Farm decision making and gender: Results from a randomized experiment in Ecuador

Abstract

Substantial resources have been devoted to mitigating the asset gender gap in developing country agriculture. Efforts have been taken to understand the role of women in decision making and in farm operations. Recommendations for best practices in eliciting information on women's roles have emphasized the importance of sex-disaggregated data collection and analysis. Collection of sex-disaggregated data is not straightforward and careful attention to context is needed. In Ecuador's highlands, chemical use in agriculture is widespread, and outreach and training programs to reduce this use are essential. These programs should target the appropriate decision makers. This paper presents results from a field experiment conducted in the Ecuador highlands where responding farm households are randomly assigned to one of three treatment groups: (i) a male respondent, (ii) a female respondent, and (iii) both adult male and female respondents (interviewed separately, but with knowledge that the other would also be interviewed). We assess whether treatment assignment affects responses to questions about decision making and responsibility for agricultural activities. Perceptions about household decision making and who is responsible for agricultural activities vary substantially by type of respondent. Men are more likely to claim sole responsibility; women are more likely to claim responsibility or that decisions are jointly made. In households where both man and woman were interviewed (separately) we found stark differences in responses about responsibilities, with men claiming sole responsibility at higher rates. Interviewing both members led to less divergence in responses, but large differences in perceptions about responsibilities remain when both are interviewed. Best interviewing practices depend on the type of information needed: for precise quantification of gender roles, complex methods may be necessary, but where qualitative information is sufficient, single-member interviews may be sufficient.