The Impact of a Bundle of Agricultural and Household Products on Farmer Welfare

The intervention provides a bundle of agricultural, health, and energy products to rural households free of charge.

2019-10-25

2020-12-30

1. Net income directly generated from products included in the Direct Giving bundle during the first nine months after households receive the bundle
2. One month non-durable household consumption measured nine months (endline) after the intervention

1. Net income directly generated from products included in the Direct Giving bundle during the first nine months after households receive the bundle, aggregated as the sum of:
• Net income from maize profits
• Total value of maize harvest – expenditure on agricultural inputs
• Total chicken net income
• Total value of chickens owned at endline + total revenues from chicken + value of eggs produced – expenditure on caring for chickens
• Income and expenditure savings on energy
• Estimated value of solar lamp at endline + revenues from charging others’ phones - expenditures on home lighting and phone charging
A net income figure will be calculated for each household by summing the various net income input measurements collected during the endline survey and monthly surveys. Values for time periods in which data for a given profit input is unavailable will be linearly extrapolated.
We only consider maize in the primary outcome measurement, since based on product test data we do not expect the Direct Giving bundle to impact profits for other crops. However, we will still collect data at endline on inputs and outputs for non maize crops to check this assumption.
--
2. One month non durable household consumption measured nine months (endline) after the intervention, aggregated as the sum of:
• Value of own produced food consumed
• Expenditure on food consumed outside of home
• Expenditure on temptation goods (alcohol, gambling, etc.)
• Expenditure on miscellaneous household items (e.g., soap, DVDs, insecticides)
• Infrequent consumption (e.g., medical, fees and taxes, field rent)
Note that the consumption outcome will not include spending on light-related energy, since we expect the intervention—specifically, the solar lamp—to reduce energy spending.
Consumption impacts measured at endline will be linearly extrapolated to the month level to estimate consumption impact during the month prior to the endline survey.

1. Average monthly household consumption across the first 12 months post distribution
2. Total value of household assets after nine months
3. Minimum acceptable diet among children between the ages of six month and 23 months
4. Utilization of products included in the Direct Giving bundle
5. Expected net income over tree lifetime from the tree seedlings included in the Direct Giving bundle, modeled using data from farmers already growing trees in Singida and initial endline data from treatment group farmers

1. Average monthly household consumption across the first 12 months post distribution
• Value of own produced food consumed
• Expenditure on food consumed outside of home
• Expenditure on temptation goods
• Expenditure on miscellaneous household items (e.g., soap, DVDs, insecticides)
• Infrequent consumption (e.g., medical, fees and taxes, field rent)
• Durable goods (i.e., the asset items listed below)
--
2. Total consumption from each month (among the households that were randomly chosen to be administered a consumption survey that month) will be used as a separate observation to estimate treatment’s impact on average monthly consumption across the 12 months of data collection. We will also calculate impacts on consumption for each specific month.
• Total value of household assets after nine months, aggregated as the sum of the values of:
• Bicycles
• Motorcycles
• Cars
• Boats
• Computers
• Kerosene stoves
• Radios
• Sewing machines
• Beds
• Tractors
• Mattresses
• LLINs
• Fridges
• Cookstoves
• Water filters
• Pots
• Tables
• Sofas
• Chair
• Cupboards
• Dressers
• Clocks
• Irons
• Televisions
• Mobile phones
• Car batteries
• Farm tools
• Livestock
• Buildings
• Latrines
--
3. Minimum acceptable diet among children between the ages of six month and 23 months (measured at endline)
• Proportion of breastfed children age 6–23 months who had at least the minimum dietary diversity and the minimum meal frequency during the previous day
• Non-breastfed children age 6–23 months who received at least two milk feedings and had at least the minimum dietary diversity (not including milk feeds) and the minimum meal frequency during the previous day
--
4. Utilization of products included in the Direct Giving bundle
• Number of adult hybrid and local chickens
• Number of eggs harvested
• Percentage of children under 5 years sleeping under LLINs
• Percentage of household members sleeping under LLINs
• Crop storage using hermetic bags
• Hybrid maize stored
• Farmers used fertilizer in previous farming season
• Solar lamp usage (for light and mobile charging)
• Trees planted, harvested, and stored
• Quantity of corn/soy blend consumed by each person in the household
--
5. Expected net income over tree lifetime from the tree seedlings included in the Direct Giving bundle, modeled using data from farmers already growing trees in Singida and initial endline data from treatment group farmers

This study aims to establish causal impact through a cluster-level RCT. Randomization occurs at the village-level. The unit of analysis is the household.

This study aims to establish causal impact through a cluster-level RCT. Randomization occurs at the village level prior to the start of baseline data collection. Within each village, one contiguous cluster (kikundi) of 10 households is randomly selected for inclusion in the study, making the kikundi the unit of treatment.

Randomization is done at the village level instead of the household level for logistical ease (the Direct Giving bundle needs to be distributed to households in fewer villages) and to minimize spillover effects between evaluation arms.

This evaluation includes a baseline survey conducted in August through October of 2019 and an endline survey conducted post harvest in the same period in September 2020. In addition, each household is administered four of 12 monthly consumption and income surveys that will take place in each of the 12 months after distribution, running from December 2019 - December 2020. Clusters are randomly assigned to receive their first monthly survey in one of months 1 to 3, and will have each subsequent survey three months later.

The endline survey is paired with semi structured qualitative interviews.

Baseline and endline survey timing has been chosen to directly follow the maize harvest in Tanzania, since the survey aims to measure consumption, assets, and income generation and many of the farmers in Singida harvest maize as a primary crop. Baseline and endline are spaced one year apart in order to allow farmers to go through a full harvest cycle with their improved agricultural inputs, grow their poultry and trees, purchase assets, and realize other potential benefits of the intervention.

It is possible that there could be additional marginal impact of the intervention bundle beyond 12 months, but the RCT is designed to end in late 2020 to allow for a decision on the future of Direct Giving within a reasonable timeframe. Longer term follow up with the pilot recipients may be considered outside the scope of this evaluation.

The results of this work will help decide if the intervention should be continued in Singida and scaled up to other geographies within and outside of Tanzania.

Randomization is done in office using Stata.

Village

Yes

192 kikundis (contiguous clusters of 10 households each)

1,920 households

64 clusters in treatment; 128 clusters in control

Calculating for 80% power and a 5% significance level, we have a minimum detectable effect size (MDES), in standard deviations of the outcome, of 0.18 when comparing between the treatment and control groups for outcomes measured at endline.