Saving for Health Expenditures in Kenya

Health remains a major barrier to economic development in poor rural areas. Access to effective health products, whether preventive or curative, has so far remained limited due in large part to poverty and the absence of financial markets that would enable poor households to invest in health on credit. Given such constraints, poor households should save in anticipation of future health shocks. However, substantial evidence suggests that they lack adequate savings products, and, as a result, households are quite vulnerable to health shocks. In order to afford medical expenditures, they resort to drawing down productive assets or business capital or to other costly risk-coping strategies.

Policy Issue

The benefits of investing in health are thought to be very high. For example, it has been estimated that 63 percent of under-5 mortality could be averted if households invested in preventative health products. Despite this, investment levels remain quite low in many developing countries. While many people point to credit constraints as the primary impediment, barriers to savings also appear to be a significant obstacle to investing in health. There are several major pathways through which savings may be constrained. Inter-household barriers may be relevant if social norms that necessitate that an individual provide support to friends and relatives if she is asked and has the cash on hand. Intra-household barriers may arise if members of a household have different spending preferences. Intra-personal barriers may arise if an individual’s saving and spending preferences are not constant over time. It is necessary to better understand these pathways and their relative importance so that we may develop more efficient health saving devices.

Context of the Evaluation

The researchers chose to work with a common social structure in the area: a ROSCA (Rotating Saving and Credit Association) - a group of individuals who make regular cyclical contribution to a fund, which is then given as a lump sump to a different member at each meeting. Recent studies reveal very high participation rates in these organizations; across Sub-Saharan Africa, average membership among adults ranges between 50 and 95 percent.i

Details of the Intervention

To estimate the relative importance of the different types of barriers to savings, the researchers randomly varied access to a set of saving devices specifically designed to alleviate one or more of the barriers discussed above. One hundred and thirteen ROSCAs were randomly assigned to five groups: four of the groups were given specific savings devices to use in addition to their regular weekly savings, while the fifth group served as a comparison.

In the first two treatment groups, members of the ROSCAs were given a locked metal box (with an opening in which deposits could be made) in which they could save at home. In the first group – the “Safe Box” group – members were given the key to the lock and could therefore take money from the box whenever they wanted, even to spend on non-health products. In the second group – the “Lock Box” group – members were not given the key and had to call the program officer in order to open the box. Once opened, the money in the box could only be used to buy health products.

The other two treatments were at the ROSCA level. In the third treatment group, individuals were encouraged to use their existing ROSCA to create a “Health Pot” in which members would contribute an additional amount during regular meetings earmarked for health products only. In the fourth group, individuals were encouraged to save in an individual “Health Savings Account” (HSA) that would be held at the ROSCA and earmarked for emergency health costs only (i.e. respondents were only allowed to withdraw this money if they needed it for a health emergency).

In all five groups, participants were encouraged to save for health savings goals. Thus, any effect of a savings product above and beyond the control group should be attributable to the product itself.

Results and Policy Lessons

Overall, the results indicate a significant demand for such savings products. Take-up of all four treatments was extremely high, suggesting that the primary effect of all treatments is simply the provision of a mechanism to protect money from others. 

In terms of health impacts, the researchers looked at two outcomes: (1) how much people invested in preventative health in the year following the program; and (2) whether people had enough money to deal with health emergencies. Note that the Lock Box and Health Pot were geared towards outcome (1), the Health Savings Account was geared towards outcome (2), and the Safe Box was geared to both outcomes.

Investments in Preventative Health: A year after the intervention, individuals in the Safe Box andHealth Pot groups had significantly higher levels of investments in preventative health products than those in the comparison group. Relative to comparison group individuals, the Safe Boxincreased investment by 67 percent, while the Health Pot increased investment by 128 percent. As expected, the Health Savings Account had no effect on this measure. Surprisingly, however, the Lock Box had no effect either. This lack of an effect is because the value of tying up money towards health is outweighed by the cost of completely limiting liquidity (for instance, to deal with unexpected income shocks). 

Coping with Health Shocks: Individuals in the Health Savings Account treatment were less vulnerable to unexpected emergencies. People in the Safe Box group also appeared somewhat less vulnerable, though the effects were not significant at conventional levels. As expected, there was no effect in risk coping in the two treatments groups that were not designed for emergency savings.

Prevalence of Savings Barriers: The results confirm the presence of all three types of savings barriers. First, inter-personal barriers are substantial - those who were previously giving assistance to others without receiving assistance in return benefited more than others. Second, intra-personal barriers also matter. Those whose savings preferences were not constant over time (as measured by survey questions) were not able to benefit from the Safe Box (because it was too easy for them to access the money). They also did not benefit from the Lock Box – this is because even though the savings in the box was illiquid, there wasn’t a strong incentive to actually put money into the box in the first place. However, they did benefit from the stronger commitment and social pressure to make deposits that was provided by the Health Pot. Third, there is some evidence of intra-household barriers. The effects of several of the interventions were larger (though not statistically significantly so) for married individuals. 


i Anderson, Siwan and Jean-Marie Baland. 2002. “Economics of Roscas and Intrahousehold Resource Allocation.” The Quarterly Journal of Economics 117 (3): 963-995

    Primary School Deworming in Kenya

    Hundreds of millions of children worldwide are infected with parasitic worms. These worms are detrimental to children's health, their cognitive development, their education and their futures. Chronic illness caused by worm infections reduces literacy and adult productivity. 

    Free deworming treatment substantially improved student attendance and health. The program also had significant "spillover" effects, improving health outcomes and attendance among students in neighboring primary schools. 

    Including the spillover benefits of treatment, the cost of keeping a child in school one additional day is only US$0.02, which makes deworming considerably less expensive than any alternative method of increasing primary school participation.

    Given the great success of this project, IPA is now working to Scale Up school-based deworming in partnership with Deworm the World.

    <--break->Policy Issue

    Intestinal helminths—including hookworm, roundworm, schistosomiasis and whipworm—infect more than one in four people worldwide and are particularly prevalent among school-aged children in developing countries. These intestinal worms are believed to have a negative impact on education, hindering child development as well as school attendance and reducing income later in life. These effects are especially pronounced in Africa, where nearly half of the total disease burden is due to infectious and parasitic diseases, including helminth infections. Existing randomized studies have focused primarily on the effects that these diseases have on cognitive performance, whereas outcomes of more direct interest to economists and policymakers—school attendance and enrollment, test scores, and ultimately, labor market outcomes—have yet to be thoroughly investigated.

    Context of the Evaluation

    Busia district is a poor and densely-settled farming region in western Kenya adjacent to Lake Victoria. Budalangi and Funyula divisions have some of the country’s highest helminth infection rates, in part due to the area’s proximity to Lake Victoria—schistosomiasis is easily contracted through contact with the contaminated lake water. Soil-transmitted helminths (STH), on the other hand, are transmitted through contact with or ingestion of fecal matter. This can occur, for example, if children do not have access to a latrine and instead defecate in the fields near their home or school, where they also play. One quarter of Kenyan student absenteeism is attributed to abdominal pains which likely due to intestinal helminth infections. In addition, older children may miss school to take care of siblings who are sick with helminth infections.

    Details of the Intervention

    This study evaluated the Primary School Deworming Project (PSDP), which was carried out by International Child Support in cooperation with the Busia District Ministry of Health. The program randomly divided 75 schools into three equal groups which were phased into treatment over three years.

    Within each group, a baseline parasitological survey was administered to a random sample of pupils. Schools with worm prevalence over 50% were mass treated with deworming drugs every six months. Girls of reproductive age (thirteen and older) were not supposed to be treated due to concerns about the possibility of birth defects. Nonetheless, 19% of girls thirteen and older also received medical treatment, partly due to confusion about pupil age, and partly because several Kenyan public health nurses administered drugs to some older girls, judging the benefits to outweigh the risks. In addition to medicine, treatment schools received regular public health lectures, wall charts on worm prevention, and training for one designated teacher. The lectures and teacher training provided information on worm prevention behaviors—including washing hands before meals, wearing shoes and not swimming in the lake.

    Results and Policy Lessons:

    Impact on Infection Intensity: Deworming reduced serious worm infections by half amongst children in the treatment groups. Pupils that received treatment reported being sick significantly less often, had lower rates of severe anemia, and showed substantial height gains, averaging 0.5 centimeters.

    Impact on School Attendance: Deworming increased school participation by at least 7 percentage points, which equates to a one-quarter reduction in school absenteeism. When younger children were dewormed, they attended school 15 more days per year, while older children attended approximately 10 more school days per year. The larger impact of treatment in lower grades may partially result from higher rates of infection among younger pupils.

    Treatment Spillover: The entire community and those living up to 6 kilometers away from treatment schools benefited from “spillovers” of the deworming treatment. Spillover effects occur because medical treatment reduces the transmission of infections to other community members. Reductions in infection in non-treated children resulted in an additional 3 to 4 days of schooling per year. Although data was not collected on adults, it is also likely that older community members were able to work more days as a result of spillover effects.

    No improvements in test scores were found as a result of the deworming. Additionally, evidence suggests that health education had a minimal impact on behavior, so that to the extent that the program improved health, it almost certainly did so through the effect of the medicines rather than through health education. Including the spillover benefits of treatment, the cost per additional year of school participation is US$3.27, considerably less than the cost of many alternative methods of increasing primary school participation.


    Transaction Costs, Bargaining Power, and Savings Account Use

    Transaction costs, such as those associated with opening, maintaining, and withdrawing funds may be a barrier to using formal savings accounts for those with low income. Couples in Western Kenya were offered the opportunity to open bank accounts, in the husband’s name, wife’s name or both names jointly.  In addition, a subsample accounts were randomly selected to come with a free ATM card, which lowered withdrawal fees and made the accounts more accessible through ATM machines. Results show that lowering costs via ATM cards significantly increased the use of savings bank accounts owned by men and accounts jointly owned by men and women. In contrast, savings accounts owned by women with low household bargaining power were used significantly less when ATM cards were provided.
    Policy Issue:
    Transaction costs such as account opening, maintenance, and operating fees pose significant barriers to the adoption of formal savings by the poor. Recent evidence on commitment savings, however, suggests that individuals may actually benefit from some transaction costs on savings accounts: Both internal and external constraints to saving such as time inconsistent preferences (valuing present over future consumption) and pressures to share resources with other members of the household and community may be reduced when money is locked away and costly to access. These observations raise a number of unanswered questions about the savings behavior of the unbanked: Would making formal sector accounts cheaper and more convenient substantially increase use of these accounts? Would the use of the savings account vary by account type (individual versus joint) or identity of the account owner (husband or wife)? To inform product design, this project aims to understand how the poor respond to reduced transaction costs on formal savings accounts.
    Context of the Evaluation:
    ATM cards are a common tool for reducing bank account transaction costs in the developed world as they facilitate withdrawals and, in some cases, are associated with reduced fees. This randomized evaluation assesses the impact of providing ATM cards for savings account holders in Western Kenya. While formal financial services in Kenya have traditionally been outside the reach of the poor, banks have recently begun to offer lower cost formal savings products marketed to a broader swathe of the population. This project was implemented in collaboration with Family Bank, a formal bank in Kenya that offers products suitable for lower income savers.  All study participants were offered Family Bank’s Mwananchi account, which has no recurring fees, a minimum balance of $1.25 US, and no deposit fees. Withdrawal fees are $0.78 US without an ATM card and $0.38 US with an ATM card. The account does not require the purchase of an ATM card, which costs about $3.75 US. In addition to reducing withdrawal fees, the ATM card enables account holders to make withdrawals at any time of the day.
    Description of the Intervention:
    Seven hundred forty nine low-income married couples were given the opportunity to open up to three accounts with Family Bank of Kenya: a joint account, an individual account for the husband, and an individual account for the wife. Each account was randomly assigned a temporary 6-month interest rate, which ranged from zero percent to 10 percent. Altogether, these couples opened 1,122 accounts. One quarter of the opened accounts were randomly selected to receive a free ATM card. The cost of the card was prohibitive for the vast majority of the study participants – consequently the “free ATM” treatment increased ATM card take-up by 89 percentage points.
    In addition to survey data from one-on-one baseline questionnaires administered during group sessions (which collected basic demographic information, as well as information on individual discount rates and time inconsistency, decision making power in the household, income, and current use of a variety of savings devices.), administrative data on bank account use was also collected for the first six months following account opening.  
    In 2012, a follow-up survey was conducted to (1) assess the longer run impacts of ATM cards, (2) determine whether the initial results reflect changes in total savings or substitution between different savings devices, and (3) delve more deeply into the results pertaining to bargaining power and women's account use.  Couples were revisited and spouses were interviewed separately about their savings, assets, income generating activities, and decision making in the household. This survey particularly focused on measuring total household saving and measuring individual bargaining power in the household. In addition, administrative data on Family Bank account use will provide a three year time series of account use for individuals in the study’s sample.
    Preliminary Results:
    Using administrative account use data of the six months following the intervention, results showed relatively low overall usage of formal accounts with only 27 percent of the couples who opened accounts saving in at least one of their new accounts.
    Data from the three-year follow-up shows that lowering costs via ATM cards significantly increased savings rates (by 8 percentage points) and average daily balances (by 74 percent) in bank accounts owned by men and accounts jointly owned by men and women. In contrast, accounts owned by women with low household bargaining power were used significantly less when ATM cards were provided.
    Preliminary results from the follow-up survey indicate the ATM cards had important long-run effects. For example, the ATM treatment significantly increased long-run use of accounts (by 4.1 percentage points). Ongoing research is studying impacts on overall income and asset levels.
    These results imply that lowering transaction costs to formal savings will increase access for many savers. However, the findings also suggest that transaction cost saving technologies that make account balances easier to view and access may favor individuals who have more bargaining power within the household, and that incorporating additional security features into transaction-cost reducing technologies (such as biometric scanning) may be a promising way of both reducing costs and making accounts more attractive to individuals with weaker bargaining positions in the household.
    Simone Schaner

    Girls Scholarship Program in Kenya

    Approximately 85% of primary school age children in western Kenya are enrolled, however only about one-third of students finish primary school. Dropout rates are typically higher for girls. Results suggest that the Girls Scholarship Program led to persistent test score gains in pupils from treatment schools five years after the program. Girls from treatment schools were also more likely to be enrolled in school and to have attended some secondary school at the time of the long-term follow up survey.

    Policy Issue: 

    In many education systems, those who perform well on exams covering the material of one level of education receive free or subsidized access to the next level of education. Such merit-based scholarships are attractive to the extent that they can induce greater student effort, assuming that pupils are motivated to strive for scholarship opportunities. However, the role of student motivation in improving education outcomes is relatively poorly understood. Policymakers have frequently focused their attention on increasing school inputs or improving teacher attendance, assuming that students are motivated to take advantage of these improvements. Merit-based scholarships for girls may offer an alternative to increase female education, and more educated women tend to have healthier children and higher incomes. However, the assumption that pupils are inherently motivated to pursue education, and the effect that educational opportunities can have on female learning, are relatively unexplored.

    Context of the Evaluation: 

    Approximately 85 percent of primary school age children in western Kenya are enrolled in school, but only about one-third of students finish primary school. Dropout rates are typically higher for girls; in 2001 the 6th grade dropout rate was 10 percent for girls and 7 percent for boys among students in this study’s comparison schools at baseline. Primary schools charge fees to cover their non-teacher costs, including textbooks for teachers, chalk, and classroom maintenance (approximately US$6.40 per family per year). There are also additional fees for school supplies, textbooks, uniforms, and activities such as taking exams, and these costs may deter parents from sending children, especially daughters, to school. This project was introduced in part to assist families of high-achieving girls to cover these costs.

    Details of the Intervention: 

    The Girls’ Scholarship Program (GSP) was carried out by International Child Support (ICS) Africa, in two rural Kenyan districts, Busia and Teso. Out of a set of 127 schools, 64 were randomly invited to participate in a program which gave merit-based scholarships to 6th grade girls who scored in the top 15 percent on tests administered by the Kenyan government. For the next two years, winning girls received: (1) a grant of US$6.40 to cover school fees, paid to her school; (2) a grant of US$12.80 for school supplies paid directly to her family; and (3) public recognition at a school awards assembly held for students, parents, teachers and local government officials.

    Academic achievement was captured in test scores, which are likely to be a good objective measure, and was not significantly affected by cheating. Exams in Kenya are administered by outside monitors, and district records from those monitors have no documentation of cheating. 


    Results and Policy Lessons: 

    Implementation: Poor existing attitudes towards outside intervention and an educationally disadvantaged population meant that some schools in the Teso district were resistant to the program. Particularly, stronger indigenous religious beliefs and a tradition of suspicion of outsiders caused implementation difficulties, which may have reduced program effectiveness there.

    Test Score Effects: The program raised test scores by 0.19 standard deviations for girls enrolled in schools eligible for the scholarship. These effects were strongest among students in Busia, where the program increased scores by 0.27 standard deviations. There are no effects found in Teso. Large positive test score gains were also found among Busia girls with low chances of winning the award, suggesting that there were positive externalities on learning. The average program effect for girls corresponds to an additional 0.2 grades worth of primary school learning, and these gains persisted one full year following the competition. There is also evidence of positive program externalities on the entire class; boys (who were ineligible for the awards) saw scores increase by 0.08 standard deviations on average. 

    Student Attendance: While the program impact on school participation is nearly zero among girls in the pooled Busia and Teso sample, the impact in Busia is positive at 3.2 percentage points. This corresponds to about a one-quarter reduction in school absenteeism.

    Teacher Attendance: The program had a large impact on overall teacher attendance; in the pooled Busia and Teso sample there was a 4.8 percentage point increase in overall teacher attendance, and if these attendance gains were concentrated only among 6th grade teachers then this would imply a 7.6 percentage point increase in attendance. Once again, effects were larger in Busia—the impact on overall teacher attendance there was 7.0 percentage points, roughly halving overall teacher absenteeism. Teachers could potentially be gaming the system by diverting their effort towards students eligible for the program, but there is no difference in how often girls are called on in class relative to boys in the program versus comparison schools, indicating that program school teachers probably did not substantially divert attention to girls. This finding suggests that greater teaching effort was directed to the class as a whole.

    Merit Scholarships and Inequality: The scholarship award winners did tend to come from relatively advantaged households, raising concerns about the distribution of benefits from this program. But in terms of student test score performance, the positive externalities affected all students, and were not concentrated amongst the most privileged. 

    Parental Involvement Effects: Anecdotal evidence from teacher interviews suggests greater parental monitoring occurred in Busia as a result of the program. One Busia teacher mentioned that parents began to “ask teachers to work hard so that [their daughters] can win more scholarships.” Another Busia teacher asserted that parents visited the school more frequently to check up on teachers, and to “encourage the pupils to put in more efforts.” When teachers were asked to rate local parental support for the program, 90 percent of Busia teachers claimed that parents were either “very” or “somewhat positive;” in Teso the analogous rate was only 58 percent. Thus, the greater improvements in both student and teacher attendance and performance in Busia as compared to Teso suggest that merit scholarships are most effective in the presence of local parental accountability and involvement, either formal or informal. 

    HIV/AIDS Prevention Through Relative Risk Information for Teenage Girls in Kenya

    Kenya's Ministry of Education has developed an AIDS curriculum for schools. However, this curriculum has not been effective in reducing the rates of infection and pregnancy. Information on the distribution of HIV infections by age and gender is not included in the official HIV curriculum for primary school. IPA evaluated the impact this information could have on teenager’s sexual decisions. Results found that girls exposed to the program were less likely to be pregnant in the next year. A follow-up survey is currently being conducted in order to measure longer-term impacts.

    Policy Issue: 

    The vast majority of HIV/AIDS cases occur in sub-Saharan Africa, where nearly 2 million people become infected with the virus every year. One quarter of these new HIV infections are among people under 25, and almost all are due to unprotected sex.  AIDS is incurable and no successful AIDS vaccine has been developed, so policymakers must focus on other preventative measures. Ensuring the adoption of safer sexual behavior among youth remains critical to preventing the transmission of this disease.

    Context of the Evaluation:

    Kenya has the eleventh largest HIV infected population in the world -- over 6 percent of Kenyans are infected.1  Children are seen as a “window of hope” in the fight against AIDS, because their sexual patterns are not firmly established. In an effort to prevent HIV infections in new generations, the Kenya Ministry of Education, Science, and Technology integrated HIV/AIDS education into the primary school curriculum in 2001. However, by 2003, this curriculum had not been fully implemented, likely due to teacher inexperience and discomfort with talking about this sensitive material.

    Details of the Intervention:

    In Kenya, as in most African countries, 25-year-old men are far more likely to have HIV than 16-year-old adolescent boys. This means that sexual relationships with older partners (often called “Sugar Daddies”) are particularly dangerous for adolescent girls.

    Information on the distribution of HIV infections by age and gender is not included in the official HIV curriculum for primary school, however. To test the impact this information could have on teenager’s sexual decisions, ICS conducted a “Relative Risk Information Campaign” in 71 schools randomly selected among 328 primary schools involved in another HIV intervention evaluation. A trained project officer visited each of those 71 schools and, with the authorization of the teachers, spoke to Grade 8 students for a 40-minute period. Students were shown a 10-minute educational video on “sugar daddies”. The video screening was followed by an open discussion about cross-generational sex. During the discussion, the project officer shared the results of studies conducted in Kenya and Zambia and Zimbabwe on the role of cross-generational sex in the spread of HIV. In particular, the project officer wrote on the blackboard the detailed prevalence rates of HIV, disaggregated by gender and age group, in the nearby city of Kisumu, a place familiar to the students.

