (0.20) Dvson dummy (1= Lar; 0 = Gthungur and Lmuru) -2.64*** (0.64) -0.63 (0.41) Dvson dummy (1= Lmuru; 0 = Gthungur and Lar) -0.89*** (0.36) -0.94* (0.55) -0.61 (0.44) Educaton of farmer (years n school) -0.03 (0.04) 0.03 (0.05) 0.02 (0.03) Man occupaton of household head (1 = farmng; 0 = non-farmng) 0.40 (0.38) 0.06 (0.43) -0.12 (0.30) Age of the household head -0.03** (0.01) -0.03* (0.02) 0.04*** (0.01) Farm sze (ha) -0.20 (0.30) 0.02 (0.25) 0.19 (0.20) Membershp to organzaton (1 = s a member of an organzaton -0.29 (0.39) 0.05 (0.72) 0.20 (0.35) Dstance to the nearest market (km) -0.08 (0.14) 0.22 (0.15) -0.08 (0.10) Off-farm ncome ( 000s of ksh.)/year -0.01** (0.003) 0.001 (0.003) 0.003 (0.002) Gender of farmer (1= female) -0.06 (0.39) 0.37 (0.55) -0.24 (0.34) Area under tea and/or coffee -7.65* (3.95) 4.63* (2.60) -2.41 (1.70) If a car s owned (1=yes) 0.19 (0.40) -0.24 (0.62) -0.15 (0.31) If a bcycle s owned (1 = yes) 0.17 (0.39) -0.32 (0.59) -0.17 (0.34) If a donkey cart s owned (1 = yes) -0.12 (0.37) 0.05 (0.59) 0.02 (0.32) McFadden R-squared 0.27 0.53 0.18 N 149 149 149 NB: (.) = Standard errors, *, **, *** are sgnfcance levels of 0.1, 0.05, and 0.01 respectvely. show that households n Lar are more lkely to sell ther kale n Narob than households n the other two dvsons. Moreover, the results show that households wth many adult members rrespectve of ther gender are more lkely to sell kale n Narob than do households wth fewer adult members. On the contrary households wth fewer adult female members are more lkely to sell kale at farm gate. Apparently avalablty of labor for marketng has an nfluence on the decsons of where to sell kale. More tme s requred to take kale to Narob rather than farm gate or the local market, and hence households wth many adult members (proxy for labor), are the ones lkely to sell n Narob. Ths of course mples a labor market mperfecton so that suffcent labor cannot be hred n for marketng purposes. There s a sgnfcant postve relatonshp between the dependency rato (number of chldren to adults) and the choce to sale n Narob. A hgh dependency rato mples that more support s requred from the adult members, and hence they seek to sell kale where maxmum returns are antcpated. The probt results also show that households wth hgher off farm ncome and households wth a larger area under tea and/or coffee (cash crops) are less lkely to sell ther produce n Narob. Ths result mples that the relatve mportance placed on a commodty as a source of ncome may determne the choce of where to sell. Households havng alternatve sources of ncome (offfarm ncome and/or cash crops) may put less effort n seekng hgher prces for kale, than households that do not have alternatve ncome sources. As already mentoned, the farmers n Lar dvson were more lkely to sell kale n Narob and fetch hgher prces and those n Gthungur dvson were the least lkely to sell n Narob. An examnaton of the farm actvtes practced by the households n Lar and Gthungur dvsons reveal that n Gthungur, 67% of the households grow cash crops (tea and/or coffee), whereas n Lar only about 2.3% grow ether tea or coffee or both. Smlarly, usng a t-test to compare the level of off-farm ncome n one dvson wth the combned level n the other two dvsons, we found that households n Gthungur had a sgnfcantly hgher (P<0.05) level of off-farm ncome compared to those of Lar and Lmuru combned. Households n Lar on the other hand had sgnfcantly lower (P<0.01) off-farm ncome compared to those n Gthungur and Lmuru combned. The off-farm ncome of Lmuru was not sgnfcantly dfferent from that of Lar and Gthungur combned. Addtonally, over 38% of households n Lar dd not have access to any off-farm ncome, whereas only 20% n Gthungur dd not have access to off-farm ncome. The nfluence of off-farm ncome and area under a cash crop (tea or coffee) combne to show that households n Gthungur had more alternatve sources of ncome compared to those n Lar. It becomes apparent that households n Lar dvson were more n need of ncome from kale, and therefore went to the trouble of supplyng t to Narob n order to fetch hgher prces, because t s ther major source of ncome. Households n Gthungur have alternatve sources of ncome and hence
