Two essays on technical efficiency of aquaculture production in Kenya: Parametric and non-parametric methodological approaches

This thesis is composed of two essays that use different methodological techniques to determine the level of technical efficiency for sample tilapia fish farmers in Kenya. The studies also identify farm specific and socio-economic factors that are correlated with the technical efficiency. The first essay uses stochastic frontier analysis (SFA) to assess the level and determinants of technical efficiency for a sample of tilapia fish farmers in Kenya in one stage-procedure. Tilapia fish production data and other relevant information in the Aquaculture Collaborative Research Support Program (ACRSP) program survey, 2011 are analyzed by first estimating the Translog production function before settling on the Cobb Douglas production function involving a model for technical inefficiency effects. The production frontier involves four input variables, including seed, feed, labor and land. The technical inefficiency model includes seven operational and farm-specific variables, namely geography (farm location at County and Regional levels), specialized training by ACRSP, number of trainings, household income, age and education of the household head. The mean technical efficiency for the sample of tilapia fish farmers, estimated by the stochastic production frontier, is low, at 47%. Seed and Land size coefficients in the SFA are estimated to be positive and significant. The results from the technical inefficiency model show that households that have been trained by ACRSP and those with high-income levels have higher levels of technical efficiency than those that have not been trained by ACRSP and those with low incomes respectively. An aged household head dummy variable shows a negative and significant impact on technical efficiency. The implications of these results are discussed. The second essay applies output-based DEA, a nonparametric efficiency measurement technique, to derive technical efficiency measures of sample fish farmers in Kenya. The study also identified farm specific and socio-economic factors associated with technical efficiency using Analysis of Variance (ANOVA), Ordinary Least Squares (OLS) and two-limit Tobit (2LT) models in the second stage analysis. The output production structure for fish farmers is defined in terms of tilapia output. The six input categories used were: seed from government sources, seed from private sources, high quality feed, low quality feed, labor, and pond size. The mean technical efficiency score for the sample fish farms was estimated to be 0.55. Of 118 farms included in the analysis, 41 (or 35%) were technically efficient. Given that the country\u27s current aquaculture production stands at roughly 8,000MT these results imply that there is the potential to increase production by more than 6,000MT (or 44%) through improved efficiency. Spatial location and specialized training of best management practice are found to correlate with fish farmers\u27 efficiency levels. Pond size and socio-economic variables are found to have no correlation with efficiency levels. Key Words: ANOVA; Aquaculture; Cobb-Douglas; DEA; OLS; Developing Countries; Kenya; Non-parametric; Parametric; Technical efficiency; Tilapia; Tobit; Translog; SFA