investment in human capital for men and women. Gender Differentials in Figure 1 shows the male and female distributions for Agricultural Production agricultural production. Note the uniform shift to the right (female  male) by approximately 20 percent at In 2011/2012, Ethiopiaʼs Central Statistical Agency every point on the distribution. (CSA), in collaboration with the World Bank, Figure 1: Holdersʼ Productivity Distribution, by conducted the first wave of the Ethiopia Rural Gender Socioeconomic Survey (ERSS), which collects detailed data on household welfare and income- 1 generating activity . The ERSSʼ sample includes 4,000 households that are representative of small towns and rural areas; wave 2 will be expanded to include urban areas and will run from 2013 to 2014. Agriculture is an integral part of household income in Ethiopia; nearly 80 percent of employed individuals 2 work in the agriculture sector . However, there is a gender gap in productivity that limits the sectorʼs full potential. This brief decomposes the gender differentials for agricultural production in Ethiopia, looking primarily at the explained variation of the differential. The analysis was carried out at the level of the holder, an individual within a household Relevant Characteristics by Gender responsible for a given extension of land. The results Table 1 outlines a few of the characteristics that are that follow reflect the analysis of 1,518 holders, of significantly statistically different for male and female which 84 percent are male and 16 percent are holders. On average, female holders have 1.27 3 fewer years of education, are nearly twice as likely to female . be illiterate, and inhabit households with both fewer Agricultural Production by Gender people and smaller dependency ratios, than their Agricultural production is defined here as the male counterparts. Additionally, female holders monetary value (in Birr) of self-reported production manage smaller areas of land, spend fewer hours per per hectare during the last agricultural season. There week on agricultural activities, and have less access is an overall gender difference in self-reported to non-labor agriculture inputs, such as number of agricultural productivity of 19 percent. The average oxen used to work the land. productivity for male holders is 7,509 Birr/Hectare, compared to 6,133 Birr/Hectare for female holders. Table 1: Mean Differences in Characteristics, by This could be due to differences in the size of land Gender of Holder holdings, number of fields managed, and other agriculture-related inputs; or, it could be due to   Male  Holder   Female  Holder   deeper differences reflecting different levels of Illiteracy   0.53   0.9                                                                                                                   1 Full dataset available for download from Years  of  schooling   1.77   0.51   http://go.worldbank.org/S4640KPFC0 Source: http://data.worldbank.org/country/ethiopia   2 Household  Size   5.59   3.85   3 The entire sample contains 3,558 holders. Holders were excluded from the analysis if they did not report a gender, Dependency  Ratio   0.67   0.54   st th produced below or above the 1 or 99 percentile, respectively, or Household  Food   were missing information for some covariate included in the Consumption   233.80   192.5   analysis.   th Total  Land  Managed   endowments as men (at the 10 percentile of (Hectares)   1.44   1.12   production), they would be 175 percent more Number  of  Fields   productive. Managed   13.22   11.9   Hrs/week  on   Figure 2: Gender Differential at Deciles of Agricultural agricultural  activities   22.99   14.44   Productivity Distribution and Share Explained by %  of  Plot  Rented  by   Endowments Holder   0.12   0.03   Oxen/Hectare   1.28   0.93   0.3   200%   ***  p<0.01,  **  p<0.05,  *  p<0.1   180%   0.25   160%   Endowment Effect 140%   0.2   As stated above, the gender differentials in Share  of  Diff.   Gender  Diff.   120%   agricultural production can be decomposed into 0.15   100%   explained and unexplained variation. Explained variation refers to the gender differential for various 80%   0.1   characteristics, or endowments, that could potentially 60%   explain the gap between male and female production. 40%   Thus, we aim to identify factors from Table 1 such 0.05   20%   that if the characteristic has a mean difference for males and females, we would expect a mean 0   0%   difference in agricultural production by gender as well. Gender  differen\al   Endowment  share  of  the  gender  diff.   The Oaxaca-Blinder method, the traditional technique used to quantify explained and unexplained variation   in a given differential, was employed here to When developing poverty reduction policies and determine how much of the gender differential in programs it is important to not only reduce the overall agricultural production can be attributed to the poverty rate, but to reduce the variance of endowments. From the characteristics in Table 1 productivity as well, ensuring improvements for above, the Oaxaca-Blinder decomposition finds that everyone. Successfully achieving the latter often household size, the dependency ratio, household necessitates identifying certain “pockets” of poverty consumption, number of fields managed, the that may not benefit from standard anti-poverty proportion of farming parcels that are rented, and the measures. Poor female landholders represent such number of hours dedicated to agricultural activities a pocket. The findings presented above suggest comprise the endowment effect. there are explainable factors driving this groupʼs low For the average gender differential in productivity of productivity, a fact which can be used to help shape 19 percent, we find that 6 percentage points (32 policy aimed at helping this subpopulation. percent) correspond to the endowment effect. In   other words, 32 percent of the gender differential in This brief is based on data collected by the Central agricultural production can be explained by factors Statistical Agency as part of the Living Standards such as differences in land size managed, non-labor Measurement Study – Integrated Surveys on inputs, and time spent on agricultural activities, etc. Agriculture (LSMS-ISA) project. The full dataset is The remaining 68 percent of the gender differential is available for download at CSA via essentially unexplained; this proportion of the http://www.csa.gov.et. differential stems from differences in returns to the endowment characteristics. The findings outlined in this brief are drawn from… Figure 2 demonstrates the gender differential for the Decomposition of Gender Differentials in agriculture productivity distribution at each decile of Agricultural Productivity in Ethiopia, by Arturo the distribution, as well as the share of the differential Aguilar, Eliana Carranza, Markus Goldstein, Talip that can be attributed to the endowment effect. The Kilic, and Gbemisola Oseni, The World Bank, June gender differential in production is largest from the 2013 th th 30 to 60 percentile. Note that, while there is not a significant difference in production at the 10 th   percentile, there is a relatively large and significant   endowment effect. Thus, if women had the same