810 Afr. J. Agrc. Res. are less careful on where to sell kale. Households who were more lkely to sell ther produce at the local market were those from Gthungur, those who were younger and those wth a relatvely larger area under tea and/or coffee. Carryng produce and spendng tme at the market s a tedous job and requres the energy of younger household heads. Only two varables are sgnfcant for the choce to sell at farm gate. They are age of the household head and number of adult females both wth a negatve nfluence. As dscussed above, older household heads that presumably have less energy wll tend to sell at farm gate. The result agan confrms that avalablty of labor was an ssue n the decson of where to sell, and t appears that female labor was more mportant for ths decson. It was expected that ownershp of a bcycle and/or a car/truck wll sgnfcantly nfluence the pont where kale s sold, but apparently they do not. None the less they have the correct sgns for the dfferent market outlets. That ownershp of a car/truck was not sgnfcant for the decson to sell n Narob may be because, from the results ts ownershp was assocated wth wealthy households and as we have seen t s the less wealthy households who are more lkely to sell n Narob. DISCUSSION Prce of a commodty s essentally a result of the forces of supply and demand. That the hghest prce was receved when kale was sold n Narob s thus a reflecton of ncreasng urban demand. Indeed as Tffen (2003) observed, a larger, more productve sector enlarges the market for farmers and stmulates them to nvest n mprovement. Most people n Narob are n offfarm employment and n most cases do not produce ther own kale, hence relatvely hgher demand and hence hgher prces. Havng the lowest prces pad at the local market on the other hand reflects the narrowness of local markets (Sadoulet and de Janvry, 1995). Most households n the neghborhood produce the same crop and supply the same market and as Tffen, 2003 notes, farmers are unable to sell any surplus they produce f all near them are smlarly engaged and have no access to other centers of demand. It was the farmers who dd not have permanent cash crops and/or those wth lttle or no off-farm ncome that were more lkely to sell n Narob. If we take the presence of permanent cash crops combned wth access to off-farm ncome to be an ndcaton of wealth, then the fndngs of ths study are n lne wth the fndngs of Fafchamps and Hll (2005) who analyzed the decson to sell coffee ether at farm gate where prces are lower or travel to the market and fetch hgher prces, and found that wealthy farmers are less lkely to travel to the market. Fafchamps and Hll (2005) n ther study explaned the lower lkelhood of wealther farmers to sell on the market where prces were hgher by the fact that the shadow value of ther tme was hgh. Explaned n another way, and probably more plausble for the current study, the utlty of extra ncome from kale s hgher for the poorer households so that sellng n Narob s a survval strategy for them. For the wealther households, the opportunty cost of ther tme s hgher than the extra earnngs from hgher prces pad n Narob. Publc nvestments that am at ncreased effcent market opportuntes wll thus promote the welfare of the poorer farmers. Dstance to the market came out as an mportant determnant of prce. Dstance to the market affects the prce through the transacton costs ncurred n transferrng the produce from the farmer to the purchaser. Transacton costs are the embodment of barrers to market partcpaton by resource poor smallholders (Holloway et al., 2000). They nclude searchng costs of; barganng wth potental tradng partners, transferrng the product, montorng the agreement to see that ts condtons are fulflled, and enforcng the exchange agreement (Staal et al., 1997). Dfferental transacton costs among households thus stem from asymmetres n access to assets, nformaton, servces and remuneratve markets. Dstance from the market and poor nfrastructure ncrease the cost of transferrng the product, whch are essentally transportaton cost. The result of ths pervasve exstence of transacton costs s that, even f perfect markets exst n a partcular dstance locaton, agents have to ncur hgh costs to access these markets, creatng wde bands between the market prce and the farm gate prce. The consequence s that each decson makng unt faces a unt specfc set of effectve prces. Optmum resource allocaton wll consequently dffer for each farmer accordng to the transacton costsdetermned effectve prces that characterze t. The average quanttes