    Results and Policy Lessons:

    Impact on Unsafe Cross-Generational Sex: As a result of this intervention, the incidence of childbearing was reduced by 28 percent (from 5.4 percent of girls getting pregnant within a year, to 3.9 percent). This suggests that the intervention reduced the likelihood that girls engage in unsafe sex. Specifically, the intervention seems to have reduced unsafe cross-generational sex: the rate of childbearing with men five or more years older fell by 61 percent, with no offsetting increase in childbearing with adolescent partners. 

    Cost-Effectiveness: This targeted approach cost US$98 per pregnancy averted. Researchers came up with several possible estimates of cost per HIV infection averted based off of various estimates of the ratio of the risk of HIV infection to the risk of cross-generational pregnancy; these estimates ranged from just under US$400 to almost US$2,000. These rough cost-effectiveness estimates compare favorably with other HIV prevention programs, such as treating sexually transmitted infections, voluntary HIV testing, and male circumcision.


    1CIA World Factbook, “Kenya,” accessed June 6, 2012.

    Selected Media Coverage:
    Pascaline Dupas

    Estimating the Impacts of Microfranchising on Young Women in Nairobi

    Youth unemployment is a major challenge in many low-income countries, and evidence suggests young women in urban areas are disproportionately affected. This study in Kenya evaluates the Girls Empowered by Microfranchising program, which connects unemployed participants with local business franchisors and provides mentoring and startup capital for participants to launch businesses. The study will measure the direct impacts of the microfranchising intervention on participants; compare program impacts to the effect of a cash grant program; and estimate the impact of new microfranchises on nearby businesses.

    Policy Issue:

    Youth unemployment is a major challenge in many low-income countries, and evidence suggests young women in urban areas are disproportionately affected. While many programs have attempted to increase young women’s physical and human capital, evaluations of these programs have generated mixed results. However, there is mounting evidence that multifaceted economic empowerment programs that combine job skills or vocational training with more holistic life skills education can have substantial impacts on the entrepreneurial activities of young women. Microfranchising is a recent policy innovation that falls in this category. Microfranchising programs connect unemployed participants with local franchisor businesses, providing motivated individuals with an established business model and the capital and business linkages needed to make their business model operational. In developing country settings where formal sector employment is relatively unavailable to young women, microfranchising programs may be especially valuable. This study is the first ever impact evaluation of a microfranchising program.

    Context of the Evaluation:

    This study targets young women aged 16 to 19 residing in slum areas of Nairobi. In Kenya, 55 percent of urban women aged 15 to 25 in the labor force are unemployed, as compared with 34 percent of young men in urban areas, 28 percent of young women in rural areas, and 18 percent of young men in rural areas.1

    The International Rescue Committee (IRC), the implementing partner in this study, which had implemented a microfranchising program in Sierra Leone, partnered with researchers and IPA to evaluate the impact of the Girls Empowered by Microfranchising (GEM) program in Kenya.

    Description of the Intervention:

    Researchers are conducting a randomized evaluation in Nairobi to measure the direct impact of the GEM program on a range of participant outcomes, compare program impacts to the effect of cash grants comparable in value to the microfranchising package, and estimate the effects of the GEM microfranchises on existing businesses.

    After IPA conducted an initial survey in 2013, 1,341 willing participants were randomly assigned to either the GEM program, a cash grant program, or a comparison group.

    Women assigned to receive the GEM program, implemented by the IRC in coordination with two community-based organizations, were invited to attend an orientation, followed by a 10-day business and life skills training course, and a several-day-long franchise-specific training. Those who completed the trainings received start-up capital in the form of equipment and supplies worth approximately US$200 to start up a new business. The micro-franchisees then launched their businesses with one of two relatively well-known firms in Kenya, a prepared foods franchisor and a hair salon franchisor. Mentors from the community-based organizations regularly visited participants, providing ongoing support over the first months after launching the business.

    Women assigned to the cash grant program, implemented by IPA, were invited to initial information sessions where they learned about the unconditional cash grants of 20,000 Kenyan shillings (approximately US$200).  Grants were distributed at subsequent meetings and participants were given the option of receiving the grants in cash or through mobile money transfers.

    In addition to comparing the impact the GEM program to the provision of comparably sized cash grants, researchers are measuring the direct impacts of the microfranchising intervention on participants approximately one year after launching a microfranchise; the indirect impacts of the GEM program on women whose friends participated in the program; the number of microfranchises that succeed or fail within the first year and factors associated with success; and the impacts of the newly launched microfranchises on pre-existing businesses in the target neighborhoods.

    Results and Policy Lessons:

    Results forthcoming


    [1] UNDP. Discussion Paper: Kenya's Youth Employment Challenge, January 2013. 

    Evaluating the African Health Market for Equity (AHME) Initiative in Ghana and Kenya

    Sub-Saharan Africa accounts for 24 percent of the global burden of disease. While private clinics are the first source of care for many Africans, the quality of care offered in private facilities is inconsistent and often weak, and the private healthcare sector faces a wide host of challenges. In this study, IPA-affiliated researchers from UC Berkeley and UCSF will evaluate the impact of a multi-pronged private healthcare initiative on healthcare utilization, quality of care, clinic financial outcomes, and child health outcomes in Kenya.

    Policy Issue:

    Sub-Saharan Africa accounts for 24 percent of the global burden of disease,[1] yet it only has 11 percent of the world's population. The health care systems of the countries in the regions are facing numerous challenges at once, including lack of training and organization, insufficient standards and quality monitoring, and high out-of-pocket expenditures.  While many efforts to address problems in the health care sector have focused on government clinics and hospitals, private providers are in fact the first source of care for many Africans. Though millions of people rely on private clinics, regulation and enforcement of quality care in private facilities is generally weak, and the private healthcare sector is not structured to ensure either quality or affordability.[2]These issues have impelled governments and NGOs to turn their attention to improving care in private facilities. Thus far, many programs have aimed to solve individual constraints to providing high-quality health care, but few have intervened on multiple fronts simultaneously. A multi-faceted approach has not been tested. This research will fill this gap by testing an initiative that addresses multiple health challenges in the region at once.

    Context of the Evaluation:

    Ghana, Kenya and Nigeria, the countries participating in the African Health Market for Equity (AHME) initiative, all have large populations, a high disease burden, high out-of-pocket payments for healthcare, and they are all working to expand the reach of their health insurance programs. The initiative was designed and is being implemented by Marie Stopes International, Population Services International, PharmAccess Foundation, Grameen Foundation, the International Finance Corporation, and Society for Family Health.

    AHME is a multi-faceted initiative that aims to improve both the supply and demand for private healthcare among the poor. Supply-side interventions aim to ensure high quality of care, while demand-side interventions aim to reduce barriers to accessing high-quality care. Evidence suggests that grouping private providers under a franchised brand with a social goal could improve both access to and the quality of some clinical medical services. Such “social franchising” entails creating a valued brand for goods or services, with a social goal, and extending the reach of that brand by leasing the right to use it. A social franchise model serves as the basis for the AHME initiative. In addition, the initiative includes training in basic clinical management and strategic planning for quality improvement and access to credit to implement improvement plans. AHME will also facilitate government registration for clinics. Finally, clinic personnel will use information and communication technology (ICT) to improve operational efficiency.

    Details of the Intervention:

    Researchers Paul Gertler (UC Berkeley) and Dominic Montagu (UCSF) will use a randomized evaluation to evaluate both the effectiveness and cost-effectiveness of the AHME program at improving quality of care, service utilization, access to high-quality care, and health outcomes. While the initiative is taking place in Ghana, Kenya and Nigeria, the initiative is being evaluated only in Ghana and Kenya, and the randomized evaluation is occurring in Kenya only.

    In Kenya, the researchers will randomly assign private clinics that meet criteria for selection into the AHME program to either a treatment or comparison group. Clinics in the treatment group will be invited to participate in AHME immediately, while clinics in the comparison group will be recruited into AHME after the evaluation is complete.

    Clinics participating in AHME will receive five interventions:

    1.     Social franchising: Private providers will be trained and certified to deliver standardized care under a franchised brand. The brand aims to signal to the client that the clinic offers high-quality services. Local and national marketing for the brand aim to build demand for the franchised services.

    2.     SafeCare: The program provides participating clinics with a standardized assessment of facility quality, support in developing quality improvement plans, and incentives for clinics to improve quality of care. 

    3.     Medical Credit Fund: The fund provides strategic planning support tied to performance-based financing to eligible SafeCare-participating clinics.

    4.     Demand-side financing: AHME will facilitate registration with Kenya’s National Health Insurance Fund for SafeCare-participating clinics that meet a minimum standard of quality.

    5.     Information and communications technology (ICT): Mobile phones and other technology will be utilized to enable clinic personnel to, among other things, collect data and directly reach clients.

    Data will be collected over a four-year period at both the clinic and household level to measure healthcare utilization, quality of care, clinic financial outcomes, and child health outcomes.

    The impact evaluation in Kenya will be a collaborative effort between Innovations for Poverty Action and researchers from the University of California, Berkeley, who will lead the overall evaluation, and the University of California, San Francisco, who will lead the accompanying qualitative evaluation.

    The qualitative evaluation will take place in Ghana and Kenya over the same four-year period to complement the quantitative findings. The qualitative evaluation will explore provider and client attitudes towards quality of health and options for care and will describe the AHME operation processes and their effects on the overall markets and institutional environments in which they function. Members of the research team will conduct in-depth interviews with providers participating in AHME and their clients and carry out focus group discussions in communities surrounding AHME facilities. The researchers will also conduct key informant interviews with AHME partner organizations and other key project stakeholders.

    Results and Policy Lessons:

    Results forthcoming.


    [1]UNEP. “Global Shortage of Health Workers.”

    [2]Prata, Ndola, Dominic Montagu, and Emma Jefferys. "Private sector, human resources and health franchising in Africa." Bulletin of the World Health Organization 83, no. 4 (2005): 274-279.

    Paul Gertler

    Challenges in Banking Poor in Rural Kenya

    Policy Issue: 
    Access to basic banking services in sub-Saharan Africa remains limited, and lags far behind other parts of the developing world. Such limited access could potentially have important repercussions on people’s lives. If lack of access to a formal bank account makes it more difficult for people to save, they will be unlikely to have enough saved up to cope with unexpected emergencies such as an illness in the household. When such shocks occur, rather than withdraw money or take a loan from the bank, people might have to take much costlier actions, such as cutting back on food consumption or removing their children from school. Lack of banking access may also make it difficult for people to save up large sums or obtain credit for start-up costs for a business, agricultural inputs, or even preventative health products like anti-malarial bednets. Over the past decade, there has been a significant push to understand these impacts more fully and to explore strategies to expand access. Comparatively little attention has been paid to the demand side—why people may choose to stay out of the formal banking system.
    Context of the Evaluation: 
    In western Kenya, large bank branches are located primarily in major towns, often leaving rural villages with very few options. Villages in the study sample have two options: a “village bank,” owned by share-holding villagers and affiliated with a microfinance organization, and a partial-service branch (essentially a sales and information office with an ATM) for a major commercial bank. Both banks have substantial minimum balance requirements and withdrawal fees, and the village bank also has an account opening fee. The village bank does not pay interest on deposits, and neither does the commercial bank, at least for the poor (interest is only paid if the account balance exceeds 20,000 Ksh, or about US$210).
    While both the village bank and the commercial bank offer credit products, the terms for borrowing vary quite a bit across the two institutions. The village bank requires the formation of a group of at least five people who approve the purpose and amount of each other’s loans, and who serve as mutual guarantors. To take out a loan, borrowers must purchase a share (valued at 300 Ksh each, or US$3.20) in the bank, and are then eligible to borrow up to four times the value of shares owned at an interest rate of between 1.25 and 1.5 percent per month. The commercial bank grants microloans to existing businesses or individuals who have had an account at the commercial bank or with another commercial bank for at least 3 months. Two guarantors and full collateral are required for each loan, which must be repaid within 6 months, at an interest a rate of 1.5 percent per month. 
    Details of the Intervention: 
    To better understand the demand for formal financial services, researchers conducted a randomized evaluation in two phases. In the first phase, 55 percent of the total sample of 989 households was randomly offered a voucher for a free savings account at either of the two local banks. Researchers paid the account opening fees, provided the minimum balance, and arranged for the banks to simplify the account opening procedures for study participants, but did not waive the withdrawal fees. The vouchers were delivered to people in their homes, at which time field officers explained how the bank and the account worked, and how to redeem the voucher.
    Nine months later, among those who had not received the savings intervention, half were randomly selected to receive information about local credit opportunities. Trained staff visited these individuals at home and delivered a detailed script explaining the rules and procedures for obtaining a loan from either of the two local institutions. Among those who had received the savings intervention previously, half were selected to receive the same financial information script as well as a voucher redeemable for one free share at the village bank, thereby removing one of the most significant barriers to getting a loan. 
    A background survey collected information on demographic characteristics of the household, sources of income, as well as access to financial services, knowledge and perceptions of available financial services, and saving practices more generally. Nine months after the start of the savings intervention, a survey was administered to a randomly selected half of the sample, asking respondents open-ended questions about their current savings practices, perceived barriers to saving, and perceptions of the various saving mechanisms available to them. For those who had received an account voucher but had not redeemed it, the survey also asked why they had not opened an account. The survey also included a number of questions about familiarity with and interest in local credit options.
    Results and Policy Lessons: 
    At baseline, knowledge of banking options was very limited—only 60 percent of adults knew of the bank branches in the area and almost no one knew the fee schedule for account opening or the conditions for applying for a loan.
    Savings intervention: While overall take-up of the savings account was 62 percent, only 28 percent of those who opened an account made two or more deposits in the 12 months after account opening. These results suggest that entry costs—be it the cost of acquiring information, the opening fees, or the administrative hassle—are only part of the explanation of the low banking rates observed in the sample. Qualitative surveys with respondents indicate that the most common concerns with available savings mechanisms were risk of embezzlement, unreliable services, and transaction fees.
    Credit intervention: Though the vast majority of respondents took the vouchers when offered, only 40 percent redeemed them and only 3 percent had even started the process of applying for a loan 6 months later. Evidence from qualitative surveys on barriers to borrowing suggests that the fear of losing one’s collateral if one cannot repay the loan is the primary deterrent. 
    Overall, the results suggest that simply expanding existing services is not likely to massively increase formal banking use among the majority of the poor unless quality can be ensured, fees can be made affordable, and trust issues are addressed. 
    Related Papers Citations: 
    Dupas, Pascaline, Sarah Green, Anthony Keats, and Jonathan Robinson. "Challenges in Banking the Rural Poor: Evidence from Kenya's Western Province." NBER Working Paper #17851, Cambridge, February 2012.


    Demand for Sanitation in Kenyan Urban Slums

    Sanitation is essential to health and welfare, but as many as 2.5 billion people in the developing world have no access to improved sanitation. In slums near Nairobi, Kenya, IPA-affiliated researchers from UC Berkeley and the University of Maryland are testing how subsidizing the cost of connecting to the sewer system and providing information about the health benefits of improved sanitation affects the number of landlords who connect to the sewer system.

    Policy Issue:

    Safe water and sanitation are essential to health and welfare, but as many as 2.5 billion people in the developing world have no access to improved sanitation. In urban areas, the lack of adequate sanitation disproportionately affects poor residents in informal settlements. Improved water supply and sanitation could provide a wide range of benefits, including longer life spans, reduced disease prevalence, and lower health costs.

    Many governments recognize these potential gains in public health and have begun investing in expensive sewer systems. However, the cost-effectiveness of these investments depends on the number of households that connect to the new sanitation infrastructure. There are large fixed costs to connection: utilities often charge a connection fee and households must purchase toilets and pipes. Poor households may not have cash on hand to cover these costs, and they may be unwilling to pay if they do not understand the relationship between sanitation and health. This study explores the reasons households may not invest in sanitation by evaluating how changing the price and providing information about the benefits of improved sanitation affect the demand for sewer connections.

    Context of the Evaluation:

    The Government of Kenya will spend US$427 million on water and sanitation infrastructure over the next two years as part of the World Bank-funded Water and Sanitation Service Improvement Project. The goal of this large project is to improve access to clean water and improved sanitation throughout the country.

    Koyole Soweto, an informal settlement in Nairobi, has a population of approximately 85,000 people. Slums like Koyole Soweto are very crowded: a single compound often houses multiple families in 6-10 dwellings. About 70 percent of landlords live in one of the dwellings in the compound, and they lease the other dwellings to different households. A single connection to the sewer system serves the entire compound, but it is the landlord who decides whether to invest in a sewer connection.

    Details of the Intervention:

    IPA-affiliated researchers from UC Berkeley and the University of Maryland are partnering with the Athi Water Services Board, which is responsible for introducing both piped water and sewerage services in Koyole Soweto. Installing a water connection is relatively cheap, so program implementers expect near-universal take-up. However, the cost of a sewer connection is much higher (US$250) and is a lower priority for most households, so researchers are evaluating how subsidizing the price of a sewer connection and providing information about the health benefits of proper sanitation affect the number of landlords who choose to connect to the sewer system. In addition, they are measuring if connecting to the sewer system ultimately affects housing rental prices.

    Approximately 2,200 compounds in Koyole Soweto will be randomly assigned to receive different subsidy amounts, sometimes paired with an information campaign. This will result in six groups of 366 compounds, receiving either:

    • A small subsidy for a sewer connection
    • A small subsidy, plus an information campaign emphasizing the relationship between sanitation and health
    • A medium-sized subsidy for a sewer connection
    • A medium-sized subsidy, plus the information campaign
    • A large subsidy for a sewer connectionA large subsidy, plus the information campaign

    Landlords who live in the compound may think about the decision to connect to the sewer system differently than those who live elsewhere, as resident landlords will enjoy the convenience and health benefits of new sanitation facilities. However, absentee landlords may be able to raise their tenants’ rent after installing a sewer connection. In order to determine how behavior varies between these two types of landlords, researchers will make sure each group has similar numbers of resident and absentee landlords.

    Results and Policy Lessons:

    Project ongoing, results forthcoming.

    Free Distribution or Cost-Sharing: Evidence from a Malaria Prevention Experiment in Kenya

    Bednets treated with insecticide are a proven way to deter mosquitoes and prevent deadly malaria. But how can we get more people to use these potentially lifesaving items? Some argue that those who pay for a good will value it more and use it more compared to those who receive it for free. We found no evidence that women receiving free nets were less likely to use them than those who paid a price for them. Charging for nets does however considerably reduce access, dropping by 75 percent when the price increases from zero to $0.75. Overall, our results suggest that free distribution is both more effective and more cost-effective than charging (even a subsidized price) for nets.

    Policy Issue:

    Malaria is one of the world’s foremost public health concerns, causing as many as 1 million deaths each year, the majority of which occur in sub-Saharan Africa.1 Malaria is often associated with poverty—the poor are most affected, likely because they have reduced access to medical services and information, and the lowest ability to avoid working in malaria epidemic areas. The disease can also perpetuate poverty—taking a high toll on households and healthcare systems and reducing GDP by an estimated full percentage point each year in malaria-endemic countries.2 The spread of malaria can be greatly reduced with the use of preventive strategies such as insecticide-treated bed nets (ITNs).

    There is a general consensus among academics and policymakers that provision of public health goods with positive externalities should be publicly financed. But this consensus coexists with a long-running debate on what proportion of the cost the beneficiaries of these public health programs should be bear. Standard economic analysis implies that goods (such as ITNs) that have a positive benefit (such as reduced malaria transmission) to the whole community when they are used by individuals should be provided at zero cost to the user. However, some argue that charging for health tools may increase their usage intensity, by screening out those who do not value the good, and inducing people to rationalize their purchase by using the good.

    Although cost sharing may lead to higher usage intensity than free distribution, it may also reduce program coverage by dampening demand. And if people who cannot afford the price are more likely to be sick, then, by selecting these people out, charging could significantly reduce the health benefits of the partial subsidy.

    Context of the Evaluation:

    In Kenya, malaria is responsible for one out of every four child deaths.3 It impacts economic growth and productivity, and almost 170 million working days are lost annually due to the disease.4 ITNs are used to prevent malaria infection and have been proven highly effective in reducing maternal anemia and infant mortality, both directly for users and indirectly for non-users with a large enough share of net users in their vicinity. ITNs have been shown to reduce overall child mortality by an average 20% in regions of Africa where malaria is endemic. Despite their proven efficacy, in Kenya only 5% children and 3% of pregnant women sleep under an ITN. Priced at US$5-7 per net, they are not affordable to most families, and so governments and NGOs often distribute ITNs at heavily subsidized prices.

    Details of the Intervention: 

    This program targeted ITN distribution to pregnant women who visited clinics for prenatal care.

    First stage: Sixteen health clinics were randomly selected to receive ITNs at a subsidized rate, with the discount varying between clinics from 90-100% of market price, and four comparison clinics were provided no ITN distribution program.

