Agroforestry--perennials planted in association with annual field crops--has potential as a method for sequestering carbon while reportedly increasing agricultural yields and farmer income. However, measuring the downstream effect of agroforestry promotion on household welfare is difficult due to the long-term nature of agroforestry's payoffs and their dependence on local agro-ecological conditions. Agroforestry evaluation often requires detailed data that can only be collected by household survey, and is therefore also not always amenable to quasi-experimental methods using only secondary data analysis. This thesis demonstrates a method for using spatial matching methodology to select a sampling frame for survey data collection in order to measure the long-term household welfare impacts of an agroforestry project in western Kenya. I find spatial matching to be a cost-effective way to assemble a sample of pre-existing farmer groups to conduct an ex-post quasi-experimental impact evaluation and present balance statistics and alternate specifications to validate the methodology. After village-level matching, selected sample is then surveyed to obtain measures of household welfare including asset wealth and expenditure. The geographies targeted by the implementing NGO are used as the indicator of treatment exposure to calculate intention to treat and local average treatment effects of the agroforestry program. The agroforestry program is found to result in modest but significant gains in asset wealth and expenditure. The pre-survey spatial matching process is also shown to decrease variance in baseline indicators, improving the statistical power of the research design and indicating its potential for broader use in impact evaluation studies.
【 预 览 】
附件列表
Files
Size
Format
View
Where trees grow, assets grow: applying spatial matching to evaluate agroforestry’s household welfare impacts in Kenya
PDF
download
878987797
028-85220240
