Impact of Agricultural Credit on Coffee Productivity in Kenya
Description
To conduct the DEA analysis, coffee output at the farm level and seven inputs were considered. Gender and level of education were included in the model as control variables. The relevance of variables of each included in the model is hereafter expounded. B_(itc_i ) is the quantity of harvested coffee in kilograms per acre by coffee farm i in year t with c_i standing for whether smallholder farmer had credit (i=1) or not (i=0). The input variable a_i for coffee farm i in year t is given as: Labor cost (a_1 ): the total annual cost of labor in Kenya shillings used by smallholder coffee farmers (SHCF), whether participating farmers in the program (PF) or non-participating farmers in the program (NPF). The structure of labor (a_2): The proportion of family labor involved in coffee farming. Fertilizer (a_3 ): The cumulative number of 50-kilogram bags of fertilizer applied per year on the coffee farm. Farming area (a_4 ): the total acreage of coffee. Age of the coffee tree (a_5 ): taken to be equivalent to from the date of transplanting of the existing coffee bushes to-date Agrochemicals (a_6 ): the total liters of fungicides and pesticides applied on the coffee farm per year. Age of the farmer (a_7 ): This variable represents how old the farmer is in years Gender (a_8 ): is a dummy variable of the sex of the household head with 1 if male and 0 if female. Education (a_9 ): This is the highest level of formal education of the head of the household. It is coded 1, 2, 3, and 4 representing no formal education, primary, secondary, and tertiary 4 respectively. Extension visits (a_10 ): This variable represents the number of times farmers received advice on sound agronomical practices from extension workers. a_10 is measured as the total number of visits to the coffee producer during the period. Coffee variety (a_11 ): is a dummy variable with a value of 1 for the farmer who has improved coffee variety and 0 for a farmer who planted traditional varieties. Cropping system (a_12 ): is a dummy variable with a value of 1 if the farmer practices a mono-cropping system and 0 otherwise.
Files
Steps to reproduce
The data was collected from participating and non-participating farmers in the Commodity Fund credit program. The data were treated in two ways. Firstly, the data on ππ»πΆπΉπ who received credit from the Fund for coffee cultivation was considered as a treatment group. The treatment was the credit received by ππ»πΆπΉπ from ComFund and not the coffee project itself. The treatment group was in three sub-counties of Githunguri, Gatundu South and Gatundu North. Secondly, the ππ»πΆπΉπ who did not receive credit from the Fund was considered as a control group. The control group was in two sub-counties of Kiambu and Thika. To minimize potential spillovers of agricultural inputs to the control group in their sublocations, the selected sublocations whereby farmers received agricultural credit were mapped. Whereas other studies have used a 1km buffer zone (for example, Chung et al., 2018), this study used a 6km buffer zone to separate the treatment and control groupsβ sublocations. This is consistent with existing literature (Banerjee et al., 2010; Miguel and Kremer, 2004).