    Second stage: Within a given clinic, a further discount is randomly offered to women who have already chosen to buy the net. This second stage is intended to allow separate estimation of the selection and sunk cost effects of price on usage discussed above.

    Administrative records at the clinics were collected; data on the number of women enrolling for and receiving prenatal care services and the percentage of prenatal clients acquiring an ITN was recorded. Individual-level data was acquired through interviews with pregnant women. Women were asked basic background questions, whether they purchased a net, and their hemoglobin level was recorded.

    Results and Policy Lessons:

    Impact on ITN Usage Intensity: No evidence was found to suggest that cost-sharing increases ITN usage: women who paid positive subsidized prices were no more likely to use nets than those who received ITNs for free. Additionally, there is no evidence that cost-sharing puts ITNs in the hands of women who need the net most: those who pay higher prices appear no sicker than the prenatal clients in the comparison group in terms of measured anemia (an important indicator of malaria).

    Impact on ITN Demand: Cost-sharing does considerably dampen demand. ITN uptake drops by 60 percentage points when the price increases from zero to $0.60, a price still $0.15 below the price at which ITNs are currently sold to pregnant women in Kenya. These results imply that demand for ITNs is 75% lower at the cost-sharing price prevailing in Kenya at the time of the study ($0.75) than it is under a free distribution scheme.Overall, given the large benefit to the community associated with widespread usage of insecticide-treated nets, results suggest that free distribution of ITNs is both more efficient and more cost-effective than cost-sharing.

    1 WHO, "10 Facts on Malaria,"
    2 African Medical & Research Foundation, (AMRF), “Fact sheet – Malaria,”
    3 The World Bank, “News & Broadcast: World Bank Intensifies Anti-Malaria Efforts in Africa”,
    4 The World Bank, “Booster Program for Malaria Control in Africa – Kenya,”


    Understanding Technology Adoption: Fertilizer in Kenya

    Why do so few people use fertilizer even though it can considerably improve yields? This project measures the increase in yield due to fertilizer and hybrid seed use in Western Kenya. It found that fertilizer is profitable, and providing information goes part of the way towards increasing fertilizer adoption. Programs that help farmers commit when they have money to use fertilizer in the future have a very large impact on fertilizer adoption. 

    Policy Issue:

    By some estimates, approximately 1.4 billion people lived on less than $1.25 a day in 2005, and many of the poor are farmers. Identifying ways to increase agricultural incomes is crucial to alleviating poverty. Such strategies are especially important in sub-Saharan Africa, a region that has suffered decades of decline in per capita food production.

    Context of the Evaluation: 

    An estimated 50 percent of the population of Kenya’s Western Province lived below the poverty line around the time of this study, which often means they are unable to afford enough food to meet their basic calorie requirements as well as their non-food needs.The majority of Kenyan subsistence farmers grow maize as their staple crop, but many have only small amounts of land and are actually net buyers of maize, purchasing it when their own supply runs out immediately before a harvest. In such an environment, improving agricultural productivity could substantially benefit the farmers’ livelihoods. A potentially important input into increasing productivity is chemical fertilizer. Numerous agricultural trials on experimental farms suggest substantial returns to fertilizer, and improved fertilizer use has been associated with the increase in agricultural incomes during the Green Revolution in South Asia. However, only 37 percent of sampled farmers in the Busia district of Western Kenya report ever having used fertilizer. 

    The overall goal of this research program is to understand why farmers do not invest in fertilizer. This part of the project first investigates whether the returns to fertilizer are actually substantial on real-world farms in real conditions.

    Details of the Intervention:

    In collaboration with International Child Support (ICS), an NGO, researchers set out to experimentally measure the returns to fertilizer amont area farmers. Farmers were selected from lists of parents at local schools, and ICS paid for fertilizer and hybrid seeds, delivered materials, helped these farmers apply fertilizer and seeds, and assisted them with the harvest. On each farm, a comparison plot was kept directly next to treatment plots, which was farmed using traditional methods. The type of seed and amount of fertilizer applied to each plant was varied by plot (see below), but farmers were instructed to tend all plots exactly the same.



    Time of Application

    # of Plots


    ¼ tsp Calcium Ammonium Nitrate

    2 months after planting



    ½ tsp Calcium Ammonium Nitrate

    2 months after planting



    1 tsp Calcium Ammonium Nitrate

    2 months after planting



    Hybrid Seeds

    1 tsp Di-Ammonium Phosphate

    1 tsp Calcium Ammonium Nitrate

    (the “full package” recommended by the Kenyan Ministry of

    At planting

    At planting

    2 months after planting



    Results and Policy Lessons:

    Impact on Crop Yield: All fertilizer treatments led to increases in yield, though in different amounts. Interventions A, B, and C led to yield increases of 28 percent, 48 percent and 63 percent respectively, relative to comparison plots. Intervention D, the Ministry of Agriculture recommended package, led to an average 91 percent increase in yield relative to comparison plots. These increases in yield are generally consistent with the results obtained in experimental farm trials.

    Rates of Return:  Rate-of-return calculations suggest that Intervention B is highly profitable, with mean returns of 36 percent over a season and 69.5 percent annualized. Implementing Intervention B on a 0.93 acre area of maize cultivation (the average acreage under maize cultivation in this sample) would increase agricultural income net of fertilizer cost by about 1,100 Kenyan shillings (US$33 PPP) compared to traditional methods—this represents a 15 percent increase in net income and more than a month’s agricultural wages. This evidence demonstrates that fertilizer use can have substantial returns, even in the absence of any changes in other farming practices, on real-world farms. However, other levels of fertilizer use, including the official recommendations of the Ministry of Agriculture, are unprofitable for the average farmer in this sample. Thus, while fertilizer can be very profitable when used correctly, one reason why farmers may not use fertilizer is that the official recommendations are not adapted to their specific context. This also suggests that fertilizer is not necessarily easy to use correctly, and may not be profitable for many farmers who do not use the right quantity.


    1 Shahua Chen and Martin Ravallion (2008). “The Developing World Is Poorer Than We Thought, But No Less Successful in the Fight against Poverty,” World Bank Policy Research Working Paper #4703.
    2 National Coordinating Agency for Population and Development (NCAPD) [Kenya], Ministry of Health (MOH), Central Bureau of Statistics (CBS), ORC Macro. 2005. “Kenya Service Provision Assessment Survey 2004”. Nairobi, Kenya: National Coordinating Agency for Population and Development.

    Selected Media Coverage:

    Chlorine Dispensers for Safe Water in Kenya

    Policy Issue:
    Two million children die of diarrheal disease each year and contaminated water is often to blame. Treating water with chlorine could substantially reduce this toll. The most common approach to chlorination in areas without piped water infrastructure is to offer small bottles of chlorine for sale to consumers.However, chlorine use has been slow to catch on in this system. In this Kenyan study area, for example, less than 10% of households regularly use chlorine at a monthly cost of approximately US$0.30, despite several years of vigorous social marketing that has raised awareness about the product
    Details of the Intervention:
    Researchers examined free provision of dilute chlorine via a point-of-collection system, which includes a container to dispense the product placed at the water source, a local promoter to encourage the product’s use, and free provision of a supply of chlorine solution packed in bulk. This bulk supply dramatically reduces delivery costs relative to the retail approach, which requires packaging chlorine in small bottles, and relative to door-to- door distribution, which in addition significantly raises marketing costs. Hence, bulk distribution to water sources makes free provision more realistic. Additionally, this delivery method makes chlorine use very convenient. Users can treat drinking water when they collect it. The required agitation and wait time for chlorine-treated water are at least partially accomplished automatically during the walk home from the source. The source-based dilute chlorine disinfection approach to water treatment makes this act salient and public, in addition to making it cheaper and more convenient. The dispenser provides a daily visual reminder to households to treat their water at the moment when it is most salient—as water is collected—and maximizes the potential for learning, norm formation, and social network effects by making the dispenser public. Potential users can see others who use the dispenser, and they have the opportunity to ask questions; they will also know that others will see whether they use the dispenser
    Results and Policy Implications:
    Take-up of chlorine provided through dispensers dramatically exceeded take-up of chlorine for treating water for in-home use. When communities were randomly assigned to receive a promoter and a community dispenser, take-up was approximately 40% in the short run (three weeks) but climbed to more than 60% by the medium term (three to six months), representing 37- and 53-percentage point gains, respectively, compared to the communities that did not receive them.
    In contrast to the take-up levels achieved with the dispensers, clinic-based coupon redemption started higher and dropped over time. More than 40% of households that were given coupons redeemed them 8 months into the program, but this figure fell to 20% by 12 months. This finding suggests that the success of the dispenser may be due not only to the zero price but also to the reduction in the psychic cost of remembering to treat water that is achieved by source-based treatment as well as other attributes, like the visual reminders. Although take-up rates are slightly lower than those achieved in some trials, the dispenser system relies far less on outside personal contact (e.g., from repeated household visits from enumerators) than do those approaches; hence, costs are significantly lower. The chlorine dispenser is extremely cost-effective, with a comparative study finding dispensers the most effective from a range of low cost approaches to reducing diarrhea. 
    The success of the chlorine dispensers at the proof-of-concept stage described here led to a concerted effort to scale the intervention up as a sustainable program. As of April 2014 over 1.8 million people were being served by chlorine dispensers, with plans to reach four million by the end of 2014.
    The program has transitioned to Evidence Action, a new organization started with the support of IPA to scale evidence-based initiatives. More information can be found on their website here.

    Exploring Early Education Programs in Peri-urban Settings in Africa: Summary findings from Nairobi, Kenya

    Innovations for Poverty Action (IPA) performed a study of preschools in a slum of Nairobi, Kenya, in May and June 2013. The study aims to present descriptive details on the access and quality of preschools in this growing sector as part of a four-city project including similar work in South Africa, Nigeria and Ghana, launched and sponsored by the UBS Optimus Foundation.
    Note: This is not an impact evaluation, but a scoping study in four African cities designed to support future research. You can see the full Nairobi report here (PDF), the full 4-city report here (PDF), and the main page with links to the other summaries here.
    Data collection was conducted in the Mukuru slum area, a large industrial community in the south east of Nairobi, where an estimated 75% of children aged 3-6 live in an informal dwelling, generally a tin structure. With the aim of documenting the scale, cost and quality and preschool education in this area, 221 household surveys, 29 headmaster surveys and 32 classroom observations were conducted.
    Large preschool participation rates, even among the poorest
    Preschools abound in Nairobi and can be found on many streets in slum neighborhoods. Over 80% of 4 and 5 year olds in the Mukuru area are attending preschools, with no significant gender gap. Children in the poorest quintiles still have participation rates over 70%. 
    These high attendance figures are achieved despite the fact that 41% of the 3-6 year-olds in the area live in households with a daily income of less than $2.50 per capita. All school-related costs come to about KES 1,500 ($18) per month per child on average.
    Many private preschool options
    The preschool sector is largely dominated by the growing private school industry: an estimated 94% of preschool students in the study area of Mukuru are attending private preschools. Parents generally give a high priority to sending children to preschool for primary school preparation, and put a great emphasis on academic study starting as soon as at age 3. 
    We also find strong evidence that parents perceive more expensive private schools as superior to low cost private schools, and private preschools to be superior to public preschools. On average, parents estimate that attending a low cost private preschool instead of a public preschool would be associated with higher educational achievement and a 33% greater income at the age of 30.  
    The average caregiver interviewed knows of 4.9 preschools that their child could walk to, which shows the large set of options that parents have when choosing a preschool. The vast majority of private primary schools have attached preschools. Amongst the major factors caregivers consider when selecting a school are proximity, teacher quality, fee level and school test results.
    Gaps in infrastructure and services
    The observed preschool classrooms had adequate infrastructure, with basic learning materials, seating options and teacher supervision in most settings. The average preschool student in Mukuru is in a class of 27 students, with a student teacher ratio of 32:1.
    Classroom observations revealed very few cases where students had insufficient seating. An estimated 50% the preschools attended by the children in our sample have access to electricity, 87% have latrines, 66% have a playground or open space, and 65% are enclosed by a fence or a wall. Many preschools have little in the way of health or nutritional provision.
    Strong emphasis on academic instruction
    Notably, classroom observations revealed that 100% of instruction was teacher led, where the teacher provided instruction at the front of the class to students at desks. Children are taught literacy and numeracy, are given exams, and are ranked within the class from as early as age 3. Learning goals at young ages significantly outstrip those in place in Europe or America, and the teaching style of preschools mimics that of primary schools. In contrast, education experts underline the importance of developing a wide range of skills in preschool years, with equal emphasis being placed on social development, creativity, problem solving and emotional development.  
    There was no shortage of very basic learning materials, with an average of 100 exercise books per class. However, materials with additional content such as textbooks, storybooks, activity books, art materials or toys were generally limited or absent, which is in line with the strong academic emphasis of all preschools. 
    These findings indicate overall that the low cost private schools movement, particularly developed in urban Kenya, seems to also be reaching preprimary students. Most parents are aware of the value of early education, even in very low income areas, and a large majority of 3-6 year olds are attending academically-oriented preschools. There is evidence suggesting, however, that cost remains a barrier to good quality preschools, and that preschools might benefit from improved facilities and a more diverse curriculum focused on developing a broader range of skills. 
    IPA is eager to identify cost-effective programs successful at improving access and quality of preschool services with both public and private sector partners. For questions on IPA’s work in the early education sector, please contact Loïc Watine (
    Loïc Watine

    Creating a Toilet Habit, Kenya

    Despite expanding access to sanitary options such as community toilets, many individuals, especially in urban slums, continue to practice open defecation. One potential explanation is that open defecation has become an ingrained habit. Applying lessons from psychology and neuroscience, researchers are evaluating whether a combination of economic incentives and a marketing campaign can foster a new habit—using hygienic latrines instead defecating in the open—among slum dwellers in Kenya.
    Policy Issue:
    Sanitation coverage for urban dwellers in low-income countries remains low. In Sub-Saharan Africa, for example, only 42 percent of the urban population has access to sanitation. Even when individuals have access to sanitation options, such as community toilets, usage of these facilities remains low and many households continue to practice open defecation. One potential explanation for the persistence of this practice is that it has become an ingrained habit that individuals continue to engage in despite having a better alternative.
    Can interventions designed using lessons on habit formation from psychology and neuroscience encourage individuals to switch to a better habit? Private sector firms have successfully used this type of interventions to foster new habits. For example, advertising campaigns that associated triggers (a feeling of uncleanness when people run their tongue across bacteria plaque on their teeth) and rewards (a tingling sensation after brushing) with brushing regularly with mint-scented toothpaste encouraged millions to adopt it as a daily routine. However, there is little evidence on whether similar interventions would work when it comes to individuals’ hygiene choices in developing countries. 
    Context of the Evaluation:
    The study takes place in Mukuru, a slum in the Kenyan capital of Nairobi. About 80 percent of the 500,000 people living in the slum do not have adequate access to sanitation. Sanergy is a social enterprise working to address this unmet sanitation need. Sanergy builds and franchises community toilets (called Fresh Life Toilets, or FLTs) to local entrepreneurs. These entrepreneurs charge pay-per-use fees for the toilets they operate and generate additional income by converting human waste to fertilizer at a central processing facility. Each FLT serves up to 100 users per day and provides personal hygiene products such as soap and water. While Sanergy has grown its network of toilets rapidly since its launch in 2010, low demand continues to be a challenge. 
    Details of the Intervention:
    Applying lessons from psychology and neuroscience, researchers are partnering with Sanergy to evaluate whether a combination of economic incentives and a marketing campaign can foster a new habitusing hygienic latrines instead defecating in the openamong slum dwellers in Kenya.  
    Sanergy will issue discount vouchers intended to instill toilet usage as a habit, and pair these with a marketing campaign that will associate a certain characteristic of the FLT (for example, water and soap) with a feeling of cleanliness. In addition to this marketing campaign, Sanergy will also advertise to a random subset of voucher recipients a financial reward that increases with the number of vouchers redeemed.
    While all 3,000 individuals participating in the evaluation will receive vouchers, the specific features of the vouchers will vary across recipients. Sanergy will randomly assign individuals to one of four treatments groups or to a control group:
    Intensity of Voucher Discount
    Control group
    Number of Vouchers
    60 percent
    100 percent
    (free usage)
    20 percent
    High number, low discount
    High number, high discount
    Nominal discount to incentivize participants to present their vouchers upon usage
    Low number, high discount
    Low number, high discount
    Within each group, Sanergy will offer a randomly selected subset of individuals time-delimited vouchers, which will be redeemable only during a specific two-hour window during the day. The remaining individuals will be offered vouchers that can be used anytime during the day. 
    Researchers will recruit households living within a two-minute walk from a FLT and collect data on how often they use their nearby FLT during the subsidy period and in the year after the subsidies end. Researchers will also track the exact times at which individuals use the toilet.
    Results and Policy Lessons:
    Study ongoing, results forthcoming.


    Mushfiq Mobarak

    Credit, Change, and Lost Sales: The Surprising Impact of Small Change on a Firm’s Profitability in Kenya

    Highlighting the importance of carrying correct change helped firms to change their behavior and increase profits.

    Policy Issue:

    Small businesses in developing countries are thought to face numerous challenges in their efforts to expand and increase profitability. While credit and human capital constraints (i.e. lack of training) have frequently been highlighted as potential barriers, another constraint may be limited attention. Most people face constant tradeoffs between investing attention in work versus in other matters, such as homelife. The poor may face comparatively greater challenges in maintaining their homefront (because of higher rates of illness, for example), which may divert attention away from their work. It is possible to test whether this limited attention reduces productivity by focusing on one particular business decision for small firms: how much change to keep on hand to break larger bills. Not having proper change can have an impact of a firm's profit level. If a firm does not have sufficient small bills or coins to give a buyer change, the buyer may choose to buy the item elsewhere and the firm would lose the sale. Evaluation estimates suggest that the average firm in Western Kenya loses 5 to 8 percent of profits due to lost sales because of a lack of small change.

    Context of the Evaluation:

    The businesses included in this evaluation, which were randomly selected from ten market centers in Western Kenya, included barbers, tailors or other artisans, market vendors, and hardware shops. The typical business was small - only 16 percent of businesses had any salaried workers - and approximately 55 percent of firms were operated by women. Losing sales because of insufficient change was a common problem for these firms. At the baseline, over 50 percent of firms reported having lost at least one sale in the previous 7 days because they did not have sufficient change. Furthermore, firms spent over 2 hours on average looking for coins or small bills in the previous 7 days. Even firms that had not lost any sales in the past week spent over an hour and a half searching for change for customers.

    Design of the Intervention:

    To understand whether firms run out of change because they do not fully internalize the profits they are losing, the evaluation proceeded in two phases. First, a field officer visited each firm on a weekly basis to administer a short “changeout” questionnaire, which asked a number of questions about change management, including the number of times they ran out of change (i.e. the number of “changeouts”), the number of lost sales due to changeouts in the previous 7 days, the value of these sales, how much time they spent searching for change, and how often they gave or received change from nearby firms. The survey also asked about total sales and profits. Although the survey did not provide any training or information about change, or any direct "reminders," it may have served as a catalyst for firms to start altering behavior, as lost sales and profits due to poor change management became more salient. To measure this effect, the start date for the changeout questionnaire was randomized across firms. This enabled an estimation of the impact of the visits themselves, by comparing lost sales between those firms that started the survey earlier to those that started later.

    The second intervention more explicitly emphasized the costs of having insufficient change. After following firms for several weeks, researchers calculated the lost sales for each firm due to insufficient change as well as the market average. This information was then presented to a randomly selected subsample of firms.

    Results and Policy Lessons:

    Impact on frequency of changeouts: Veteran firms, meaning firms that had joined the survey early, were, on average, 6 percentage points less likely to experience a changeout in a given week than firms who we had just begun the changeout survey. Firms who were randomly selected for the information intervention were similarly 8 percentage points less likely to experience a changeout than those not selected.

    Impact on lost revenue and profits: Veteran firms, because they had fewer changeouts, also lost less income due to lost sales. Specifically, lost revenue for veteran firms decreased by 32 percent and lost profits decreased by 25 percent. Additionally, they also lost fewer sales whileaway from their shop to get change during the day. The information intervention also reduced lost revenue by 43 percent and lost profits by around 33 percent.

    Impact on behavior: Firms that had been in the survey longer seemed to bring in more change to work each morning, but the results were not statistically significant. These veteran firms also visited nearby firms for change on average 2.4 fewer times per week and shared change with other businesses on average 1.1 fewer time per week. Similarly, upon receiving the information, intervention firms began receiving change 1.6 fewer times per week and sharing one fewer time per week. Estimates indicate that overall, behavioral changes resulted in a 12 percent increase in profits.

    As the weekly surveys provided no skills training, nor any direct information, it is most plausible that they served as a reminder which made the importance of changeouts and the amount of money being lost more salient. While the information intervention provided some new information (the average behavior of other firms), the firm-specific information would have already been known to firms if they had processed the information. Thus, a likely explanation for the results is that firms were not paying attention to the lost sales to change, and the interventions reduced the cost of processing the information already available to them.

