Using a matching decomposition to examine the gender technology gap in Tanzanian agriculture

ABSTRACT

The promotion of gender equality in access to productive resources relies on multiple valid arguments. However, whether equal resource endowment would result in equal agricultural productivity is controversial. Using panel data from Tanzania, we examine the agricultural technology levels of male and female managed plots, and find that female farmers operate on a significantly lower technology level than their male counterparts. Thus, even if female farmers used the same amount of agricultural inputs and were equally efficient as male farmers, they would probably still attain lower yields. We employ a matching and decomposition approach to analyse whether the difference in technology between male and female managed plots originates from observations outside the common support or different distributions of productive resources, household characteristics, or crop choice within the common support. We find that most plots under female and male management are well comparable and fall into the common support, but different distributions of considered characteristics cannot sufficiently explain the observed differences in levels of technology among female and male farmers. Only among the few farmers that apply inorganic fertilizer, female and male farmers work on a similar technology level.

KEYWORDS:

• Gender
• technology gap
• nonparametric methods
• decomposition methods
• Tanzania

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The endowment effect was estimated based on the mean difference in endowment and β*, which is the coefficient of an Ordinary Least Squares regression of the pooled sample of all plots. Therefore, β* is derived assuming one production function for both female and male managed plots.

² There are, of course, multiple combinations of potential drivers of technology. In this study, we only present three different combinations, including those drivers often targeted in agricultural interventions and socio-demographic characteristics. Robustness checks referring to other combinations revealed similar results and are available upon request.

³ Anomalies are the difference between the mean precipitation/temperature of the corresponding months previous to the interview and the mean precipitation/temperature of that time frame between 1979 and 2013, divided by the standard deviation during these 35 years.

⁴ Within the first two panel rounds it was only asked whether seeds were purchased for the particular crop and planting season and, if so, were from an improved variety. The survey of the third panel round gathered information on the main seed variety for each crop and their costs. We only capture if non-recycled improved seeds were used in the third panel round. In general, the inclusion of the seed variety does neither considerable change the overall findings nor coefficients and corresponding significances of all other considered variables. Respective estimates are available upon request.

⁵ We thank the anonymous reviewer for suggesting this to us.

⁶ This refers to female managed plots as the reference. All results after matching male managed plots to the average of all female managed plots with the same characteristics, can be found in the appendix in Table A 1 and Table A 2. The corresponding findings simply differ in referring to the male sample as reference.

Additional information

Funding

This work was supported by the German Academic Exchange Service (Deutscher Akademischer Austauschdienst - DAAD) [57381316].