Can Asset Transfer & Asset Protection Policies Alter Poverty Dynamics in Northern Kenya?

Intervention 1: REAP (Rural Entrepreneur Access Project)
REAP is a two year, graduation-style intervention that was developed and introduced in northern Kenya in 2009 by the non-governmental organization known as the BOMA Project. The REAP program aims to enable poor women graduate from poverty by helping them start sustainable businesses. The program includes financial coaching, business training and seed funds. Eligible women are selected through a rigorous process of initial wealth ranking of each household by community members, followed by household level validation facilitated by BOMA staff. Uniquely, the program facilitates the formation of three-women business groups that are the foundation of each intervention. As of mid 2017, the BOMA project had helped to launch over 5,200 businesses and 830 savings groups which have directly involved over 15,600 women.

Concretely, the graduation program targets the most vulnerable women and helps them to establish a sustainable income and savings through a two-year program of sequenced interventions:
1. Targeting of participants for the development of three-woman business groups;
2. Two years of mentoring;
3. Business skills training;
4. A seed capital jump grant of $US 200 to start the business;
5. Savings group training and a $US 100 progress grant (if the business remains in operation); and,
6. Access to credit through membership in savings groups


Intervention 2: Index Based Livestock Insurance (IBLI)
IBLI has been sold as a commercial insurance product in northern Kenya since 2010. The insurance relies on an NDVI satellite index to measure relative forage availability and make insurance payments when levels are much below normal for that region during that period. The index is calculated at the end of each rainy season and payouts are made when the index indicates forage is likley to be scarce during the following dry season. Since its inception, IBLI has expanded to include seven counties in Kenya and the Borena Zone of Ethiopia. The government of Kenya launched the Kenya Livestock Insurance program (KLIP) in 2015, which purchases IBLI insurance coverage for over 18,000 targeted pastoralists.

2018-02-01

2022-02-28

This study will measure the impact of the interventions jointly and individually on three groups of outcomes: material well-being, subjective well-being, and expenditure patterns sensitive to intra-household power dynamics.

We consider each of these groups of outcomes a ‘family’ for the purposes of hypothesis testing, and will use best practice methods to control the familywise error rate (e.g. Holm-Bonferroni or similar).

- Material Well-being -
We will be measuring the impact of the two programs on two outcomes in the ‘material well-being’ family: income and productive assets.

Household income will be measured by calculating the value of animal products and any other agricultural products (which are rare in this study) produced based on sale prices from the dataset and combining that with reported earnings from other sources.

The BOMA program is designed to directly increase ownership of productive assets by women, and by extension their households. To measure total change in productive assets, we will calculate the total value of those assets using value and price information collected in the survey. Livestock are by far the most commonly-owned assets, and for those we can use sale prices. The value of other assets is reported in the survey, and will be combined with livestock values to calculate total asset holdings.

- Subjective Well-being -
In the category of subjective well-being, we will focus on two outcomes: reported 10-item CES-D score and locus of control, which measures the extent to which respondents feel their own choices affect life outcomes.

The CES-D Score will be calculated by summing numeric values of answers to the 10 CES-D questions, as is standard practice with this score.

The internal locus of control index will be calculated using questions LOC014, LOC011, LOC003, LOC017, and LOC012 in the survey. For each question where the respondent answers ‘Agree’ or ‘Strongly Agree’ the index increases by one, and for each in which they answer ‘Disagree’ or ‘Strongly Disagree’ it decreases by one. We’ll also calculate and test for changes in the overall locus of control index, which uses answers from all questions.

- Women Empowerment -
The women empowerment index will be constructed based on the International Food Policy Research Institute’s guidelines for their Women Empowerment in Agriculture survey, though our questions differ slightly, as they were drawn on the BOMA project’s monitoring and evaluation program. Rather than averaging the numerical results, we will follow IFPRI in using a threshold based approach for each of two categories: expenditures and assets. The expenditures category will include questions d1-d5 in the survey (five total questions), while the assets category will include questions d6 onward (nine questions). For each category, we will count the share of questions to which the respondent answered that they have at least some say (specifically, they answered “Somewhat – I make some decisions or have a say in some decisions” or better). Results from each of the two categories will then be averaged to generate the empowerment index.

- Expenditure Patterns/ Nutrition -
We also expect that the interventions may alter household expenditure patterns, particularly toward school and nutrition. We will focus on effects in Household Dietary Diversity Scores (HDDS) and the share of boys/girls attending primary school.

This study takes place in Samburu County, Kenya. The sampling and treatment randomization process were developed to integrate well with existing processes used by BOMA. Specifically, BOMA uses a Participatory Rural Appraisal (PRA) for classifying households in each village into wealth categories. The poorest—usually around 50%--women are then screened for interest and potential barriers to participation. Those that pass the second screening are classified as REAP-eligible. This study samples from the REAP-eligible and the group of women that were placed in the wealth group just above the REAP-eligible threshold by their community, which we call the Vulnerable women.

Using BOMA’s rosters from the study region, half of the REAP-eligible study participants will be allocated into the control group and half will be allocated into the treatment group. This allocation will be stratified by community. Those in the REAP treatment group will then be allocated across five REAP treatment waves, which will be rolled out every 6 months over the first 24 months of the project. This approach aligns with BOMA’s standard operating procedure and will introduce variation in treatment durations and period since graduation, allowing us to use a continuous treatment approach for estimating impacts. Further, we have introduced variation in the distribution of REAP treatments across waves between communities, so that there will be between-community variation in the saturation of REAP businesses during midline, offering insight into spillover effects between businesses.

Insurance will be sold by the private insurance company Takaful Insruacne of Africa, starting in February 2018 and will available to everyone in the study area. Through stratified randomization within wealth group and community, half of study participants (those from the REAP-eligible group and the Vulnerable group) will be allocated into the IBLI-control group and 50% will be allocated into the IBLI treatment group. Within the IBLI treatment group, one third will be allocated into each of the three subsidy schemes. Insurance can be purchased during insurance windows that occur twice each year. Those in control can purchase insurance at the commercial rates. Those in the IBLI treatment arms will receive discount coupons which they can submit to an insurance agent to receive their discount. IBLI policies last for 12 months and overlapping coverage can be purchased.

This research will examine the impacts of insurance coverage and test 3 subsidy schemes for their impact on uptake. One treatment arm will receive a subsidy scheme similar to that employed by KLIP. One insurance subsidy scheme is similar to that offered by KLIP--100% subsidy for the first few animals and zero subsidy for additional animals. The two other IBLI-treated arms will face subsidy schemes that vary in response to the amount of coverage purchased.

Computer: A random number was generated in STATA for each individual in the roster. That random number was used to allocate participants into the study and across treatment arms.

For the allocation into REAP or IBLI subsidies, the unit of randomization was the individual. For the allocation into saturation scheme, the unit of ranomization was at the manyatta level. For the purpose of this registration, we will focus on our primary analysis, which focuses on the impacts of each program and relies on the non-clusterd randomization.

No

Women=1980

Women=1980

Treatment arm by wealth group:

REAP-eligible (poorest)
- Control: 280
- REAP: 380
- IBLI Subsidy: 280
- REAP + IBLI Subsidy: 380

Vlunerable (wealth group just above the poorest)
- Control: 165
- IBLI subsidy: 165

Pre-analysis+Plan+AEA+RCT+Registry+Complete.docx