    The Role of Mobile Banking in Expanding Trade Credit and Business Development in Kenya

    Policy Issue:

    Access to finance is a critical constraint for small businesses everywhere.  Credit provided by up-stream suppliers to down-stream firms (“trade credit”) can relax the constraints on capital. Trade credit can help small businesses, like retail shops and kiosks, to purchase non-perishable goods for resale and free up resources for short- and long-term uses.  However,the provision of this type of credit may be limited by high transaction costs, up-stream liquidity constraints, and concerns over repayment.  As trade credit agreements in low-income countries usually involve small amounts, judicial systems are unlikely to enforce repayment of loans in court. Without a system to distribute small loans in an economically feasible manner and manage repayment, suppliers have little incentive to extend this service. This project evaluates a new method of extending trade credit facilitated by mobile banking and inventory management technologies and will shed light on its potential to foster small business development in a developing country context.

    Context of the Intervention:

    Working with Financial Sector Deepening (FSD), a Kenyan Trust focusing on development of financial services for the poor, researchers will evaluate a trade credit product that uses a mobile network to increase the efficiency of loan origination and repayment.  In collaboration with FSD, a large supplier of non-perishable products (the Coca Cola Bottling Company (CCBC)), and a Kenyan bank (Equity Bank), researchers will conduct a randomized evaluation of the new trade credit product.

    This technology has the potential to overcome two particular challenges.  First, by reducing the transactions costs of making repayments, new mobile technologies make it economically feasible to offer trade credit products requiring small, frequent repayments. Second, the centralized information system allows centralized monitoring of both credit and repayment histories.

    Description of the Intervention:

    CCBC uses 240 independently owned distributors to deliver its products to about 40,000 retail outlets in Kenya. These retailers typically make purchases (in cash) and take delivery of product once every few days, depending on expected demand and available cash on hand.  There is presently no pre-ordering of any sort in the supply chain, and no short-term credit. All payments are made in cash at or just prior to the time of delivery.

    CCBC will automate their supply chain, enabling every case of product to be recorded and tracked at the retailer level.  A natural next step in the automation process is to integrate financial transactions.  This project takes advantage of this advance in supply chain automation to build in a trade credit product.  In particular, the tracking system will allow real-time monitoring of both cash and mobile phone-based transactions, and hence enable more efficient administration of credit contracts. Critically, the trade credit will be provided not by the independently owned distributors, but by Equity Bank via its in-house mobile banking platform. This is the feature that makes the trade credit product viable for a larger number of retailers.

    The project will involve working with 1,200 retailer selling Coke products in and around Nairobi, Kenya. Of these, two thirds will receive the trade credit while one third will serve as a comparison group.  While all credits will be repayable to Equity Bank, the distributors of Coke products will be given explicit incentives to ensure repayment for half the retailers to whom the credit is offered.  The study will assess the commercial viability of the product, the role of distributors in administering it, and its impact on business development and employment creation.  If the intervention is profitable for lenders and borrowers, the project partners are keen to expand the credit product at a much larger scale and to other suppliers.

    Results and Policy Lessons:

    Results forthcoming.

    Savings Accounts for Rural Micro Entrepreneurs in Kenya

    Testing the impact of formal savings accounts on savings, productive investment and expenditures among small-scale entrepreneurs in rural Western Kenya.

    Policy Issue:

    Hundreds of millions of people in developing countries earn their living through small-scale businesses with very low levels of working capital. Approximately a quarter  of households living on less than US$2 per day have at least one self employed household member. Enabling small-scale entrepreneurship has long been identified as a mechanism to alleviate poverty, and substantial attention has been paid to microcredit as a means to promote entrepreneurship. However, the impact of microcredit schemes on business outcomes, especially for the very poor, is still largely unknown, and many banks which target the poor realize low or negative profits. In this context, some have argued that the focus needs to be put on savings instead of credit, since evidence suggests that individuals should be able to save their way out of credit constraints. But this strategy demands accessible opportunities for people to save securely – an uncertain prospect for the vast majority of the poor who still lack access to formal banking services of any kind.

    Context of the Evaluation:

    In Kenya, small enterprises have been estimated to account for more than 20 percent of adult employment and 12-14 percent of national GDP, but only 2.2 percent of surveyed microentrepreneurs had a savings account with a commercial bank prior to the study. Some individuals have demonstrated a willingness to pay a premium to save securely, often receiving negative interest or tying their funds up in illiquid savings and credit associations. The fact that people take up these costly strategies suggests that the private returns to holding cash at home are even lower, possibly due to the risk of theft, appropriation by one’s spouse or other relatives, or because individuals tend to over-consume cash on hand.  In the village of Bumala, a market center along the main highway connecting Kenya to Uganda, a community-owned bank sought to increase access to formal banking by offering savings accounts to villagers. Still, two years after opening, only 0.5 percent of daily income earners had opened an account, citing lack of information about the bank and the inability to pay the account opening fee as primary reasons for low take-up.  

    Description of Intervention:

    Working in collaboration with the Bumala village bank, researchers studied the importance of savings constraints for self-employed individuals in rural Kenya. Field workers identified market vendors, bicycle taxi drivers, and self-employed artisans who did not already have a savings account, but were interested in opening one. Of the eligible individuals, 163 were randomly selected to be offered the option to open a savings account at no cost, with a minimum balance that could not be withdrawn. These accounts offered no interest and included substantial withdrawal fees. Thus, the de facto interest rate on deposits was negative. A comparison group of 156 individuals was not barred from opening an account but was offered no assistance in doing so.

    To test the prevalence and impact of savings constraints, researchers examined 279 self-reported daily logbooks kept by individuals in both the treatment and comparison groups. These logbooks included detailed information on market investments, expenditures and health shocks, making it possible to examine the impact of the savings accounts along a variety of dimensions. Field workers met with respondents twice per week to verify the logbooks were being filled out correctly, and paid respondents a small amount for each week the logbook was completed correctly. This information was supplemented with administrative data on savings from the bank itself.

    Results and Policy Lessons:

    Impact on Savings Account Take-up: Eight percent of respondents refused to even open an account, while another 39 percent opened an account but never made a deposit. Of those who did utilize the savings accounts, women made significantly larger deposits, a median of 100 Ksh,(US$1.42, equivalent to 1.6 times average daily expenditure) compared to the median deposits for men of 50 Ksh ($0.71).  This gender difference increased for those who deposited more.  Account usage was very strongly correlated with wealth, suggesting that the accounts were mostly useful for people above subsistence levels.

    Impact on Savings Behavior: Reported average bank savings were higher in the treatment group. Females in the treatment group did not decrease other forms of savings in animal stock and ROSCAs (informal groups that require members to make regular contributions to a savings pot that is periodically given to one member).  There are various possible explanations for the continued use of ROSCAs by women. It is possible that ROSCAs are valuable as a source of credit and emergency insurance; that they provide a form of savings commitment through social pressure; or that changes in ROSCA participation could not be captured during the study due to the long  savings cycles (up to 18 months).

    Impact on Business Investment: Four to six months after they were offered, bank accounts had substantial positive impacts on business investment for women, with a 37.5 percent increase in average daily investment. This suggests women faced large negative returns on money they saved informally, and those constraints were important for the businesses they run.  While very large on average, this treatment effect is also quite heterogeneous: only 57 percent of women in the treatment group made at least one deposit within the first 6 months of opening the account, and only 43 percent made at least two deposits within that timeframe.

    Impact on Private Expenditures: Findings suggest that higher business investment in the treatment group led to higher profits, as measured through household expenditures.  The accounts had a significant positive impact on expenditures on the entire sample, with this effect most strongly concentrated for market women. About 6 months after having gained access to the account, the daily private expenditures of women in the treatment group were on average 37 percent higher than those in the comparison group. Daily expenditure on food was also significantly higher.


    Strategic Household Savings in Kenya

    How important are differences of opinion within the household for making financial decisions? In this study, married couples in rural Kenya were given the opportunity to open joint and individual bank accounts at randomly assigned interest rates. Researchers assessed if couples with different preferences worked together to save in the highest return account, or if these differences led to poor financial choices. Results indicated when savings preferences in the household diverged, individuals were more likely to prefer individual accounts, and made less efficient financial decisions. 

    Policy Issue:

    Despite their low incomes, individuals in developing countries save using a wide variety of informal savings devices like illiquid rotating savings and credit associations. Researchers have widely noted the popularity of these informal devices and an attendant puzzle: these devices are often risky, complex, and costly when compared to simple alternatives, such as storing savings at home. What then, makes these costly savings practices attractive? Anecdotally, many informal savers cite the need to protect savings from misappropriation by other members of the household, particularly spouses. How important is this need in determining individual savings choices? Does it become more important as individual preferences for how much to save diverge? And how much are individuals willing to sacrifice to gain additional control over household savings levels?


    While formal financial services in Kenya have traditionally been outside the reach of the poor, banks have recently begun to offer lower cost formal savings products marketed to a broader swathe of the population. This project was implemented in collaboration with Family Bank, a formal bank in Kenya that offers products suitable for lower income savers. Family Bank offers savers the option of both individual accounts, which can only be accessed by the account owner, and joint accounts.  When spouses jointly own an account, either member can make deposits and withdrawals at will.

    Description of Intervention:

    All married couples participating in the intervention were given the opportunity to open up to three accounts with Family Bank: an individual account for the husband, an individual account for the wife, and a joint account. Each account was randomly assigned a temporary 6-month interest rate, which ranged from zero percent (the norm for Family Bank accounts) to 10 percent. The interest rate intervention consequently created random variation in not just the absolute rate of return available to a couple, but also the relative rates of return between the three different accounts. This offered a simple way to measure efficient savings behavior: an efficient couple should always choose to save in the account with the highest rate of return.

    In addition to the interest rate intervention, half of couples who opened at least one individual account were randomly selected for an information sharing treatment. The goal of this treatment was to test whether individual accounts were used to hide information from spouses. This treatment enabled the spouse of an individual account holder to retrieve information on the balance of the individual account at the bank (provided both the spouse and the account holder consented to the information sharing treatment).

    Finally, all couples in the intervention were asked a series of questions at baseline to measure levels of patience and preferences over savings levels. These questions were used to construct a measure of preference heterogeneity in the household. Individuals were also asked about their own and their spouse’s use of a variety of savings devices – this information was used to construct a measure of how well informed spouses were about one another’s finances.


    Responses to Experimental Interest Rates

    All couples responded robustly to the experimental interest rates. However, those couples with badly aligned savings preferences (the “poorly matched”) were more than twice as likely to save in individual accounts and 34 percent less likely to save in joint accounts. Furthermore, poorly matched couples were insensitive to relative rates of return between the three different accounts on offer. In contrast, those couples who were well matched on savings preferences responded robustly to relative rates of return. Consequently, poorly matched couples sacrificed at least 52 percent more potential interest rate earnings when compared to their well-matched peers.

    Responses to the Information Sharing Intervention

    Over 40 percent of couples selected for the information sharing intervention did not consent to it, which suggests that individual accounts are used to hide resources from spouses. Furthermore, those couples who were poorly informed about each other’s finances at baseline were significantly less likely to consent. These poorly informed couples were also significantly more likely to save in individual accounts and marginally less likely to save in joint accounts. However, measures of intrahousehold information sharing and preference heterogeneity were uncorrelated with one another.

    Overall, these results suggest that when savings preferences in the household diverge, individuals strategically exploit secure savings devices to control the overall level of household savings. However, even when preferences over how much to save are well aligned, individuals may still value secure accounts if these accounts allow them to hide savings from others.

    To learn more about this Simone Schaner's research, click here

    Simone Schaner

    The Return to Capital for Small Retailers in Kenya

    Throughout the developing world, the family owned business is the most common form of enterprise. Though these types of businesses are prevalent, there is tremendous heterogeneity in the success of such firms. For instance, in the retail sector, some firms hold large inventories and earn significant profits, while others hold minimal stocks and provide little more than subsistence income for their owners. Given the importance of small enterprises in poor countries, it is important to understand why some firms are more successful than others and to identify potential ways to address the constraints that keep some firms from becoming more profitable.

    This project takes advantage of the characteristics of the retail industry to explicitly estimate the rate of return to a marginal increase in inventory for a set of small retail firms in rural Kenya. The empirical strategy is based on the fact that retailers should set inventories such that the marginal benefit from holding the last unit (the expected profit) is exactly equal to its marginal cost (the opportunity cost of holding capital). We estimate the marginal rate of return in 2 ways. First, we measure sales lost due to insufficient inventories. Second, we use an administrative dataset to observe whether firms buy enough inventory to qualify for quantity discounts.


    Preliminary results suggest that returns to inventory capital in the Kenyan retail sector are likely far greater than returns to investment in developed country equity markets and suggest that these returns likely differ significantly across firms. Implementation of other surveys and interventions is ongoing.

    Powering Small Retailers: the Adoption of Solar Energy under Different Pricing Schemes in Kenya

    The majority of people living in Sub-Saharan Africa do not have access to electricity. Traditional power companies often find it too costly to bring electricity to rural and suburban areas, but in recent years, the cost of alternative energy sources like solar power has fallen dramatically. Providing small businesses with access to reliable electricity through off-grid solar power systems could potentially help small retailers earn more by keeping their businesses open longer and introducing new services. This randomized evaluation tests how price and payment method affect the adoption of off-grid solar power among small retailers near Nairobi and if access to electricity can improve their businesses’ performance.

    Policy Issues: 
    Nearly 70 percent of people living in Sub-Saharan Africa lack access to electricity. Most traditional power companies find it too costly to extend the electric grid to many rural and suburban areas. Without access to power, households and small businesses typically use kerosene-powered lanterns or candles to provide light at night. Access to electricity could potentially help small retail businesses earn more revenue by extending their hours of operation or offering other services to customers, such as mobile phone charging facilities. 
    Many have proposed solar power as a way to bring safe and reliable electricity to small businesses and households that cannot access the electric grid. Yet for small retail businesses the cost of off-grid solar power systems may still be prohibitive. How do price and different payment methods affect the adoption and use of off-grid solar power systems among small retailers and how does access to electricity affect their business performance? 
    The small businesses participating in this study typically sell food, drink, clothing, or other household goods. Access to electricity could potentially allow them to increase their evening operating hours, offer mobile phone charging services in their stores, and save on their own mobile phone charging costs. 
    Angaza Design is a company that markets off-grid solar power systems to consumers and businesses in East Africa. Their main product is an LED light unit with integrated mobile phone charging and a detachable 3-watt solar panel to charge the unit’s battery. While the total cost of this solar power system is often too high for small retailers to purchase all at once, Angaza Design allows people to purchase the solar unit for a small down payment and then use a mobile money platform to pay for energy output in affordable increments by “topping up” device credit, just like they currently purchase mobile phone airtime. The device can be disconnected if payments are not made. Regular payments are applied towards paying off the full cost of the device, after which it is automatically “unlocked” and can be used without purchasing additional device credits.
    Description of Intervention: 
    Researchers are conducting a randomized evaluation in partnership with Angaza Design and SunnyMoney to estimate the impact of different pricing schemes, payment schedules, and enforcement methods on the adoption of off-grid solar power and the impact of access to electricity on small retail businesses’ revenue and profits. From a sample of 1,849 small retail businesses operating in the outskirts of Nairobi, researchers randomly assigned some businesses to receive one of four different offers to purchase the Angaza Design solar power system and some businesses to serve as the comparison group. Those offered the solar power system received marketing visits in which the salesperson read a script describing the features of the solar power system, the payment process, and penalties for late payments. The salesperson then gave the customer a voucher needed to purchase the solar power system from a sales agent, under one of four different payment schemes: 
    • Offer 1 provides the customer with a pay-as-you-go solar power device at 15 Kenyan shillings (KSH) per hour (about US$0.17) of electricity used. Customers are sent one text message per week to remind them to purchase more solar power time. 
    • Offer 2 instead allows the customer to make weekly payments of 130 KSH (about US$1.70) for unlimited use of the solar power system. The customers are sent a reminder to pay the day before their next payment is due. If a payment is missed, the solar power system automatically shuts off until the payment and a 50 KSH (US$0.56) penalty are paid.  
    • Offer 3 is identical to Offer 2 except that while customers are told about the 50 KSH penalties for non-payment during the initial marketing visit, after they receive the solar power system they are told it will not be applied. If customers fail to make a payment, the solar power system will not work until the retailer pays the weekly installment, but no penalty is charged. 
    • Offer 4 is identical to Offer 2 except that after customers receive the solar power system, they are told that neither the late payment penalty nor the shutoff will be applied if they fail to pay their weekly bill. Under this condition, there are no penalties for failing to pay and the device continues to work.
    Results forthcoming. 

    Encouraging Adoption of Rainwater Harvesting Tanks Through Collateralized Loans in Kenya

    Policy Issue:
    Over the last decade, billions of dollars have been invested in the development of new agricultural technologies, which have the potential to contribute to economic growth and poverty alleviation amongst the poor. However, in Sub-Saharan Africa, the adoption of these promising agriculture technologies has been limited. Low adoption rates may be explained in part by credit constraints, which prevent many smallholder farmers from making the often large, initial investment in a new technology. A sizeable initial deposit or a guarantor is often required to receive a loan, restricting access for most. In developed countries, guarantor requirements are uncommon; instead, loans for purchase of houses, vehicles, and small business equipment often use the assets themselves as loan collateral. Using assets as collateral is rare in developing countries, where credible and efficient institutions and processes to seize collateral may be lacking. But in certain environments or for certain products—like water tanks, which are large and difficult to hide or relocate, making repossession feasible—can asset collateralization expand access to credit without affecting repayment rates?
    Context of the Evaluation: 
    Almost 80 percent of the Kenyan population lives in rural areas, where many suffer from significant water constraints. About 96 percent of cropland in Kenya is rain-fed and the rains are both seasonal and unpredictable. In recent years, much of rural Kenya has been hit by extreme droughts, which can be particularly problematic for dairy cattle, which require a consistent and regular water supply. Dairy farming is common in Kenya’s Central and Rift Valley provinces, the sample area for this study. The smallholder dairy farmers in the study sample had an average herd size of two cows, but 32 percent of households reported loosing at least one cow to drought in the year prior to the study, and 52 percent of households had a cow get sick. Without easy access to water, sample farmers reported spending a considerable amount of time (ten hours per week) taking cows to a water source. Technologies, such as rainwater harvesting tanks, that help to give dairy farmers reliable and convenient access to water, could improve their productivity and other economic outcomes.
    Details of the Intervention: 
    Researchers used a randomized evaluation to examine the effect of asset collateralized loans on the take-up of rainwater harvesting tanks and any subsequent impacts on dairy production, time use, or girls’ enrollment in school.
    In partnership with a local dairy cooperative, researchers offered smallholder dairy farmers in central Kenya a loan to purchase a durable, 5,000-liter rainwater harvesting tank. The market price of the tanks at the time of the study was Ksh 24,000 (US$320), or roughly 13 percent of annual total household consumption in the sample. A full tank would last between 17 days and 6 weeks depending on whether the water was used for livestock, domestic use, or both.
    From a sample of 3,000 farmers who sold milk to a local dairy cooperative, researchers randomly assigned 1,804 to receive one of four initial loan offers. The value and terms of the loan were the same for all four offers: the loan could only be used to purchase a water tank, the value of the loan was set equal to the market price of the tank (US$320), and all borrowers had to make a minimum deposit of Ksh 1,000 (US$15).
    The four different loan offers varied in the deposit and guarantor requirements, and whether the asset was pledged as collateral. 
    (i) 100 percent secured joint-liability loan - required borrowers to make a deposit equivalent to one-third of the value of the loan in their savings account. Three guarantors were also required to insure two-thirds of the loan amount through savings or shares in the dairy cooperative. This offer was essentially the same as the standard contract used by the cooperative when it gave out loans.
    (ii) 4 percent deposit loan - used the water tank asset as collateral but carried no other up-front requirement apart from the small, Ksh 1000 (US$15) deposit, equivalent to 4 percent of the loan amount, required of all borrowers.
    (iii) 25 percent deposit loan – used the water tank asset as collateral and required a larger deposit of Ksh 6,000, equivalent to 25 percent of the loan amount.
    (iv) 25 percent guarantor loan – used the water tank asset as collateral and also required a single guarantor to pledge 21 percent of the loan amount (Ksh 5,000), in addition to the Ksh 1,000 deposit from the borrowing farmer.
    Researchers anticipated that take-up rates would vary across these groups due to differences in the perceived profitability of borrowing under the given terms, differences in access to guarantors, and differences in farmer productivity and characteristics. In order to distinguish between these factors, farmers in groups iii and iv were randomly divided into two subgroups. Half of the farmers who received offer iii or iv had the requirements waived after the loan contract was signed; in group iii, the deposit amount in excess of the minimal 4 percent deposit was returned, and in group iv, the guarantor was absolved of any responsibility for the loan and his deposit was returned.
    For all six resulting treatment groups, the timeline for the loan was 24 months and monthly repayments consisted of a fixed amount (Ksh 1,000) plus interest (1 percent per month) on the declining principal balance.  
    The default procedures were the same across all treatment groups. When a farmer fell two months behind, he received a letter warning him that he was pending default and had two months to pay off the late amount. If the payment was still outstanding after 60 days, the cooperative applied any deposits held to the balance. If a balance still remained, the farmer was given an additional 15 days to clear it, after which the cooperative would repossess the tank and sell it to cover the farmer’s loan obligations. 
    Results and Policy Lessons: 
    Take-up rates: Take up of the 100 percent secured joint-liability loan was low, with only 2.4 percent accepting the offer. Asset collateralization significantly increased take up—nearly 28 percent of those who received the 25 percent deposit loan offer accepted, 23.5 percent of those who received the 25 percent guarantor offer accepted, and 44 percent of those who received the 4 percent deposit offer accepted.
    Loan repayment rates and timing: About 63 percent of the farmers who took out a loan were late on their payments at least once during the course of the loan, but repayment performance did not vary significantly between treatment groups. The only major difference was the time taken to repay the loan: the average time taken to repay the 100 percent joint-liability loan was only nine months, while the other loans took between 17 and 22 months to pay off. No farmers defaulted on their loans and no tanks were repossessed. These preliminary results suggest that access to credit can be improved at little cost to lenders.
    Impact on economic and welfare outcomes: Preliminary results show that access to asset-collateralized loans (through the 4 percent deposit offer) improved milk production specifically for those without access to piped water before the start of the study. In addition, it reduced the time girls in treatment households spent fetching water by 35 percent, and increased the probability that girls were enrolled in school by 4 percentage points. This latter effect is striking because enrollment rates in the sample area were already high—girls’ enrollment at baseline was 94.6 percent. This was not the result of households being able to borrow to pay school fees (a common use of microfinance loans), but rather the ability to invest in an asset that reduced the demands on girls’ time.

