Mainstreaming gender in agricultural research has remained a key priority for the CGIAR funded research and development programs. The CGIAR Gender Research Network has recently issued a ‘minimum standard guideline’ that criticizes the traditional approach of conducting gender studies by comparing male- versus female-headed households. The guideline recommends random selection of a male or a female farmer to foster a better understanding of gender gaps in agriculture. In this context, we present our experience of a gender inclusive household survey from coastal Bangladesh which included a sample of 30 percent women farmers. The objective of the study was to understand maize farmers’ preferences for index-based, savings-linked weather insurance products. Hypothetical product bundles ranging from pure insurance to pure savings containing a mix of savings and insurance were offered to the farmers using a choice experiment framework. The findings reveal a significant preference heterogeneity across the male and female farmers in terms insurance demand and the traits of the insurance contract. These findings imply that weather insurance products need to be differently designed and priced to attract women farmers. We conclude that gender inclusive research is eminent for guiding socially sustainable and equitable agricultural policy. Finally, our experience of reaching out to the sampled women farmers who lived and operated within the boundary of a conservative and patriarchal society suggests that pragmatic and adaptive field management practices are important for mainstreaming gender in agricultural research.
Original Title
Gender mainstreaming in agricultural research: Evidence from a household survey in coastal Bangladesh
Mainstreaming gender in agricultural research has remained a key priority for the CGIAR funded research and development programs. The CGIAR Gender Research Network has recently issued a ‘minimum standard guideline’ that criticizes the traditional approach of conducting gender studies by comparing male- versus female-headed households. The guideline recommends random selection of a male or a female farmer to foster a better understanding of gender gaps in agriculture. In this context, we present our experience of a gender inclusive household survey from coastal Bangladesh which included a sample of 30 percent women farmers. The objective of the study was to understand maize farmers’ preferences for index-based, savings-linked weather insurance products. Hypothetical product bundles ranging from pure insurance to pure savings containing a mix of savings and insurance were offered to the farmers using a choice experiment framework. The findings reveal a significant preference heterogeneity across the male and female farmers in terms insurance demand and the traits of the insurance contract. These findings imply that weather insurance products need to be differently designed and priced to attract women farmers. We conclude that gender inclusive research is eminent for guiding socially sustainable and equitable agricultural policy. Finally, our experience of reaching out to the sampled women farmers who lived and operated within the boundary of a conservative and patriarchal society suggests that pragmatic and adaptive field management practices are important for mainstreaming gender in agricultural research.
Mainstreaming gender in agricultural research has remained a key priority for the CGIAR funded research and development programs. The CGIAR Gender Research Network has recently issued a ‘minimum standard guideline’ that criticizes the traditional approach of conducting gender studies by comparing male- versus female-headed households. The guideline recommends random selection of a male or a female farmer to foster a better understanding of gender gaps in agriculture. In this context, we present our experience of a gender inclusive household survey from coastal Bangladesh which included a sample of 30 percent women farmers. The objective of the study was to understand maize farmers’ preferences for index-based, savings-linked weather insurance products. Hypothetical product bundles ranging from pure insurance to pure savings containing a mix of savings and insurance were offered to the farmers using a choice experiment framework. The findings reveal a significant preference heterogeneity across the male and female farmers in terms insurance demand and the traits of the insurance contract. These findings imply that weather insurance products need to be differently designed and priced to attract women farmers. We conclude that gender inclusive research is eminent for guiding socially sustainable and equitable agricultural policy. Finally, our experience of reaching out to the sampled women farmers who lived and operated within the boundary of a conservative and patriarchal society suggests that pragmatic and adaptive field management practices are important for mainstreaming gender in agricultural research.
survey in coastal Bangladesh In association with Sonia Akter (IRRI), Fahmida Khanam (IRRI), Tim Krupnik (CIMMYT), Frederick Rossi (CIMMYT) In association with Comparing male and female headed households is not gender analysis. Differences between these diverse household types cannot necessarily be attributed to the sex of the household head. Background For gender analysis, it is indispensable to interview both men and women. This does not necessarily mean interviewing twice as many people or that men and women in the same household must be interviewed. For some research questions, it may be preferable to interview one person per household and randomly choose whether it is a man or woman.
Identifying the appropriate respondents In association with Arguments against gender inclusive social survey 1. Difficult to hire female enumerators: In societies where women are not allowed to speak with men due to religious restrictions, female enumerators are required to gain access to women respondents. Given low education and empowerment, recruiting female enumerators is challenging or even impossible in some contexts. 3. Women lack information: Although women take active part in farming practices in some capacity, often they are not fully equipped with the cost and return information. 2. Men oppose women's participation: In highly patriarchal societies, men tend to undermine womens preferences and opinions and hence, either prohibit women from participating in the survey, or if they allow participation, they tend to intervene during the interview. Objectives of the study Objective 1: Understanding maize farmers preferences for index- based, savings-linked weather insurance products.