sold to mddlemen were much hgher compared to those sold at the open market for any gven destnaton. Ths could mply that households are not able to handle large quanttes of kale at the open market. Kale beng pershable, t s rsky to handle large quanttes at the open market partcularly because they do not have coolng facltes to store what s not sold. The households therefore choose to transfer part of the rsk to mddlemen who possbly have better transport and coolng facltes to keep the kale fresh longer. Ths agan s n agreement wth the fndngs that smallholders are squeezed out as ndrect supplers and are left to operate n the tradtonal system, whch fetches lower prces (Neven and Reardon, 2004). Ths study dd not capture kale sells to super markets whch s ncreasngly becomng mportant because at the tme of data collecton, sells to supermarkets were not there. A follow up study to nclude supermarkets s necessary. CONCLUSION AND RECOMMENDATIONS The man determnants of the prce of kale were; the pont
Salasya and Kees Burger 811 of sell (market outlet), dstance to the market, and the locaton of producton. The prce of kale was hghest when t was sold n Narob. The decson to sell kale n Narob where prce was hghest was postvely nfluenced by avalablty of labor (both male and female) and number of dependants, and negatvely nfluenced by age of the farmer, amount of off farm ncome and area under cash crops. Wealther farmers (those wth permanent cash crops and off farm ncomes) are less lkely to sell n Narob. On the contrary poorer households wth no offfarm ncome or a permanent cash crop are more lkely to seek hgher prces n the cty. Ths mples that mprovng marketng of kale, partcularly accessblty to the cty and provdng coolng facltes could be a sure way of targetng the less wealthy households and mprovng ther welfare. There s a need for publc nvestments towards ncreased effcent market opportuntes to avod decreased commodty prces and farm ncome especally for poorer households. ACKNOWLEDGEMENTS Ths research used data generated by the PROSAM, a research project fnanced by the Ecoregonal Fund admnstered by the Internatonal Servce for Natonal Agrcultural Research (ISNAR). We are grateful for the fundng and we are also grateful to many players n the PROSAM project for ther contrbutons. REFERENCES Baley D, Peterson MC (1991). A Comparson of Prcng Structures at Vdeo and Tradtonal Cattle Auctons. Western J. Agrc. Econ. 16: 392-403. Central Bureau of Statstcs (CBS), (2001). The 1999 populaton and housng census, (1): Populaton dstrbuton by admnstratve areas and urban Centers. Central Bureau of Statstcs (CBS), Mnstry of Fnance and Plannng, Narob, Kenya. Combrs P, Lecocq S, Vsser M (1997). Estmaton of Hedonc prce equaton for Bordeaux Wne: Does qualty matter? Econ. J. 107: 390-402. Couton C, Gardes F, Thepaut Y (1996). Hedonc prces for envronmental characterstcs n the French car market. Appled Economcs Letters 3: 435-440. Neven D, Reardon T (2005). Supermarkets n Urban Kenya: General Research Approach. Polcy Bref 13/01/2005. Neven D, Reardon T (2004). The Rse of Kenyan Supermarkets and the Evoluton of ther Hortculture Product Procurement Systems. Dev. Polcy Rev. 2(6): 669-699. Fafchamps F, Hll RV, (2005). Sellng at the farm gate or travelng to the market. Ame. J. Agrc. Econ. 87(3): 717-734. Holloway G, Ncholson C, Delgado C, Staal S, Ehu S (2000). Agrondustralzaton through nsttutonal nnovaton, transacton costs, cooperatves and mlk-market development n the East-Afrcan hghlands. Agrc. l Econ. 23: 279-288. Jaetzold R, Schmdt H (1983). Farm Management Handbook of Kenya: Natural condtons and farm management nformaton. Mnstry of Agrculture, Narob, Kenya. Kmenye L (1995). Kenya s experence n promotng smallholder producton of flowers and vegetables for European markets. Afr. Rural Urban Stud. 2(2-3): 121-141. Lansford Jr NH, Freeman DW, Toplff DR, Odell L Walker (1998). Hedonc Prcng of Race-Bred Yearlng Quarter Horses Produced by Quarter Horse Sres and Dams. J. Agrc.Bus. 16: 169-185. Mnot N, Ngg M (2002). Hortcultural development n Kenya and Ivory Coast. A paper prepared for the IFPRI workshop on Successes n Afrcan Agrculture. Lusaka, June 10th 12th 2002. Republc of Kenya (RoK) (2005). The Economc Survey, 2005. Central Bureau of Statstcs (CBS), Mnstry of plannng and Natonal Development. Government prnters, Narob, Kenya. Rosen SM (1974). Hedonc prces