    Unconditional Cash Transfers in Kenya

    While cash transfers have long been a subject of interest as a tool to fight poverty, the overall impact of large transfers of cash, given without conditions, to poor households has yet to be measured. This study evaluated a cash transfer program administered by the NGO GiveDirectly, which gave an average of USD 513 to poor families in rural Kenya. The transfers led to significant increases in income, assets, psychological well-being and female empowerment. Variations in the format and size of the transfers led to differences in outcomes.
    Detailed results can be found in the policy brief, available here and at the researcher's site, here.
    Policy Issue:
    Programs designed to alleviate poverty often focus on delivering goods or services (e.g. productive assets, training, bed nets, etc.) or capital conditional on certain behaviors to poor households. While these types of programs may be effective in achieving specific goals, they do not provide poor households with the choice and flexibility of allocating resources to meet the needs they find most pressing. An alternative approach to delivering support in-kind is to simply give money to poor households. However, this approach is sometimes received with skepticism, as there is no guarantee that money will be spent to achieve the specific impacts that donors desire. This evaluation studied what happened when poor families were given cash without any stipulations.
    Context of the Evaluation:
    This study took place in Kenya, a country at the forefront of the mobile money revolution. Since the launch of M-PESA, a mobile-phone based transfer service, in 2007, Kenya has become the country with most extensive retail payment network[1]. GiveDirectly is a non-profit organization that leverages the low costs of mobile money to deliver cash transfers to poor households, reducing the cost of delivery to only 10 percent of each donated dollar. Beneficiaries receive donated money on SIM cards and can visit a local M-PESA agent to exchange the mobile credit for cash. Residents in the area of Rarieda, where the study took place, generally live on an average of approximately 1 US dollar per day, and 64 percent of those surveyed said they did not have enough food in their house for the next day.
    Description of the Intervention:
    Households were eligible for the unconditional cash transfer program if they had roofs constructed from non-solid materials (mud, grass, etc). Study villages in the Rarieda district of Kenya were randomly assigned to a treatment or pure comparison group. GiveDirectly identified 1,000 eligible households in treatment villages. Within treatment villages, 500 eligible households were then randomly assigned to receive unconditional cash transfers. These households were compared to 500 control households in the same villages, which did not receive transfers. In addition, the 500 control households were compared with 500 households in pure comparison villages to identify any spillover effects of the intervention. Study households in treatment villages received a baseline survey before randomization, while pure control households where surveyed only at endline.
    Transfers were randomly assigned to go to women or men, and further randomized to be in the form of a single lump sum transfer of 25,200 KSH (about USD 287), or monthly transfers for nine months with the same total value. A third treatment arm was selected to receive a large sum, with 137 households each given an additional KES 70,000 (about 798 USD) in seven monthly installments of KES 10,000 each. 
    Money was transferred to beneficiaries using Safaricom’s M-PESA mobile payment system—sending money from the GiveDirectly account to the recipients’ SIM cards. Recipients were required to register with M-PESA, and received a text message when the funds were transferred. They could then visit a local M-PESA agent to transfer mobile credit to the agent’s phone in exchange for cash. Most households in the sample were within a 30-45 minute walking distance to an M-PESA agent. Households were given SIM cards to allow them to redeem money, and the opportunity to purchase a mobile phone if they did not have one, with the cost deducted from their transfer. 
    A follow-up survey one year after the first transfers collected data on income sources, investment, consumption, food security, school enrollment of children, and mental and physical health. These surveys were complemented by the collection of salivary cortisol levels to measure the impact of unconditional cash transfers on stress.
    Results and Policy Lessons:
    Assets and income: Assets and holdings for those who received transfers were 58 percent (USD 279, in purchasing power parity, or PPP) higher, primarily in home improvements (such as metal roofs, which are far less costly to maintain), and livestock holdings. The transfers increased income for recipients by 33 percent (USD 15 PPP), coming from sources such as livestock and non-agricultural businesses. There is little evidence that cash transfers change the primary source of income for recipients, but they do increase expenditures in non-agricultural enterprises by 10 USD PPP per month, with revenues 11 USD PPP higher. 
    Consumption: Monthly consumption for recipients of the transfers was 23 percent higher (USD 36 PPP), spread across nearly all categories measured: food, medical and educational expenses, home improvements, and social expenses such as weddings and funerals. The exceptions were temptation goods: there was no increased spending on alcohol or tobacco. The largest increase was in food consumption, which was 19 percent (USD 20 PPP) higher.
    Food security: Food security index scores were .25 standard deviations (SD) lower for transfer recipients. Specifically, they were 30 percent less likely to have gone to bed hungry in the preceding week, 20 percent more likely to have enough food in the house for the next day, and the number of days children went without food was 42 percent lower.
    Health and education: Spending on health education increased for transfer recipients, but from relatively low levels. There were no observed increases in health or education outcomes.
    Psychological and neurobiological measures: Overall there was a .20 SD increase in psychological well-being index, stemming from a .18 SD increase in happiness scores, a .15 SD increase in life satisfaction, a .14 SD reduction in stress, and a .99 SD reduction on a depression questionnaire. Levels of cortisol, a stress hormone as measured in saliva samples did not differ across the groups overall, but large transfers and transfers to women lowered levels for both men and women significantly. 
    Female empowerment: Positive spillover effects for female empowerment were observed, with an increase of .23 SD on an index of several measures not only for the treatment households, but for all households in the village, suggesting that the transfers can improve the standing of women in general.
    Types of transfers: With the exception of well-being outcomes mentioned above, there were no differences observed comparing transfers made to women versus men in the household. Monthly transfers were associated with a .26 SD increase in food security relative to lump sum payments, while lump sum payments were associated with higher asset values. Large transfers led to approximately twice the value of assets as small transfers, as well as higher scores on psychological well-being and female empowerment measures.
    More information can be found in the full policy brief, available here and at the researcher's site, here.

    [1] Klein, Michael, and Colin Mayer. May 2011. “Mobile Banking and Financial Inclusion: The regulatory lessonsWorld Bank Policy Research Working Paper 5664.

    HIV Prevention Among Youths: Evidence from a Randomized Controlled Trial in Kenya

    Policy Issue: 

    The vast majority of new HIV infections occur in sub-Saharan Africa, where nearly 2 million people become infected with HIV/AIDS every year. Forty-five percent of these new HIV infections occur among people under 25 years old, and nearly all of them are due to unprotected sex. Ensuring the adoption of safer sexual behavior among youth is critical to keeping the new generations free of HIV.

    The objective of this project is to examine, through a large randomized controlled trial, the impact of two HIV prevention strategies among a youth population in Kenya. The two strategies to be tested are: Voluntary Counseling and Testing for HIV (VCT), and condom distribution.

    • VCT is a critical entry point for access to HIV/AIDS treatment and care, and is being scaled up in many countries. But VCT could also be a powerful prevention tool. By providing personalized counseling as well as information about high-risk behaviors, VCT could motivate people to adopt safer sexual behavior and prevent transmission of HIV. This could be particularly important for adolescents and young adults, who typically have had their sexual debut but might not have perfect information about HIV risk. They are often still HIV-negative, and might be better able to change their sexual behavior.

    • Despite strong evidence of the biological effectiveness of the male condom as an HIV prevention strategy, condom use continues to remain low in many countries. Several factors, such as low availability, cost, lack of education about condoms and how to use them, and relationship factors contribute to low usage.  This study examines whether free and easy access to a large quantity of condoms can result in a reduction of risky behaviors and a decline in transmission of STIs among youth.

    Context of the Evaluation: 

    Kenya has the 10th largest HIV infected population in the world – nearly 7% of Kenyans are infected.1 The study is being implemented in four districts of Kenya’s Western Province (Butere, Mumias and Bungoma South and Bungoma East), spanning an area of approximately 50,000 square kilometers. Since about 45% of all new infections worldwide occur in youth aged 15-24 years, this study focuses on young people (both men and women). The sample is composed of approximately 10,000 youths (half of them female) who were 17 to 22 years old in 2009.

    Despite the expanding implementation of VCT, an estimated 80% of Kenyans living with HIV are unaware of their status. Take-up of VCT in traditional settings (such as government health centers) is low. As such, several alternative models of VCT service provision, including mobile VCT, workplace VCT and home-based VCT are being explored. This study has used both mobile VCT and VCT within homes, and has achieved a very high take-up (85% of the people who were offered VCT accepted it).

    Condom usage in Kenya is also relatively low. Only 24% of women aged 15-49 who reported multiple partners in the last 12 months used a condom during their last sexual encounter. Despite significant efforts to increase availability of free male condoms, recent data suggest that condom distribution remains low, with on average 0.71 condoms distributed per eligible person per year.

    Details of the Intervention: 

    A detailed baseline survey was administered to 10,420 youths (about ½ of them girls) between March 2009 and July 2010. All respondents were tested for Herpes (HSV-2) and for HIV (via anonymous linked testing) during the baseline. The prevalence rates for HSV-2 and HIV were 8.5% and 0.5%, respectively.

    Among those surveyed at baseline, 25%  had been randomly pre-selected to be offered VCT, 25% had been randomly pre-selected to receive free condoms, and 25% had been randomly pre-selected to receive both VCT and free condoms.

    Those pre-selected for VCT were offered VCT right after the baseline survey had been administered. Eighty-seven percent of them consented.  The consent rate was slightly higher among girls than among boys.

    Those pre-selected for free condoms were offered three boxes of 50 condoms each, right after VCT (if also sampled for VCT) or right after the baseline survey had been administered. Not all respondents offered condoms took them. Seventy-one percent took all 150 condoms, 19% took only some of them, and the remainder, 10%, refused to take any condoms. The acceptance rate was much higher among boys. While 87% of boys took all the condoms and only 5% took none, only 52% of girls took all 150 condoms and 15% took none.

    A follow-up survey will be conducted in 2011-2012. The survey will include detailed questions on sexual behavior, including sexual debut, number and type of partners, and condom use, as well as detailed questions on beliefs regarding HIV transmission, own HIV status, and own exposure to risk. Crucially, the follow-up survey will also include HSV2 and HIV testing.


    Results forthcoming.

    1 CIA World Factbook, “Kenya”. Available at

    Teacher Incentives Based on Students' Test Scores in Kenya

    Does linking teachers' pay to students' test performance improve educational outcomes, or just increase “teaching to the test”? This study examines the effects of a teacher incentives program on both teacher behavior and student test scores in Kenya. Student test scores increased significantly during the study period, but evidence suggests that this improvement came through test-preparation sessions outside of normal class hours. Test score improvements dropped off after the program was completed, implying there were little to no spillover effects of the test preparations onto actual learning.

    Policy Issue:

    Over the past decade, many developing countries have expanded primary school access, energized by initiatives such as the Millennium Development Goals which call for achieving universal primary education by 2015. However, these improvements in school access have not been accompanied by improvements in school quality. Poor learning outcomes may be due, in part, to high absence rates and low effort among teachers, who often lack strong incentives to perform well at work. Some argue that linking teachers' pay to students' performance may be a way to increase teacher effort; opponents argue this will result in “teaching the test” rather than better teaching of the curriculum.

    Context of the Evaluation: 

    Learning outcomes in Kenya are poor: although the vast majority of Kenyan children attend primary school, less than half complete their primary education in rural areas. Government officials and policy makers are hopeful that improved education quality will enable individuals to attain higher levels of education. But more schools and an improved curriculum can only go so far – students must be met with a motivated teacher in their classroom. This is often not the case in Kenya; in Busia and Teso districts, teachers are absent on average 20 percent of the time. Teachers’ salaries depend on their education and experience, with no opportunity for performance-based promotion, which appears to result in a system with no incentives to teach well.

    Details of the Intervention:

    This study examines the effects of a teacher incentives program on both teacher behavior and student test scores in Kenya. Out of 100 schools which the Ministry of Education designated as particularly in need of assistance, 50 were randomly selected for the treatment, while the other 50 served as a comparison group.  In collaboration with International Child Support (ICS), researchers designed and evaluated a program in Busia and Teso districts that provided prizes to teachers in grades 4 to 8 based on the performance of their school as a whole on the annual district exams. ICS offered in-kind prizes that ranged in value from 21 to 43 percent of the typical teacher’s monthly salary, values that were comparable to merit pay programs in the United States.

    All teachers who taught grades 4 through 8 in the 50 treatment schools were eligible for prizes. The program awarded “Top-Scoring,” and “Most Improved” prizes, as well as second, third and fourth place awards to teachers in winning schools. The program penalized teachers for dropouts by assigning low scores to students who did not take the exam, preventing them from selecting only the most qualified students to take the test. In all, prizes were awarded to 24 of the 50 participating schools.

    Data were collected on many types of teacher effort – attendance, homework assignment, pedagogical techniques, and holding extra exam preparation sessions – as well as on student test scores after the end of the program. The program ran for two years beginning in 1998, with 1996 exam scores used to measure improvements.

    Results and Policy Lessons:

    By the program’s second year, student test scores increased significantly in treatment schools. However, evidence suggests that this improvement did not necessarily occur through the intended channel of regular classroom teaching. Teacher attendance and student dropout and repetition rates did not improve, and no changes were seen in either homework assignment or pedagogy. Instead, teachers were more likely to conduct test-preparation sessions outside of normal class hours. Prior to the program, treatment schools were slightly less likely to offer test preparations, but after the introduction of the program, treatment schools were 4.2 percentage points more likely to conduct prep sessions in the first year and 7.4 percentage points more likely in the second. Furthermore, there was evidence that much of the increase in test scores was due to improved test-taking skills such as strategies for handling multiple-choice questions or “cramming” for tests that are prone to memorization, as opposed to increased overall knowledge.  Lastly, test score improvements dropped off after the program was completed, implying there were little to no spillover effects of the test preparations onto actual learning.

    Michael Kremer

    Teacher Training and Free Uniforms for HIV Prevention in Primary Schools in Kenya

    Policy Issue:

    The vast majority of HIV cases occur in sub-Saharan Africa, where nearly 2 million people become infected with HIV/AIDS every year. One quarter of these new HIV infections are among people under 25, and almost all are due to unprotected sex.  AIDS is incurable and no successful AIDS vaccine has been developed. Ensuring the adoption of safer sexual behavior among youth is critical to preventing the transmission of this disease. However, there is surprisingly little evidence concerning the relative effectiveness of different programs to reduce risky sexual behavior.


    Kenya has the 10th largest HIV infected population in the world – nearly 7% of Kenyans are infected.[i] Children are seen as a “window of hope” in the fight against AIDS, because their sexual patterns are not firmly established. In an effort to prevent HIV infections in new generations, in the late 1990s UNICEF and the Kenya Institute of Education jointly developed an AIDS education curriculum, including student and teacher handbooks. However, by 2003, this curriculum had not been fully implemented, likely due to teacher inexperience and discomfort with talking about this sensitive material.

    Description of the Intervention:

    This evaluation tested two interventions to reduce risky sexual behavior: training teachers on the existing HIV curriculum, and reducing the costs of schooling by providing free uniforms. The 328 study schools were randomly assigned to one of four groups of about 82 schools. Each of the four groups of schools received a different set of programs:

    In groups 1 and 3, three teachers were trained on HIV/AIDS and on how to teach the HIV curriculum. The curriculum covers facts about the disease, and encourages abstinence until marriage and faithfulness afterwards. It also teaches life skills, such as how to say “no” to unwanted or unsafe sexual relations.

    In groups 2 and 3, children already enrolled in sixth grade classes were given a free uniform. Implementers also announced that students still enrolled in school the following year would be eligible for a second uniform, and distributed uniforms again the following year.

    All in all, group 1 schools received the teacher training program only, group 2 schools received the uniforms program only, group 3 schools received both programs, group 4 received no program at all and thus served as a comparison group.

    To evaluate the impact of the two programs on sexual behavior and sexual health, survey data was collected on youths’ sexual behavior. Such survey data can be subject to reporting biases, however. It was therefore important to complement this data with an objective measure of the incidence of unprotected sex, which is the main mode of HIV transmission in Kenya. Two such measures were considered: (1) childbearing rates and (2) STI infection rates. Childbearing rates were monitored regularly between 2003 and 2010. STI infection rates (specifically, Herpes and HIV infection rates) were measured during a long-term follow-up in 2009-2010.


    Impact of Teacher Training only: Training teachers on how to implement the national HIV/AIDS curriculum greatly increased the likelihood that teachers teach about HIV in the classroom. Two years after the training students whose teachers had been trained had greater knowledge about the disease and also reported more tolerant attitudes toward those with AIDS. However, the intervention did not reduce childbearing rates among girls, suggesting that it did not decrease the likelihood that girls engage in unprotected sex. It also did not reduce the risk of STI as measured after 6-7 years.

    Impact of Free Uniforms only: Free school uniforms led students to stay enrolled for significantly longer, and reduced the incidence of teen marriage and teen pregnancy. Girls who benefitted from free uniforms were not less likely to have an STI after 6-7 years, however, suggesting that some of the adolescent girls in the free uniforms program, while less likely to engage in committed relationships that lead to pregnancy and marriage, might have engaged in casual relationships.

    Combined Impact: In schools that received both free uniforms and teacher training on the HIV/AIDS curriculum, the reduction in drop-outs and teenage pregnancy among girls was lower than that observed in schools that received free uniforms only. This suggests that the curriculum’s emphasis on abstinence until marriage may have persuaded some girls who would have delayed marriage thanks to the free uniforms to instead privilege committed relationships, where pregnancies are more likely. On the other hand, the two programs combined led to a significant reduction in the risk of STI. This suggests that among girls who chose to delay marriage in order to stay in school with the free uniform, the HIV curriculum convinced some to abstain altogether in order to avoid the STI risk associated with casual relationships.

    [i] CIA World Factbook, “Kenya” (accessed August 25, 2009). 

    Source Dispensers and Home Delivery of Chlorine in Kenya


    Policy Issue: 

    Diarrheal diseases are a leading cause of morbidity and mortality in the developing world, killing an estimated 2.6 million people per year between 1990 and 2000. Children under 5 experience an average of 3.2 diarrheal episodes per year1 and diarrheal diseases account for 20 percent of deaths in this age group.2  Even when diarrheal episodes are not fatal, they can have long-term impacts on children’s cognitive and physical development. Diarrheal diseases are often transmitted when a water supply is contaminated with fecal matter, and may be endemic in places where the water supply is irregular. Practices from handwashing to water source protection are proven to reduce diarrhea episodes, yet the adoption of such practices has been slow in regions across the developing world.

    Context of the Evaluation: 

    Despite widespread awareness of the dangers of drinking unsafe water, there is extremely low adoption of sanitation or clean water practices in rural Western Kenya. While three quarters of households have heard of point-of-use water chlorination and 70 percent admit that drinking dirty water causes diarrhea, only 5 percent of households report that their main drinking water supply is chlorinated. The most common method of water chlorination is through the individual purchase of chlorination products, which must be added to water at home. Community level chlorination has been considered as another strategy to increase chlorine take up. Much cheaper than individually packaged bottles, point-of-collection chlorine dispensers can be used at the sources where people collect their water. Here, social pressure may be maximized by making each individual’s sanitation choice publicly known.

    Details of the Intervention: 

    Researchers sought to examine the impact of factors including price, persuasion, promotion and the chlorination products themselves with a two-phase study. Prior to the study baseline surveys were administered to a random selection of households.

    In the first phase, households were given seven WaterGuard bottles, an individual water treatment product, each sufficient for one month’s supply of clean water. They were also provided with improved drinking water storage pots with a tap to prevent contamination and detailed instructions on use. One third of this group received twelve coupons for a 50 percent discount on WaterGuard bottles, each valid for one month during the next year, and calendars with reminders. Another third received additional verbal persuasion messages beyond the basic WaterGuard instructions, and another third received no additional coupons or messages. To estimate social networking effects, the free WaterGuard bottles were distributed in different percentages in each community, allowing researchers to see if higher community levels of use increased individual adoption. A follow-up survey was administered between 2 and 7 months after the free WaterGuard was distributed.