Hypotheses: Set 1 1. Demand for savings-linked insurance product is higher than pure insurance or pure savings products. 2. Demand for insurance varies across weather risks (i.e., flood, storm, hail). 3. Demand for insurance varies across risk and time preference.
Hypotheses: Set 2 1. It is harder to recruit female enumerators than male enumerators. 2. Womens participation in the survey will be lower than men. 3. Women fail to provide complete and reliable cost and return in formation.
Objective 2: Identifying the barriers and opportunities for including female respondents in mainstream agricultural surveys. Attributes Levels Insurance type (1) Pure insurance (i.e., good time payment=0) (2) Pure savings (i.e., good time payment=deposit) (3) Insurance-savings mix (i.e., good and bad time payment>0) Deposit=Premium Tk 100 (US$ 1.2) to Tk 4,000 (US$ 50) Bad time payment Tk 1,000 (US$ 12.5) to Tk 5,000 (US$ 62.5) Good time payment Tk 0 (US$ 0) to Tk 4,000 (US$ 50) Risk type
Methodology: Choice experiment Flood Wind Hail Pure insurance Pure savings Pure insurance Insurance-savings mixed Example choice questions Survey Set-up Male sample 299 Female sample 134 Male enumerator 12 Female enumerator 6 Survey team 6 Supervisor 6 Sub-district 3 District 1 Survey Coordinator Supervisor 2 Male enumerators 1 Female enumerator Training Implementation Management Data quality GPS coordinates Village level logistics Farmers list Enumerators list Sub-district level logistics Local collaborator Data collection Survey Team Groups Commuting Training Monitoring Interviews Sample Description Male (69%) Female (31%) Average age (max-min) 45 (18-85) 35 (20-70) Head of household 87% 23% Household size 6.58 5.57 Illiterate 21% 18% Electricity connection 22% 29% Cultivable land (in decimal) 103 45 Non-land asset (in Taka (US$)) 152,000 (US$ 1,948) 96,000 (US$ 1,230) Agricultural decisions are made By the respondent alone 74% 10% In consultation with the spouse 10% 86% Household expenditure decisions are made By the respondents alone 65% 8% In consultation with the spouse 17% 88%
An additional module to track decision making process
Male (%) Female (%) Was any member of the household present during the interview? 45 83 Was the spouse present? 16 43 Did the respondent consult with the spouse? Not at all 25 2 Sometimes 67 58 Almost all the time 8 40 Was other household members present? 29 41 Did the respondent consult with the spouse? Not at all 30 13 Sometimes 65 74 Almost all the time 5 13 Decision to purchase insurance Male (%) Female (%) Full-sample 62 38 Pure insurance 63 38 Pure savings 62 40 Insurance-savings mix 62 33 Spouse present 60 37 Agricultural decisions are taken by the respondent alone 63 30 Agricultural decisions are taken in consultation with the spouse 60 40 HH expenditure decisions are taken by the respondent alone 64 32 HH expenditure decisions are taken in consultation with the spouse 50 38 Latent class logit model: main-effects Explanatory variables Group 1 Group 2 Average class probability 46% 54% Pure insurance a -0.40 1.10*** Pure savings a -3.60*** 0.93*** Savings-insurance mix a -1.60*** 1.06*** Premium -0.0035*** -0.0018*** Good time payment 0.0035*** 0.00125*** Bad time payment 0.0003 0.00044*** Risk type Flood b 0.68*** 0.12 Wind b 0.61*** 0.10 Note: a Base category=None; b Base category=Hail Latent class logit model: Intra-group heterogeneity Explanatory variables Group 1 Group 2 Explanatory variables
Group 1 Group 2 Gender gap Farming characteristics and spatial variation Female*Non-status quo -0.21 -1.85*** Revenue*Non-status quo 0.010 0.034** Female*Time*Non-status quo -0.003 0.05*** Land size (high)*Non-status quo 0.0005 0.001 Female*Spouse*Non-status quo 0.25 -0.65* Land size (med)*Non-status quo 0.0015* 0.007*** Risk aversion Land size (low)*Non-status quo -0.0015 -0.0017 Risk averse*Pure insurance 0.16 -0.80*** Sub-district 2 c *Non-status quo -0.45 1.31*** Risk averse*Pure savings 0.76** -0.60* Sub-district 3 c *Non-status quo -0.96*** -1.07** Risk averse*Mix -0.16 -0.41 Model fit statistics Time preference McFadden Pseudo R-squared 0.20 Time preference*Pure insurance -0.53** -0.36 Number of groups 433 Time preference*Pure savings 0.36 -0.12 Number of observations per group 6 Time preference*Mix -0.04 0.015 Chi squared 1113.17 [p<0.0001; 61 d.f.] 1. Revisiting the arguments against gender inclusive social survey: 1. Lack of female enumerators 3. Women lack information 2. Men oppose womens participation Conclusions 2. Pragmatic and adaptive filed management practices are crucial for gender inclusive research.
3. No evidence of resentments against gender inclusive surveys was found at the local level. Next step: A follow-up survey