Whole farm technical efficiency : the case of Ethiopian smallholders: time-variant stochastic frontier model

Sammanfattning: This study uses a time-varying random-effects stochastic frontier model to estimate level of whole farm technical efficiency using 3,465 observations of sample smallholder farmers located in Tigray, Amhara, Oromia, and Southern Nation and Nationalities (SNNP) regions of Ethiopia. A baseline econometric analysis has been done by employing Corrected Ordinary Least Square (COLS) and cross-section data models prior to the panel data model analysis as a robustness check. A panel data, composed of the three waves of the Ethiopian Socioeconomic Surveys (ESS) conducted between 2011 and 2016, is used where each smallholder farmer in the sample is observed three times. The mean of estimated level of householdspecific technical efficiency is estimated to be 53 percent with individual efficiency scores ranging from 0.14 to 89 percent. This indicates that if the average smallholder farmer was to achieve the technical efficiency level of the most efficient farmer, it could be possible to realize a 40 percent (1-[0.53⁄0.89]) increment in value of output by average farmer. The study further reveals access to credit, beekeeping, crop rotation, and time-trend variables are important determinants of technical inefficiency. Thus, agricultural policies would be in a better position to achieve increased smallholder productivity by promoting these activities. Rural women empowerment and activities that minimize smallholders’ vulnerability to natural shocks play a key role to boost crop and/or livestock productivity. Regional mean technical efficiency scores are significantly different from each other. Thus, it would be important to have some sort of experience or best-practice sharing platforms among regions so that smallholders in different locations have opportunities to increase their productivity up to the level of best performing farmers. The time-trend variable and estimated level of technical efficiency found to be positively correlated. This implies that lessons from actions by development initiatives in each year need to be documented, disseminated using appropriate communication tools, and scaled up to a wider range of farming community. Further studies are recommended to investigate the major location-specific causes for such huge gaps between the most technically efficient and inefficient stallholder farmers.