    In the second phase researchers compared six different treatments designed to increase WaterGuard adoption. For the first three treatments, scripted promotional messages were delivered at either the (1) household level, (2) community level, or (3) both. The second two treatments included repeated promotion of chlorination through a home visit by a community elected promoter. Despite volunteering to work for free, the promoter was paid either a (4) flat rate, or was (5) paid based on how many households had chlorinated water at follow-up visits. The last treatment (6) combined the incentivized promoter model with an unlimited supply of free WaterGuard delivered through a point-of-collection chlorine dispenser at the local water source. Follow-up surveys were conducted 3 weeks and 3-6 months after the start of the study.

    Results and Policy Lessons: 

    Impact of Free Home Distribution: Most households have a low willingness to pay for chlorine, despite its well known benefits. After receiving a free 7-month supply, chlorine was detected in 58 percent of households, much more than the 2 percent starting level. Still, only 10 percent of the distributed coupons were redeemed. Where WaterGuard bottles were distributed freely, additional persuasive messages had no effect on take up, and in retail markets they only had short-term effects. There appeared to be no “social networking” effects of living in a community with a higher level of chlorination, and no evidence was found that price was an effective screening mechanism to target households who are more likely to benefit from cleaner water.

    Impact of Persuasion: Hiring local community members at a low wage to promote chlorine use among their neighbors is highly effective at increasing use. Chlorine was detected in 40 percent of households visited by a promoter, compared to only 4 percent in those who weren’t visited. Incentivizing these promoters had only modest effects. Communities with point-of-collection chlorine dispensers in combination with promoters saw 61percent of households chlorinate their water, up from only 2 percent prior to the study, suggesting that this is a highly cost-effective way to promote take up.

    Scale-Up: Investments in marketing campaigns and coupon schemes proved to be ineffective strategies to encourage point-of-use chlorination. Free chlorination dispensed at water sources along with community promoters provided the most effective strategy to improve water cleanliness, potentially preventing diarrheal incidence in areas such as rural Kenya.


    1 Disease Control- Priorities Project, “Public Health Significance of Diarrheal Illnesses,”

    2 Parashar, Umesh, et al. “Global Illness and Deaths Caused by Rotavirus Disease in Children,” Emerging Infectious Diseases. Vol. 9. May, 2003.

    Cleaning Natural Springs in Kenya


    Policy Issue: 

    Diarrheal diseases are a leading cause of morbidity and mortality in the developing world, killing an estimated 2.6 million people per year between 1990 and 2000. Children under 5 experience an average of 3.2 diarrheal episodes per year,1 which accounts for 20 percent of deaths in this age group.2 Even when diarrheal episodes are not fatal, they can lead to severe dehydration and have long-term impacts on children’s cognitive and physical development. Diarrheal diseases are often transmitted when a water supply is contaminated with fecal matter, and may be endemic in places where the water supply is irregular.


    Context of the Evaluation: 

    Diarrhea is widespread in rural Kenya, where 43 percent of the population gets their drinking water from nearby springs, usually transported in 10 to 20 liter jerry-cans. Landowners have no incentive to improve the sanitation of water sources on their property because custom requires them to allow everyone free access to springs. As such, water is often contaminated by surface rainwater runoff as it seeps from the ground, and this contamination is spread to the population who collect drinking water at the spring. However, people are often reluctant to change their habits to include behaviors that may reduce diarrhea incidence such as hand washing or daily chlorination of water, resulting in low uptake for these “point of use” interventions.

    Details of the Intervention: 

    Researchers sought to find an inexpensive way to improve water quality at the source, and thereby reduce the burden of diarrhea by making improvements to springs. Working with a local NGO, they identified 200 springs in the Busia district of Kenya, and persuaded each local community to contribute 10 percent of the costs of the improvement project, usually in labor. At a cost of about US$1,000 per site, half of these springs had their sources encased in concrete, forcing water to flow through a pipe rather than seeping from the ground, thus preventing contamination from groundwater. NGOs conducted community meetings at which user committees comprised of local residents were selected and placed in charge of maintaining the protected springs. These committees were responsible for performing basic maintenance, including patching concrete and clearing drainage ditches, at an average cost of US$35 per year via community contributions.

    Household characteristics such as income, education and health were approximately equal among the two groups at the start of the program, suggesting that there were no systematic differences between communities that had their springs protected and those that didn’t. Throughout the program, statistics were collected on the level of water contamination and diarrheal disease in all communities and by examining changes in these measures, impacts of the intervention could be assessed.

    Results and Policy Lessons: 

    Impact on Diarrhea: The simple infrastructure investment of “spring cleaning” significantly reduced both water contamination and the incidence of diarrhea. There was 66 percent less E-coli contamination in treated springs than in untreated ones, and an average of 24 percent less contamination in users’ home water supplies among households who collected water from multiple springs and those who only used protected springs. This incomplete transfer of benefits may be due to the fact that households may transport and store water in contaminated vessels.

    Despite these mitigating factors, diarrheal incidence in children under 3 years old fell by 4.7 percentage points, or 25 percent, though there was no significant effect on children ages 5 to 12. Interestingly, diarrhea reduction was disproportionately concentrated among girls, suggesting that spring cleaning could be an effective tool for the improvement of female child survival.

    Behavior Modification: Possibly due to the apparent benefits of using protected springs, families began increasing their use of protected springs for drinking water, relative to other sources. However there were no significant changes in water transportation, home water chlorination, boiling or hygiene practices, implying the experience with significantly cleaned water did not increase people’s taste for water improvement.

    Willingness to Pay: Researchers used the information gathered about changes in use of protected springs to estimate how much time households were willing to expend to get cleaner water, and by extension how much they were willing to pay for it. These empirical estimates were approximately one third of what households report they are willing to pay for clean water and less than one-tenth of the value that policymakers often use when assessing social programs. The travel habits of residents of Busia, Kenya suggest that they are willing to spend at most 10.1 work days, or about US$0.89 to avert a diarrhea case. Using a high estimate for the value of time, this suggests a valuation of US$2,715 per averted child diarrhea death, far below the estimated value of a statistical life, and the cost-effectiveness cutoffs usually used in analyses of health projects in less developed countries. This implies that people either do not understand the causal link between clean water and diarrhea, which anecdotal evidence suggests they do, or that they place a lower value on improving infant and child health than typically assumed.


    1 Disease Control- Priorities Project, “Public Health Significance of Diarrheal Illnesses,” 

    2 Parashar, Umesh, et al. “Global Illness and Deaths Caused by Rotavirus Disease in Children,” Emerging Infectious Diseases. Vol. 9. May, 2003.


    Limited Insurance within the Household in Kenya

    Policy Issue:

    Individuals in developing countries are often subject to considerable financial risk, but most lack access to formal financial services that would allow them to insure themselves against unexpected income shocks like medical expenses or natural disasters. Instead, households often use informal systems of gifts and loans from friends or family to cope with large unplanned expenses. While these informal networks do provide some protection against shocks, they can also face problems of information asymmetry and payment enforcement. Many individuals may therefore attempt to cope with risk within the household where information sharing and enforcement are presumably easier. Whether intra-household insurance mechanisms are effective in insuring against risk remains, however, an important question. If members of a household do not insure each other against risk, then programs which impact the ability of individuals to cope with risk (such as formal savings accounts or microinsurance programs) could substantially increase people’s welfare.

    Context of the Evaluation:

    The towns of Busia, Sega, and Ugunja in Western and Nyanza Provinces of Kenya are semi-urban areas located along a major highway. Though many people in the area earn their living from agriculture, a substantial fraction earns at least some income from self-employment. The individuals in this study were drawn from a group of daily income earners (men who work as bicycle taxi drivers - called boda bodas in Kiswahili - and women who sell produce and other items in the marketplace). Daily income earners were targeted in this study because their informal employment made them more susceptible to transitory income shocks.

    A very small minority of the sample had access to formal savings: just 2 percent of men and 1 percent of women had savings accounts. However, informal savings and credit sources were common. Sixty-three percent of men and 44 percent of women participated in Rotating Savings and Credit Associations (ROSCAs) -a group of individuals who make regular cyclical contribution to a fund, which is then given as a lump sump to a different member at each meeting. Men and women were equally connected to informal credit groups (around 90 percent of both men and women received a loan in the past year and around 85 percent gave a loan).

    Description of Intervention:

    This study presents results from a field experiment in Kenya designed to test whether intra-household risk-sharing mechanisms (such as financial transfers within the household) operate efficiently. One hundred and forty-two married couples were followed for eight weeks. Every week, each individual had a 50 percent chance of receiving a 150 Kenyan shilling (US$2) income shock, equivalent to roughly 1.5 days’ income for men and 1 week’s income for women. As these shocks were, by definition, random, the experimental design made it possible to test for efficiency by comparing theresponsiveness of private consumption to shocks received by an individual and to those received by his spouse. Assuming that household members were risk averse, failing to insure temporary shocks (such as those administered in this study) would leave potential gains from trade unexploited. For example, if a woman is sick, without insurance (i.e. a transfer from her spouse), she may be unable to see treatment and subsequently will be unable to work and make a daily income, and to fulfill her responsibilities as caregiver for her family, producing inefficiency in the household. If the household pooled risk efficiently, increases in private consumption should be the same for shocks received by an individual and those received by their spouse.

    The shocks were announced to each spouse, so that it was not possible for one member of a household to hide an income shock from the other. Payments were made privately, however, and individuals were told that they could spend the money however they chose.

    Each week both spouses were visited separately by a trained enumerator who administered a detailed monitoring survey that included questions on consumption, expenditures, income (and income shocks), labor supply, and transfers given and received over the previous seven days. In addition, a background survey and a survey to measure risk aversion were administered.

    Results and Policy Lessons:

    The study rejects the unitary theory of the household which suggests that the household behaves like a single decision maker, and the collective model of the household in which members fully insure each other against shocks. In weeks in which they received the shock, men spent, on average, 16.9 percent of the increase on private expenditures. However, private expenditures did not change in weeks in which their wives received the shock.

    In contrast, for women, private expenditures did not respond to shocks (received either by herself or her husband). Women did, however, transfer 16.3 percent of the shock to their husbands, whereas husbands transferred none of their shock to their wives.

    If people spend windfall income differently than their regular labor income, the results may not be generalizable. The researcher addresses this possibility by examining how private expenditures respond to weekly fluctuations in labor income and finds that both men and women increase private expenditures in weeks in which their labor income is higher. This suggests that the experimental findings were not necessarily specific to the experiment.

    While the welfare consequences of failing to insure small shocks over a short time period are not likely to be very large, they suggest that intra-household risk-sharing mechanisms are ineffective, which could have important welfare effects. Overall, the findings suggest that programs which provide more formal risk coping mechanisms could have large effects on household welfare.

    Jonathan Robinson

    Barriers to Fertilizer Use: Evidence from a Field Experiment in Kenya

    Policy Issue:

    Agricultural outputs in Africa have stagnated over the past decades: although total output has risen, food production has not kept up with the increase in Africa’s population. The number of chronically undernourished people in Africa has increased to 200 million in 1997-99.1 When used correctly, chemical fertilizer can substantially raise agricultural yields, yet usage of fertilizer remains low in Sub-Saharan Africa. Past studies suggest that usage is low because farmers have difficulty saving harvest income to purchase fertilizer for the next growing season, have limited information on the benefits of using fertilizer properly, and the fact that knowledge about fertilizer is not passed from one farmer to another. This project attempts to address all three issues.

    Context of the Evaluation:

    This project focuses on small-scale subsistence farmers in rural Western Kenya, many of whom grow maize as their staple crop. All farmers in this population are extremely poor subsistence farmers, earning on the order of $1 per day. Previous research in this area has shown that when used correctly, top dressing fertilizer can increase yields by about 48%, amounting to a 36% rate of return over just a few months. However, only 40% of sampled farmers in the Busia district of Western Kenya report ever having used fertilizer.

    Details of the Intervention:

    This experiment looks at a complex intervention with several components meant to increase fertilizer use and dissemination of knowledge. Farmers were recruited to the study through meetings at primary schools and randomly divided into four groups.

    The first group received small, time-limited discounts which were valid within a 3 week window right after harvest, redeemable at a local shop. Farmers received coupons for a discount of about 15% of the price of fertilizer, for up to 25 kilograms.

    The second group was encouraged to form farmers’ cooperative with their friends and neighbors to talk about fertilizer and agricultural practices. The researchers organized the groups and coordinated the first few meetings, but did not provide any direct information to the groups.

    The third group participated in both the coupon scheme and the cooperatives.

    A fourth group received none of these services, and served as a comparison.

    Researchers will examine the changes in fertilizer usage between the different groups and whether farmers in the treatment groups talk to each other about agriculture more than others.  

    A separate intervention was designed to investigate the spread of information and technology when provided only to a subset of farmers in the treatment and comparison groups. The research team visited the randomly selected farmers and provided them with ½ teaspoon measuring spoons, as well as information about the returns to using ½ teaspoon of fertilizer per plant. To enable diffusion of this technology to others in the community, the spoons were made available in nearby fertilizer shops to other farmers for a nominal fee. In addition, when distributing the measuring spoons, the farmers were given vouchers for spoons which they could give to their friends. This intervention will test the hypothesis that the fertilizer discount intervention and the cooperative intervention could lead to greater diffusion of information about fertilizer.

    Results and Policy Lessons:

    Results forthcoming.


    1 Ernest Harsch, “Agriculture: Africa’s Engine for Growth”, available at

    2 Duflo, et. al., “Nudging Farmers to Use Fertilizer: Experimental Evidence from Kenya”, available at

    Vocational Education Voucher Delivery and Labor Market Returns in Kenya

    How effective is vocational training in Western Kenya? This study leverages a longitudinal survey to examine the impact on individuals with different childhood and background characteristics. Individuals are invited to apply (and then randomly selected) for a tuition voucher. The study will also measure the demand for vocational training, and the difference between public and private sector training institutes. 

    Policy Issue: 

    Youth underemployment, especially among less educated populations, has the potential to create significant social unrest and perpetuate poverty. However, little is known about how best to help youth find jobs and smooth the school-to-work transition, particularly in less developed countries. One would-be tool for expanding the labor market opportunities in these settings is vocational education, which could help students learn a trade and acquire the skills needed to take advantage of employment opportunities, and create successful small businesses. However, credible research on the economic returns to vocational schooling in poor countries remains scarce.

    Context of the Evaluation:

    The introduction of free primary education in Kenya in 2003 prompted a large influx of pupils previously not enrolled in school. As these pupils complete their primary schooling in the coming years, Kenya will face unprecedented numbers of primary school graduates competing for limited seats in traditional secondary schools. Vocational education is one promising avenue for these youths. Increasing vocational education has the potential to help address unemployment issues for new primary school graduates by providing students with the skills needed for employment, even without obtaining a secondary school diploma. Vocational training may also help to address the problem of underemployment among youth who exited the academic schooling system before 2003. However, there is little empirical evidence on the impacts of vocational education on employment, salary, or consumption for individuals with different characteristics and backgrounds, for instance, by gender, family background, or individual cognitive ability.

    Details of the Intervention:

    The program was launched in 2008 with the recruitment of 2,163 out-of-school Kenyan youths (roughly 18 to 26 years old) who had participated in the Kenya Life Panel Survey. In order to be included in the sample, applicants were required to attend two meetings on the program, participate in a short survey regarding their expected returns, complete a form ranking their preferred schools and courses, and submit a letter of support from either a local official or a school official. Applicants were then entered into a lottery to determine if they would receive a vocational training voucher worth up to approximately US$325, an amount sufficient to fully, or almost fully, cover the tuition costs for most private vocational education programs and government-run rural village polytechnics.

    Among the voucher winners, a randomly selected half received vouchers that could be used only in government supported public vocational training institutes, while the other half received unrestricted vouchers that could be used in either public centers or in the local private vocational training sector. Once winners were informed of which type of voucher they were eligible for, these young adults were able to access the tuition voucher, greatly reducing the cost (close to zero) of vocational education. The impact evaluation tracks these individuals (both voucher winners and non-winners) over time to evaluate take-up of vocational education, and its impacts on labor market and other life outcomes. The division of the treatment group into those with restricted versus unrestricted vouchers, combined with detailed panel data collection of individuals and institutions, will allow researchers to estimate the additional labor market returns of having access to the private vocational training sector. 

    Results and Policy Lessons: 

    Results forthcoming.


    Finding Missing Markets: An Agricultural Brokerage Intervention in Kenya

    Policy Issue: 

    In much of the developing world, farmers grow crops only for local or personal consumption, despite export options which are thought to be much more profitable. There are several plausible reasons why farmers might choose to grow crops for local markets, forgoing the opportunity to make more money through export crops. There may be information gaps about profitability of export crops, lack of access to the capital needed to make the switch to export crops, inadequate infrastructure to bring crops to urban centers, concern over risky export markets, or misinterpretation by researchers as to the true profit opportunities. 

    Context of the Evaluation: 

    Kenya’s horticultural sector, which includes fruit and vegetable production, has received a great deal of attention over the past decade due to the rapid and sustained growth of its exports to Europe. Although Europe’s appetite for Kenyan agriculture exports has been great, small farmers have largely failed to cash in on this opportunity. Many farmers instead receive below-market prices for their crops by selling them at the local market or to intermediaries, who then resell the produce at regional market centers or to export firms. In the study sample, about half of the household income came from agriculture, and most owned the land they cultivated, which was usually about one acre. Farmers grew subsistence crops (beans, maize, potatoes, and kale) 50 percent of the time and cash crops (coffee, bananas, and tomatoes) 34 percent of the time. Only 12 percent of farmers grew any export crops.  

    Details of the Intervention: 

    In collaboration with DrumNet, a Kenyan NGO, researchers evaluated whether a package of services could help small farmers adopt, finance, and market export crops, and thus make more income. DrumNet tried to link smallholder farmers to commercial banks, retail farm suppliers, transportation services, and exporters. To be a member of DrumNet, a farmer had to be a member of a registered self-help group (SHG), express interest in growing export crops marketed by DrumNet (i.e. French beans, baby corn, or passion fruit), and have irrigated land. 

    In 2003, researchers randomly divided the 36 active self-help groups (SHGs) in the Gichugu area into three equal groups: the first treatment group received all DrumNet services; the second treatment group received all DrumNet services except for credit; and the third served as the comparison.

    All individuals in the two treatment groups received a four-week orientation course, which explained the financing and selling process, and good agricultural practices. In additional, all treatment individuals opened a personal savings account with a local commercial bank to accommodate possible future business transactions. Individuals in the credit treatment group also contributed the equivalent of a week’s labor wages to an insurance fund, which would serve as partial collateral for a line of credit. After being organized into groups of five, which were jointly liable for individual loans taken out, individuals in the credit treatment group received an in-kind loan from a local agriculture supply store.

    At harvest time, for individuals in both treatment groups, DrumNet negotiated prices with an exporter and arranged a produce pickup. Once the produce was delivered to the exporter, the exporter payed DrumNet who, after deducting any loan repayments, credited the remainder to the individual savings accounts that each farmer opened when they registered. 

    Results and Policy Lessons: 

    Impact of the DrumNet Program: One year after the program began, treatment individuals were 19.2 percentage points more likely to be growing an export crop, but there were no significant gains in income for the full sample. However, among first-time growers of export-oriented crops, program participation led to a 31.9 percent increase in income.

    Out of the twelve SHGs in each treatment group, ten decided to take advantage of DrumNet services when credit was offered, compared to only five of twelve when it was not, implying that farmers perceived credit as an important factor for cultivating export crops. However, access to credit had no effect on income gains compared to no-credit SHG groups. 

    Long-term Consequences: Unfortunately, one year after the evaluation ended, the exporter refused to continue buying from the DrumNet farmers since none of the SHGs had obtained EU export certifications. This led to DrumNet’s collapse as farmers’ export crops were left to rot and loans went into default. Farmers returned to growing for local markets, underscoring the original concerns over export market risk. 


    Extra Teachers in Kenya: Peer Effects, Pupil Teacher Ratios and Teacher Incentives

    Policy Issue: 

    The introduction of free primary education has raised primary school enrollment in many developing countries. However, the resulting overcrowding of schools, as well as the influx of new students with little or no preparation, poses new challenges to policymakers. For example, increased enrollment has often not been matched by increased numbers of government-salaried teachers. One method of lowering the pupil-teacher ratio, versions of which have been used by many governments, is to hire low-paid local contract teachers to supplement government-salaried teachers. However, there are concerns that these teachers may be less experienced and therefore less effective. Empowering the local community to monitor teachers' performance may help to overcome this problem, and may also increase civil-service teacher effort and subsequently student learning.