and mplct markets: product dfferentaton n pure competton. J. Polt. Econ. 73: 537-541. Sadoulet E, Alan de Janvry (1995). Quanttatve development polcy analyss. The John Hopkns Unversty Press, Baltmore and London. Salasya BDS, (2005). Crop Producton and Sol Nutrent Management: An Economc Analyss of Households n Western and Central Kenya. PhD. Thess, Wagenngen Unversty. Staal S, Delgado C, Ncholson C (1997). Smallholder Daryng Under Transactons Costs n East Afrca. World Dev. 25: 779-794. Swernberg K (1995). Hortcultural exports from Kenya. Hortcultural Trade J. 3: 3-5. Tffen M (2003). Transton n Sub-Saharan Afrca: Agrculture, Urbanzaton and Income growth. World Development 31(8): 1443-1366. Verbeek M (2003). A gude to Modern Econometrcs. John Wley & sons Ltd, The Atrum, Southern Gate, Chchester, West Sussex PO19 8SQ, England. Weatherspoon DD, Reardon T (2005). The Rse of Supermarkets n Afrca: Implcatons for Agrfood Systems and the Rural Poor. Develop. Econ. Rev. 21(3): 333-355.
812 Afr. J. Agrc. Res. Appendx 1 Descrpton of varables ncluded n the probt model Level of educaton of the household head: Ths s the number of years spent n school. The level of the farmer s educaton s hypotheszed to be postvely related to the decson to sell kale n Narob because educated farmers are better able to access market nformaton. Age of the household head: Older household heads have less energy to transport produce to further markets and hence a negatve coeffcent s expected for the decson to sell n Narob and/or the local market and a postve one for the decson to sell at farm gate. Gender of the household head: Ths s a dummy that takes a value of 1 for female farmers and a value of zero for male farmers. Female-headed households are hypotheszed to have fewer sources of off farm ncome and hence should be more motvated to seek hgher prces n the cty. On the other hand they are less lkely to have access to nformaton on prces at dfferent markets outlets. Ths coeffcent could therefore go ether way. Number of adults: It s a proxy for avalablty of labor. Labor s requred to transport kale to the market and a postve sgn s expected for the decson to sell n Narob and a negatve sgn s for the decson to sell at farm gate. For more clarty ths varable s sub dvded nto adult females, adult males and the dependency rato chldren to adults. Occupaton of the household head: Ths s a dummy ndcatng whether a household head s a full tme farmer or not. It s hypotheszed that fulltme farmers have a hgher utlty for farm ncome and are more lkely to sell ther produce n Narob. Membershp of an organzaton: A dummy varable that equal 1 f a farmer s a member of an organzaton and zero otherwse. Members of organzatons such as farmer groups may have better access to nformaton on prces n varous markets and may organze themselves to collectvely transport produce to where the prces are hghest. A postve coeffcent s therefore expected for decson to sell n Narob. Dstance to the nearest market: Proxmty to the market often means proxmty t where publc transport can be avaled. Farmers nearer a local market are hypotheszed to be more lkely to sell n Narob and/or at the local market hence a negatve coeffcent expected. A postve coeffcent s expected for the decson to sell at farm gate. Area under a permanent crop: If a farmer has permanent cash crops such as tea and/or coffee they are lkely to have a lower utlty for cash from kale and are thus less lkely sell n Narob. A negatve coeffcent s expected for Narob and a postve one for farm gate and for the local market. Off-farm ncome: Ths s the amount of non-farm ncome a household receves n a year. Households havng access to hgher amounts of off-farm ncome are expected to have a lower utlty for farm ncome and are thus less lkely to seek hgher prces n the cty. A negatve coeffcent s expected for the decson to sell n Narob and a postve one for the other destnatons. Ownershp of a bcycle, a donkey cart, or a car: Ths are dummes where 1 = owns and zero otherwse. A household ownng any of these assets s more able to transport produce to the market. A postve coeffcent s thus, expected for the decson to sell n Narob. Farm sze: Farm sze s a proxy for wealth. Households wth larger peces of land are hypotheszed to be less lkely to sell the produce n Narob hence a negatve coeffcent s expected and a postve one for sellng at farm gate and local market.