    Context of the Evaluation: 

    Historically, Kenyan schools have had two types of teachers – those hired through the Ministry of Education and those hired locally and informally by Parent Teacher Associations (PTAs). For civil-service teachers, promotion, transfers, and disciplinary measures are decided through the Ministry of Education, rather than by more locally accountable bodies. These teachers are represented by a strong union, have civil-service protection, and receive wages and benefits considerably above market levels. PTA teachers, while typically paid much less than their civil-service counterparts, have much stronger incentives, partly because they do not have union protection, but also because a good track record as a contract teacher can help them obtain a civil-service job. 

    Incentives for civil service teachers, while weak, are largely based on their students’ scores on the primary school exit exam. Teachers may, therefore, have an incentive to focus on students they believe will perform better on the exam, while ignoring the weaker students who may be too far behind to catch up. 

    In 2003, Kenya eliminated school fees for primary school. This led to an almost 30 percent increase in enrollment and greater heterogeneity in student preparation. Many of the new pupils were first generation students and had not attended preschools. Since parents were no longer required to pay fees, local school committees were generally unable to raise the funds necessary to hire PTA teachers to match the influx of students and bring the new students up to speed.


    Details of the Intervention: 

    In collaboration with International Child Support (ICS), researchers evaluated three nested interventions that addressed the large class sizes and heterogeneity in student preparation in the Kenyan school system: (1) the addition of locally hired contract teachers to reduce the pupil-teacher ratio; (2) the empowerment of parents within PTA committees through School-Based Management (SBM) training; and (3) the sorting of students by initial level of preparedness to reduce heterogeneity within the classroom.  

    Out of 210 primary schools, 140 were randomly assigned to receive the Extra Teacher Program (ETP). The ETP program provided funding to hire a local contract teacher to address classroom overcrowding. The remaining 70 schools served as a comparison group. In each school, ICS held a meeting with parents and teachers to explain the program rules regarding the hiring of an additional teacher. School committees were responsible for hiring the contract teachers and were free to replace or keep the original contract teacher based on performance. The contract teachers were paid approximately one-quarter of the salary of regular civil service teachers, but had the same educational qualifications. 

    In 2005, the first year of the program, the ETP contract teacher was assigned to grade 1. Two sections were created for grade 1, one taught by regular civil service teachers (with multiple teachers rotating in an out of the class, teaching different subjects) and one taught exclusively by the contract teacher. As a result, average class size in grade 1 was only 44 in ETP schools, compared to 82 in comparison schools. In the second year of the program, the ETP teacher moved to grade 2, such that the cohort of first graders that benefitted from the ETP program continued to benefit once in grade 2.

    While in half of the ETP schools (the “non-tracking ETP schools”), students were divided into sections at random, in the other half (the “tracking ETP schools”), students were divided into sections based on students’ level of preparedness (as measured by exam scores during the first term). For all schools, which section was taught by the ETP contract teacher was then decided by random draw.

    Finally, half of the schools assigned to ETP, including both tracking and non-tracking schools, were randomly selected to receive School-Based Management (SBM) training. The training was designed to empower parents (within the PTA committee) to ensure a fair and objective recruiting process for the ETP teacher, as well as to monitor teachers’ performance. Two parents of grade 1 students were asked to perform attendance checks on teachers on a regular basis. 

    Standardized test covering math and literacy questions were administered to each school just before the program ended and again one year later. Four unannounced visits were also made to each school to measure teacher effort and to observe the classroom environment. 


    Results and Policy Lessons: 

    Learning Impacts: Providing school committees with funds to hire an extra teacher on a short-term contract had a positive effect on learning. Test scores of students in ETP schools were on average 0.22 standard deviations higher than those of students in comparison schools. However, not all students benefitted equally. The program impact varied a lot with the details of how it was implemented. In particular: 

    • The impact was much larger for students assigned to the ETP contract teacher (0.31 standard deviation gain) than for students assigned to the civil-service teachers (0.13 standard deviation). 
    • The impact was larger when the ETP program was accompanied with SBM training (0.25 standard deviation) than when it was implemented alone (0.19 standard deviation).
    • The impact was larger in tracking schools (0.28 standard deviation) than in non-tracking schools (0.16 standard deviation).
    • There was relatively rapid fade-out of these learning gains in the year after the program ended, except for the gains in tracking schools and the gains in SBM schools.


    Teacher effort: A crude measure of teacher effort is how often the teacher is found in class and teaching during a random spot check. This rate is 58 percent among civil-service teachers in non-ETP schools, suggesting relatively low levels of efforts. Despite their low pay, contract teachers were found 27.8 percentage points more likely to be teaching during a random visit. As a result, their students learned more. In contrast, civil-service teachers in ETP schools responded to the introduction of the ETP program by decreasing their own effort: the probability that they would be found in class teaching during a random visit decreased from 58 to 45 percent. As a result, their students did not significantly gain in learning over students in comparison schools, despite being in much smaller classes. One of the reasons why the SBM program led to increased learning gains is that it mitigated the negative effort response by civil-service teachers.

    Selection of Contract Teachers: Because a contract teacher position is a stepping stone for permanent civil-service positions, it is highly valuable (despite the low pay). As a result, existing civil-service teachers may want to hire their own relatives as contract teachers. In ETP schools without SBM training, 31 percent of contract teachers hired were relatives of existing teachers in the school. Those relatives tended to perform less well than other contract teachers. The second reason why the SBM program led to increased learning gains is that SBM increased the transparency of the contract-teacher recruiting process. Specifically, it reduced the number of contract teachers who were hired because they were related to an existing teacher by about half. 

    Homogeneous classrooms: Compared to students in non-tracking ETP schools, students in tracking ETP schools were in more homogeneous classes. This homogeneity benefitted them, irrespective of whether they had started at the top, middle, or bottom of the distribution. Homogeneity is beneficial in part because it allows teachers to better tailor their materials to the level of their students, but also because it increases teachers’ attendance in the classroom. For those who started at the middle of the distribution, being assigned to the bottom track (instead of the top track) had no detrimental effect on their learning, because the decrease in peer quality was compensated by a positive effect of being at the top of the class. 

    Overall, the results suggest that while hiring supplementary contract teachers can be part of the solution to Kenya’s teacher shortage, to get the most out of these teachers, implementation details matter. The biggest gains come when local school committees are empowered to oversee the recruitment process and to effectively monitor teachers, and when classes are structured so as to target instruction to students’ initial achievement level. 

    Flipcharts and School Inputs in Kenya

    Flipcharts are thought to promote learning in several ways, and may appeal to a broader range of students with different learning styles. Flipcharts may be particularly attractive in the rural Kenyan setting, where textbooks are too expensive for most students and many students have limited proficiency in English, the medium of instruction in Kenya and the language in which all Kenyan textbooks are written.

    The estimated impact of flipcharts on student test scores was found to be close to zero. A parallel non-randomized analysis estimates impacts five to ten times larger than the estimates based on the randomized trial. This suggests that the non-randomized estimates are subject to serious upward bias

    Policy Issue:

    While 85% of primary school-aged children are enrolled in school worldwide,1 these higher levels of enrollment may not be translated into increased learning for students in developing countries, where education quality is often low. Poor education may be due, in part, to lack of proper school materials, which are often undersupplied to poorer or more remote areas. There is a widespread belief that the provision of school materials can substantially improve educational outcomes in rural, underdeveloped areas, however there is little empirical evidence to confirm this.


    Context of the Evaluation: 

    The vast majority of Kenyan children attend primary school, although in rural areas less than half complete their primary education. The schools in this study are located in Busia and Teso, two neighboring agricultural districts on the border with Uganda, both of which have below-average income for Kenya. Flipcharts and other visual aids are rare in schools in these areas, and less than one third of the schools had any flipcharts before the study. Even textbooks are rare: in grade 8, which is selective, about 40% of students had textbooks in math and English, but 15% or less had textbooks in science and other subjects. In lower grades, textbooks are even rarer.

    Flipcharts are thought to promote learning in several ways, and may appeal to a broader range of students with different learning styles. Flipcharts may be particularly attractive in the rural Kenyan setting, where textbooks are too expensive for most students and many students have limited proficiency in English, the medium of instruction in Kenya and the language in which all Kenyan textbooks are written.

    Most analyses of the effect of educational inputs are based on retrospective studies, which compare schools with different levels of inputs. One potential weakness of this approach is that observable inputs like textbooks and flipcharts may be correlated with unobserved variables that affect educational outcomes. For example, if parents who provide better home environments (a typically unobserved characteristic) tend to organize to obtain more observed school inputs for their children, then estimates of the effect of those school inputs on test scores may be biased upwards. This evaluation compares retrospective and prospective estimates of the effect of flipcharts in Kenyan primary schools.


    Details of the Intervention: 

    In the prospective study, a Dutch NGO, ICS, provided five sets of flipcharts—large, poster-sized charts with instructional material that can be mounted on walls or placed on easels—to each of 89 randomly selected Kenyan primary schools. The intervention included flipcharts covering mathematics, geography, health, and two charts for science (agriculture and general science), as well as a teacher’s guide for science. The charts are not kept in the classroom, but rather are brought in when they are relevant to the day’s lesson, and can therefore be used in more than one classroom on any given day. In practice, the grade 7 and 8 teachers have priority over the usage of the charts, and account for roughly 60-75% of total use.

    The data available for this study are the 1997 and 1998 scores from the Kenya Certificate of Primary Education (KCPE) exam, which is taken at the end of grade 8, and the scores of grade 6-8 students on practice exams taken throughout 1997 and 1998. Test scores from schools that received the flipchart intervention were compared to the comparison group. Pre-intervention data on school performance across all subjects was also examined.

    The retrospective analysis uses data for 100 schools involved in a separate study that provided textbooks and grants for flipcharts to randomly selected schools. The effect of flipcharts and other inputs was estimated using the 1998 practice and KCPE exam scores for grades 6-8.


    Results and Policy Lessons:

    The comparison between retrospective and prospective estimates of the effect of flipcharts in Kenyan primary schools finds that randomized prospective estimates are much smaller than retrospective estimates. Despite a large sample size and two years of follow-up data, in the prospective study the estimated impact of flipcharts on student test scores was found to be close to zero.

    In contrast, using the retrospective framework, several conventional ordinary-least-squares (OLS) estimates suggest that flip charts raise student test scores by 20 percent of a standard deviation. This suggests that retrospective estimates are subject to serious upward omitted variable bias, even when controlling for other observable inputs. The relationship between school inputs and educational outcomes is critical for educational policy; however policy should reflect prospective study findings as retrospective studies may produce misleading results. 

    1 UNICEF, “Child info, Statistics by Area/Education,”

    Michael Kremer

    The Illusion of Sustainability: Comparing Free Provision of Deworming Drugs and Other "Sustainable" Approaches in Kenya

    Policy Issue: 

    Intestinal helminths—including hookworm, roundworm, schistosomiasis and whipworm—infect more than one in three people worldwide and  at least 800 million of these are school-age children. Worms are believed to have a negative impact on child development, and can contribute to lower educational attainment and income later in life. Intestinal worms can be effectively treated with low-cost drugs, but treatment must be continued indefinitely to prevent re-infection. Finding sustainable approaches to providing deworming drugs is a pressing research question, as most deworming interventions are currently financed by external institutions. Practices such as health education or cost-sharing may be able to increase program sustainability, but there is little systematic evidence on this matter.

    Context of the Evaluation: 

    Busia district is a poor and densely-settled farming region in western Kenya adjacent to Lake Victoria. This area has some of the country’s highest helminth infection rates; upwards of 90 percent of children aged 6-18 are infected. This is in part due to the area’s proximity to Lake Victoria—schistosomiasis is easily contracted through contact with the infected lake water. Other types of helminths can be transmitted through contact with or ingestion of fecal matter. This can occur, for example, if children do not have access to a latrine and instead defecate in the fields near their home or school, areas where they also play. 

    The prevention and treatment of infectious diseases such as worms is a priority for health officials, and more efficient and sustainable programs could enable the delivery of health care to a larger number of people. Advocates of improving sustainability concentrate on health education, community mobilization, and cost-recovery from program beneficiaries, to complement the more standard practice of subsidizing health products.

    Details of the Intervention: 

    This study evaluated the Primary School Deworming Project (PSDP), which was carried out by NGO International Child Support in 75 schools, randomly divided into three groups (1, 2, and 3) and phased into treatment over several years. All schools with helminth prevalence over 50 percent were treated periodically with albendazole, as well as with praziquantel if the local prevalence of schistosomiasis was high enough.

    Cost-Sharing:  In 2001, 25 of the 50 Group 1 and Group 2 schools were randomly selected to pay user fees for deworming treatment. Two thirds of the schools participating in cost-sharing received albendazole at a cost of US$0.40 per family, and one third received both albendazole and praziquantel (depending on the local prevalence of schistosomiasis) at a cost of US$1.30 per family. The per family basis of the fee introduced within-school variation in the per-child cost of deworming, since households have different numbers of children.

    Health Education: In addition to medicine, all treatment schools received regular public health lectures, wall charts on worm prevention, and training for two teachers from each school. The lectures and teacher training provided information on worm prevention behaviors—including washing hands before meals, wearing shoes, and not swimming in the lake. 

    Verbal Commitments: A verbal commitment "mobilization" intervention asked people to verbally commit in advance to adopt the deworming drugs. 

    Social Learning: A questionnaire was conducted in 2001 to test whether households with more “social links” to schools which received early treatment would be more likely to take deworming drugs. Respondents were asked about the friends and relatives they speak with most frequently about child health issues, and the degree of “linkage” to treatment schools was established on this basis.

    Results and Policy Lessons: 

    Cost-Sharing Intervention: The introduction of a small fee for deworming drugs led to an 80 percent reduction in treatment rates, consistent with the hypothesis that people have low private valuation for deworming. Take-up dropped sharply when going from a zero price to a positive price, but was not sensitive to the exact positive price level, suggesting that it may be counter-productive to charge small positive prices for the treatment of infectious diseases. 

    Health Education & Verbal Commitment Impact: An intensive school health education intervention had no impact on worm prevention behaviors. Child health is likely to be worsened to the extent that funds are diverted from medical treatment into health education in this setting. Asking people in advance whether they planned to take deworming drugs also had no impact on adoption.

    Social Learning: Individuals in treatment schools who had more extensive social networks, and therefore presumably had more information about deworming drugs, were significantly less likely to consent to take the drugs. For each additional social link to a family that had already received treatment, a family’s child was 3.1 percentage points less likely to take the drugs, and  these individuals were also more likely to believe the drugs are “not effective.” Negative social effects on take up are especially large for families with more knowledge about deworming, which may be due to overly optimistic prior beliefs about the net private benefits of the drug. A high proportion of deworming benefits flows not to the treated child or her family, but to others in the local community through positive externalities; the lower take-up among those with more deworming knowledge may reflect their understanding of this fact. 

    Overall, findings suggest that socially desirable health technologies with low private benefits may not spread on their own, due to low private estimations of the benefits that are reinforced through social networks.

    Textbooks and Test Scores in Kenya

    Policy Issue: 

    Education in developing countries is often hindered by large class sizes and low teaching quality. Over the past decade many developing countries have expanded primary school access, energized by initiatives such as the United Nations Millennium Development Goals, which call for achieving universal primary education by 2015. But increased quantity has not always been matched with improved quality, and only 60 percent of appropriately aged children gain the skills necessary to attend secondary school worldwide. In sub-Saharan Africa, that number drops to only a quarter.1 There is a widespread belief that the provision of textbooks can substantially improve educational outcomes in developing countries, but there is little empirical evidence as to whether providing textbooks alone can overcome the effects of other systemic problems such as high teacher-student ratios and frequent absenteeism.

    Context of the Evaluation: 

    In Kenya, 50 percent of families live below the poverty line, and high teacher absenteeism is reported in schools.2 Nevertheless the overwhelming majority of children start primary school. But some schools appear to promote only strong students to grade 8 in order to maintain high average scores on secondary school entrance exams, and those not promoted often drop out. Many students also drop out in earlier grades – 35 percent of the surveyed students in grade 3 dropped within the next three years. 

    Primary school is provided at a low cost, with some expenses, such as textbooks, paid by parents. Because of this cost, schools in Busia and Teso usually have textbooks for teachers to use, but few textbooks for children. At the start of this study, 80 percent of students in the sample were in classrooms with less than one English textbook for every 20 students, and the analogous figures for math and science textbooks were 78 percent and 89 percent, respectively. Government officials and NGOs are hopeful that textbook provision will improve test scores, encouraging higher grade attainment and overall improved educational outcomes.

    Details of the Intervention: 

    In 1995, International Child Support (ICS) began a program to improve primary education in Kenya’s Busia and Teso districts by providing additional official government textbooks. J-PAL affiliates, in an effort to determine the impact which textbooks have on students, teachers, and overall learning, evaluated this program. ICS randomly divided 100 needy, low performing schools into four even groups. In each year, program schools were compared to those not yet receiving the program.

    Program Design


    Distribution of Textbooks

    School Year


    Group 1

    Math books provided to grades 3, 4, 5, 6 and 7

    English provided to 3, 5 and 7

    Science and Agriculture provided to grade 8



    Group 2

    Grant equal to US$2.65 per student, 43% of which was spent on textbooks



    Group 3

    Similar grant in 1998 school year



    Group 4

    Similar grant in 2000 school year



    Sharing textbooks is common in Kenya, and two or three students typically share a workspace. Hence, a 60 percent textbook per pupil ratio was provided in English and science, and a 50 percent ratio was provided in math, giving nearly all students shared access to a textbook. A pre-test in the form of a district exam was administered to all schools before the textbooks were distributed, and district exam scores were collected at the end of each of the subsequent program years for comparison data. Classroom activities, attendance and dropout rates were also monitored.

    Results and Policy Lessons: 

    This study found no evidence that textbook provision increased average test scores, or that it reduced either grade repetition or dropout rates. Textbooks increased progression to secondary school for eighth graders but did not reduce grade repetition or raise attendance in lower grades. This provides evidence for the hypothesis that the program mostly benefited strong students, since only strong students reach grade 8 and have a hope to progress to secondary school, while many students in the lower grades were actually unable to read the textbooks. Students in grade 8, a selective and academically strong group, were 5 percentage points more likely to enter secondary school in the second program year than comparison school students.  

    Kenyan students are extremely heterogeneous in their family background, preparation for schooling, and economic circumstances. The Kenyan curriculum remains oriented to elites with language of instruction in English, most students' third language. Textbooks are meant to pair with this curriculum, arguably better suited toward elite students and ill-suited for academically weaker students. The evidence from this evaluation supports such a picture of the Kenyan education system, and suggests that better suited materials might produce achievement gains in a wider section of the population. 

    1 UNICEF, “Basic Education and Gender Equality,”
    2 As of 2000. CIA World Fact Book, “Kenya,”

    Michael Kremer

    What Matters (and What Does Not) in Malaria Prevention in Kenya

    Insecticide-treated bed nets have been proven highly effective in preventing malaria, reducing maternal anemia, and infant mortality, both directly for users and indirectly for non-users in their vicinity. Despite their proven impact, less than half of Kenyans sleep under a bednet. This study tested willingness to pay by households and a range of marketing effects. The demand for bed nets is very sensitive to price - an increase in price from free to $1 leads to a drop of 35 percentage points in take up. However gaining access to a free bednet increases households likelihood of buying one in later years. The marketing messages had no impact. 

    Policy Issue: 

    Over 10 million children under 5 die every year in the world. It is estimated that nearly two thirds of these deaths could be averted using existing preventative technologies, such as vaccines, insecticide-treated materials, vitamin supplementation or point-of-use chlorination of drinking water. A key policy question is how to increase availability and adoption of these technologies. In particular, what are the roles of prices, social networks and marketing in the adoption of such products? A commonly proposed way to increase adoption in the short-run is to distribute those essential health products for free or at highly subsidized prices. The rationale for some subsidization is evident for health interventions that generate positive health externalities. In addition, when the majority of the population is poor and credit-constrained, subsidies might be necessary to ensure access to the technologies.

    For products like vaccines, one-time adoption is sufficient to achieve eradication of the corresponding disease -- every child needs to be immunized only once. But other products, such as water treatment kits or anti-malarial bednets, require sustained adoption and use to generate the hoped-for health impact. A key question is whether policies aimed at achieving immediate adoption of such technologies increase or dampen their long-term use. It is often argued that free or highly subsidized distribution may generate a “dependency” effect, whereby beneficiaries anchor around the subsidized price and refuse to pay for the product once the subsidies are lifted. Furthermore, if people do not put free products to good use, incorrect information about the quality of the product might diffuse through the community. In this context, marketing messages might be important to increase adoption.

    Context of the Evaluation: 

    In Kenya, malaria is responsible for one out of every four child deaths.1 It impacts economic growth and productivity, and almost 170 million working days are lost annually due to the disease.2 Insecticide-treated bed nets (ITNs) are used to prevent malaria infection and have been proven highly effective in reducing maternal anemia and infant mortality, both directly for users and indirectly for non-users with a large enough share of net users in their vicinity. ITNs have been shown to reduce overall child mortality by 18% and reduce morbidity for the entire population. Despite their proven efficacy, less than half of Kenyans sleep under an ITN. Priced at US$5-7 per net, they are unaffordable to most families. Recently, a new generation of ITNs was invented: the long-lasting ITN (LLIN), which keeps its insecticide properties for its entire lifespan (typically 3-4 years).

    Details of the Intervention: 

    Households were given a voucher for a LLIN at a randomly assigned subsidy level, ranging from 40-100%. The final prices ranged from 0 to US$4.60 and households had three months to redeem their voucher. Twelve months after the distribution of the first LLIN voucher, households received a second LLIN voucher, redeemable at the same retailer as the first LLIN voucher received a year earlier. Unlike the first voucher however, all households faced the same price (US$2.30) for this second voucher. By comparing the take up rate of the second, uniformly-priced voucher in the second phase price groups, researchers are able to test whether being exposed to a large or full subsidy dampens or enhances willingness to pay for the same product a year later.

    This study also evaluated the effects of two interventions based on behavioral models derived from psychology: varying the framing of the perceived benefits; and having individuals verbally commit to purchase the product. At the time they received their first voucher, households were exposed to a randomly assigned marketing message. The “health framing” group emphasized the morbidity and mortality due to malaria which could be avoided by using the net. The “financial framing” group emphasized the financial gains households would realize (from averting medical costs and loss of daily income) if they could prevent malaria. A third group received no marketing message. Finally, a randomly selected half of all the households were asked to verbally commit to buy the ITN, and state who would sleep under it once they had bought it.

    Results and Policy Lessons: 

    Price Sensitivity: The demand for malaria-preventing bed nets in Western Kenya is relatively price sensitive; an increase in price from $0 to $1 leads to a drop of 35 percentage points in take up, and an increase from $1 to $2 leads to a further drop of 25 percentage points. Although the price effects are large, the price-elasticity observed here is lower than that found in other similar studies, possibly because households in this experiment had three months to redeem their voucher, and therefore time to save for it.

    Diffusion Effects: Gaining access to a free or highly subsidized LLIN in the first year increased households’ reported, as well as observed, willingness to pay for a second LLIN. This positive experience effect trickles down to others in the community: households facing a positive price were more likely to purchase the LLIN when the density of households around them who received a free or highly subsidized LLIN was greater.

    Marketing Effect: Neither of the two framing options (health or financial) had any impact at all on LLIN take up, and women to do not appear to have a different price-elasticity than men. Likewise, the verbal commitment treatment had no impact on actual investment behavior, despite a 92% initial agreement to purchase the LLIN.

    1The World Bank, “News & Broadcast: World Bank Intensifies Anti-Malaria Efforts in Africa”, (Accessed August 26, 2009)
    2The World Bank, “Booster Program for Malaria Control in Africa – Kenya,” (Accessed September 14, 2009)


    Pascaline Dupas

    The Latrine Training Mat Project

    In many countries, sanitation facilities, such as simple pit latrines are common and are helpful for maintaining sanitation and preventing illness. However, young children often continue to defecate in the open long after they are old enough to use the latrine finding open pit latrines intimidating and challenging to use. Innovations for Poverty Action has developed a simple, affordable, and scalable tool called the Safe Squat ™ latrine training mat for use in such contexts. Our training mat promotes good sanitation practices from an early age and fosters a life-long habit of latrine use by converting the latrine floor into a child-friendly, easy-to-clean surface.  The Latrine Training Mat Project has piloted several prototypes of the mat in rural Western Kenya with promising results and is currently working to pilot the tool in new locations. 

    Policy Issue:

    The World Health Organization estimates that 1.5 million children die each year from diarrheal disease.[1] Evidence demonstrates that families using latrines are less likely to have children with diarrhea than those who dispose of feces improperly, in the trash, or in the open near the household. .[2] However, access to a latrine is not enough to ensure safe disposal of children’s feces. Young children, particularly those under the age of five, often do not use latrines even when they have consistent access to one.[3]Latrines, often simple pits in the ground, can be difficult for young children to use, discouraging proper sanitation practices.

    Evaluation Context:

    Among world regions, Sub-Saharan Africa has the highest proportion of basic sanitation use, including open pit latrines (without a slab or platform), bucket latrines, hanging latrines, pour flush latrines that are not connected to a sewage system, and open defecation. Open pit latrines without a slab or platform can be particularly intimidating and challenging for children to use.  Based on a series of in-depth interviews with mothers in rural Western Kenya, IPA found that children defecate in the open long after they might be capable of using a latrine for two main reasons.  First, the hole of a pit latrine is often wide enough to frighten a small child, if not pose a serious safety risk.  Secondly, mothers are reluctant to promote latrine use for young children, since the messes they create make the latrine unpleasant for other families to use.  Ironically, frequent cleaning of the latrine floor can exacerbate the problem, as the size of the hole grows when the floor is scrubbed and the mud erodes away.

    Description of the Intervention

    IPA has designed a latrine training mat (LTM) called the Safe Squat ™  that is a flat, square slab made of plastic or treated wood, approximately 60 cm across with a tapered hole about 13cm wide at the center. It is elevated a few centimeters from the ground on risers and temporarily fits over the existing latrine hole.  The Safe Squat ™ training mat is designed to safely promote good sanitation practices from an early age, while saving mothers valuable time that might otherwise have been spent cleaning the latrine or disposing of feces.

    IPA piloted the mat among two villages in Matungu District of Western Kenya.  Three prototypes were designed and tested among 12 households (six from each village). The prototypes were made of treated wood, temporary plastic, or permanent plastic, with the objective of determining the most acceptable material and design for the mat.  Both wooden and temporary plastic models were designed solely for children, and could be placed on and off the hole as needed. The permanent plastic mat was designed for the whole family to use, but the hole in the center of the mat was approximately 5 cm wider than the other two prototypes.  Data collected was qualitative in nature, consisting of in depth interviews and focus group discussions. This type of data collection assured a detailed and nuanced understanding of the participant’s experience with the latrine training mat. Following an in depth interview regarding her child’s defecation practices, mothers from each household received one of the three mat prototypes, and agreed to use the training mat with her child for at least one week. Researchers used the Trials of Improved Practice (TIP) methodology to assess whether the method of intervention delivery would influence the way in which the intervention was used and tested the intervention presentation in two ways.   In the first village, field officers merely explained that the mat was a sanitation tool to help young children use the latrine. In the second village, participants received the mat along with a detailed description of its main features, as well as explicit instructions on how the mat should be cleaned and stored.


    Pilot Results:

    The mat was well received by intervention participants in the Kenya based pilot.  The mothers that participated in the pilot reported that they liked the tool, and reported that it saved valuable time otherwise spent cleaning the latrine, or disposing of feces. They also reported that their children liked and used the mats regularly, and that other household members approved of the tool as well. Although pilot households preferred the permanent plastic mats for the whole family’s convenience, the temporary plastic mat remained the most acceptable choice for children under the age of five. There were no observed differences in mat use, between the village that received messaging, as opposed to the one that did not.  Based on these promising results, the Latrine Training Mat Project plans to conduct future pilots in new countries. If future pilots are successful, the Latrine Training Mat Project hopes to test the mats as part of a larger randomized controlled trial. 


    [1]World Health Organization. “Diarrhoeal Disease” Fact Sheet No. 330. August 2009. Retrieved from November 16, 2011.

    [2]Mertens, T; Jaffar, S; Fernando, M.A; Cousens, S.N.; Feachem, R.G. Excreta disposal behavior and latrine ownership in relation to the risk of childhood diarrhea in Sri Lanka. International Journal of Epidemiology. 21 (6); 1157-1164, 1992.  


    [3]Gil, A; Lanata, C; Kleinau, E; and Mary, P. Children's feces disposal practices

    in developing countries and interventions to prevent diarrheal diseases: a literature review.

    Environmental Health Project (EHP). 2004.


    The Impact of Distributing School Uniforms on Children's Education in Kenya

    Policy Issue:

    Children in developing countries face numerous barriers to accessing basic education. According to the World Bank, school fees (fees for books, materials and some exams) are among the major obstacles to universal primary education in developing countries. Several countries in sub-Saharan Africa have taken strides towards meeting the Millennium Development Goal of universal primary education by 2015 by eliminating school fees,but other significant costs remain, including the cost of providing a school uniform for a child. Many governments and NGOs have sought to overcome this barrier by offering free or subsidized uniforms to particular students, often as part of a “child sponsorship” program. However, no prior studies have examined the impact of providing school uniforms on school participation and education outcomes.

    Context of the Evaluation:

    In Kenya, students were required to pay school fees to attend school through 2002. In January 2003, a new government policy eliminated fees, which had amounted to approximately US$4 in the Busia area, and also provided basic textbooks. Students were, however, still required to purchase and wear uniforms, and historically, students who did not wear uniforms could be sent away from school at the discretion of the principal. School uniforms cost between US$4.33 and $7.33 for girls and between US$5.40 and $7.33 for boys. Amongst sampled primary schools in rural Busia, 78% of students were likely to be wearing uniforms, but only 5% to be wearing shoes.

    Description of Intervention:

    The NGO ICS-Africa operates a Child Sponsorship Program (CSP) in Western Kenya, in which children sponsored by donors in the Netherlands and elsewhere receive school fees and school uniforms. Some portion of the sponsorship fee paid by the donors is used to give general assistance to the school: grants for construction, health care programs, and agricultural programs. In 2002 ICS-Africa planned to phase CSP into several new schools and agreed to randomize the aspects of the program directed at individual children. ICS used a lottery to distribute some of the child sponsorships within each of twelve primary schools in western Kenya.

    The Kenyan school year runs from January to December. In mid-January 2002, ICS organized a census of children in standards one through four of the twelve selected schools. Based on that census, ICS selected all children who had experienced one or both parent deaths (orphans) to receive sponsorships.It then used a lottery to randomly select the remaining beneficiaries, and studied the effects on this population.

    Schools immediately received some basic benefits from being sponsored. A pair of ICS nurses visited each school several times a year and provided basic first aid to any child or local adult who requested it. An agricultural representative organized student clubs to grow crops on the school grounds. In fall 2002, each school received a sizeable grant for classroom construction, for desks, and for books. The only individual benefit that sponsored children received was a school uniform; in June 2002, uniforms were distributed to all sponsored children who were still in school.

    An attendance dataset recorded pupil attendance at each of the twelve schools from 2002 through the end of 2004. Attendance was gathered as field officers made unannounced visits to each school multiple times each year and recorded whether each child was present. From these multiple visits, an annual child attendance average is collected.

    Results and Policy Lessons:

    Results indicate a strong positive impact of receiving a school uniform on student school participation. Giving a uniform reduces school absenteeism by 6.4 percentage points (43%) from a base of 15% school absenteeism. The effect is 3.4 percentage points larger for students who did not have a uniform at the baseline. This is a major reduction in absenteeism from a baseline school attendance level of 85%.

    The program appears to have had a positive impact on test scores in 2003, raising average test-scores of recipients by one quarter of a standard deviation. While the average test scores of uniform recipients are still observed to be higher two years after the program started by one fifth of a standard deviation, the effect is less precisely estimated.

    The average effect of the program is an increase in school participation of 0.064 years per treated child. The average cost of a school uniform is 436.86 Kenyan shillings (US$5.82). Thus, the cost of increasing school by one year is $5.82 / 0.064, or US$90.94. While this is more than other interventions, such as deworming, which were tested in this area, it is still considerably less than the cost of many programs which give beneficiaries cash, rather than goods.

    Michael Kremer

    Nudging Farmers to Use Fertilizer: Experimental Evidence from Kenya

    Use of inorganic fertilizer has the potential to dramatically increase yields and, if used correctly, is a highly profitable investment. So why do so few farmers in sub-Saharan Africa use it? Is it lack of information about profitability, lack of money to purchase the product, or an inability to save for the purchase? Researchers designed an intervention to offer farmers the ability to save harvest income for future fertilizer purchase. An ICS officer visited farmers immediately after the harvest, and offered to sell them a voucher for fertilizer, at the regular price, with free delivery later in the season. The program was popular, and in the first season increased usage by 14 percentage points. In the second season, the increase was even bigger, increasing usage by 18 percentage points. These effects are comparable to those obtained from a 50% price subsidy.

    Policy Issue:

    By some estimates, there are approximately 1.4 billion people living on less than $1.25 a day,1many of whom are farmers. As such, identifying ways to increase agricultural incomes is crucial in meaningfully alleviating poverty. Such strategies are especially important in sub-Saharan Africa, a region in which agricultural yields have been low and remained stagnant for many years. Use of inorganic fertilizer has the potential to dramatically increase yields and, if used correctly, is a highly profitable investment. Why do so few farmers in sub-Saharan Africa use fertilizer- is it lack of information about its profitability, lack of money to purchase the product, or an inability to save for the purchase?

    Context of the Evaluation: 

    An estimated 66% of the population of Kenya’s Western Province lives below the poverty line2where the majority of small farmers grow maize as their staple crop. Improving agricultural productivity through increased use of inorganic fertilizers could have substantial benefits for the livelihoods of these subsistence farmers. Numerous agricultural trials on experimental farms, as well as experiments where farmers were given full fertilizer, showed that adding fertilizer once the plants had sprouted (as top dressing) generated a 70% annualized return in Western Province. However, only 40% of sampled farmers in the Busia district report ever having used fertilizer. The overall goal of this research program is to understand why farmers do not invest in fertilizer. This part of the project investigates whether difficulty in saving harvest income until the time that inputs are needed is a significant barrier to adoption.

    Details of the Intervention:

    In collaboration with the NGO International Child Support (ICS), researchers designed an intervention to test if providing mechanisms to save harvest income for future fertilizer purchase could be effective in increasing usage. The intervention was called the Savings and Fertilizer Initiative (SAFI). The design of the experiment allowed researchers to test the impact of the SAFI program against various other strategies to improve usage, in particular fertilizer subsidies.

    The following interventions were tested over two seasons among a sample of farmers:

    1. Basic SAFI: An ICS officer visited farmers immediately after the harvest, and offered to sell them a voucher for fertilizer, at the regular price, with free delivery later in the season. The farmer had to decide during the visit whether or not to participate in the program, and could buy any amount of fertilizer.
    2. SAFI with ex ante Choice of Timing: An ICS officer visited the farmers before the harvest and offered them the opportunity to decide when, during the next growing season, they wanted the officer to return to offer them the SAFI program. They were then visited at the specified time, and offered a chance to buy a voucher for future fertilizer use (as in the Basic SAFI program, as described above).
    3. Free Delivery Visit Later in the Season: Same as SAFI program, but farmers were visited later in the season. An ICS officer visited farmers 2-4 months after the harvest (when it is time to apply fertilizer as a top-dressing to the next crop), and offered them the opportunity to buy fertilizer, at the regular price, with free delivery. This program was identical to SAFI, except that it was offered later.
    4. Subsidy Later in Season: An ICS officer visited the farmers 2-4 months after the harvest (when it is time to apply fertilizer to the next crop) and offered to sell them fertilizer, at a 50% subsidy, with free delivery.


    Results and Policy Lessons:

    The SAFI program was very popular. The basic SAFI was offered in two seasons. In the first season, the program increased usage by 14 percentage points, on a base of 23 percentage points. In the second season, the increase was even bigger, increasing usage by 18 percentage points. SAFI with ex ante timing choice was also successful, increasing usage by 22 points.

    These effects are comparable to those obtained from a 50% subsidy offered later in the season: the subsidy increased usage by 14 points. They are also larger than an undiscounted offer later in the season: the free delivery visit later in the season had no significant effect on usage.

    Consistent with a savings problem, enrollment in the SAFI program did not cause farmers to use fertilizer in subsequent seasons (as would be predicted if farmers were learning about fertilizer through the program). This suggests that it was the lack of commitment mechanism that was preventing farmers from purchasing and using fertilizer.

    Overall, the results suggest that offering farmers small, time-limited discounts on fertilizer may substantially increase usage without inducing overuse among farmers who are already using fertilizer, at relatively low cost.

    1 Shahua Chen and Martin Ravallion (2008). “The Developing World Is Poorer Than We Thought, But No Less Successful in the Fight against Poverty,” World Bank Policy Research Working Paper #4703.
    2 National Coordinating Agency for Population and Development (NCAPD) [Kenya], Ministry of Health (MOH), Central Bureau of Statistics (CBS), ORC Macro. 2005. “Kenya Service Provision Assessment Survey 2004”. Nairobi, Kenya: National Coordinating Agency for Population and Development.

    Balancing Health Benefits and Risks of ACT Subsidies for Africa

    Policy Issues:

    Malaria is one of the world’s foremost public health concerns, killing close to 1 million people every year.  In many malaria-endemic regions, resistance has developed to all but one class of antimalarial drugs, called artemisinin combination therapies (ACTs). ACTs sold in the retail sector are unaffordable for the poor, and although heavy subsidies can make them accessible, the benefits of treating more people and lowering transmission rates must be balanced against the risk of overtreatment, which can hasten the development of drug resistance. A new malaria testing technology, the rapid diagnostic test (RDT), has made it possible to perform malaria testing in the retail sector, but these tests are not demanded at prevailing prices. How can policymakers tailor the prices of subsidized drugs and diagnostic tests to target those truly sick with malaria and prevent those who do not need ACTs from taking them?

    Context of the Evaluation:

    In Kenya, ACTs are now the only effective class of antimalarial drugs. The incidence of malaria in Western Kenya is very high, with nearly 70 percent of households self-reporting an episode of malaria in the month before baseline. ACTs at government health centers in Kenya are nominally free, but health centers feature long waits and limited hours, are often stocked out of medication, and the remotely-located poor often cannot afford to travel the distance to get there. Consequently, many people opt to purchase cheaper, less effective anti-malarials over-the-counter at drug shops located closer to home.

    The Affordable Medicines Facility for malaria (AMFm) is an initiative currently being considered by major international aid agencies in which ACTs would be heavily subsidized to first line buyers throughout Africa (the subsidy policy is currently being piloted in 8 countries). But such a high subsidy could lead to overtreatment.  If people who do not have malaria presumptively buy ACTs without a diagnosis, this can contribute to drug-resistance, waste subsidy money and delay appropriate treatment for the true cause of illness.  Improving diagnostic access could considerably reduce overtreatment, but there has been little research so far in this area.

    Description of Intervention:

    The intervention took place in the districts of Busia, Mumias and Samia in Western Kenya. Researchers distributed vouchers redeemable at local drug shops to all households within four kilometers of four rural market centers. The households were randomly assigned to one of three groups:



    I. No subsidy

    II. ACT Subsidy Only

    III. ACT and RDT Subsidy

    Households received a voucher to buy ACTs at the market price of 500 Ksh (about US$6.25)

    Households received a voucher for ACTs giving them either an 80 percent, 88 percent, or 92 percent subsidy.

    Households received an ACT voucher as in group II, and also received an RDT voucher with either an 85 percent or 100 percent subsidy.

    In order to find out what fraction of people buying ACTs were truly malaria positive—a proxy for how well the ACT subsidy under the AMFm would be targeted—the   researchers also selected a random subset of the households in treatment II and III to receive the offer of a “surprise” free RDT after they completed their transaction at the drug shop.

    Trained study officers were posted at each of the four participating drug shops during opening hours every day throughout the study period. When a household member came into a drug shop to redeem his or her voucher,  study officers recorded details such as medicines bought, symptoms, patient characteristics, and true malaria status in case an RDT was administered.  

    Results and Policy Lessons:

    An ACT subsidy increases access, especially for the poorest households (measured by the literacy status of the household head). Without subsidies, literate-headed households take ACTs for 37 percent of illness episodes as compared to 11 percent of episodes in illiterate-headed households. With subsidies, the coverage rates become 45 percent and 38 percent, respectively.

    A lower subsidy level improved targeting to malaria-positive individuals without compromising access to ACTs among those who need them. At the two lower subsidy levels (80 and 88 percent), households were less likely to use an ACT voucher for adults, but no less likely to use a voucher for young children, who are much more likely to actually have malaria and for whom malaria is most dangerous. As a result, targeting improved.

    At the two lower subsidy levels, about 75 percent of patients for whom care was sought at the drug shop tested positive for malaria, while at the 92 percent subsidy level only 56 percent tested positive. Households were very willing to take RDTs, even when asked to share some of the cost: when the test was available, over 80 percent of the households who sought treatment at the drug shop chose to take an RDT before deciding what medication to purchase. However, many chose to buy an ACT even if they tested negative.

    This evaluation suggests that an information or marketing campaign about the reliability of rapid diagnostic tests might help convince people to use and comply with them – but even without such a campaign, moving from the target AMFm subsidized price (92 percent of the retail cost) to an 80 percent ACT subsidy with RDTs could increase the share of ACT takers who are malaria positive at the drug shop by 24 percentage points.

    While these results suggest that a slightly lower ACT subsidy than the one proposed by the AMFm would improve targeting without compromising access, the results also make it clear that a large ACT subsidy is needed in order to increase access among the neediest. Taking some of the planned ACT subsidy money away from ACTs and putting it towards subsidizing and promoting RDTs could improve targeting and be particularly effective among adults, especially if adherence to test results can be improved.

    Kenyan Life Panel Survey

    The Kenyan Life Panel Survey (KLPS) builds on an existing longitudinal dataset of educational, health and demographic information for approximately 6800 pupils in Western Kenya collected from 1998-2003, and extends it for another 6 years. In particular KLPS seeks to examine the long-run impact of a recent school-based health program - the Primary School Deworming Project - which provided free treatment for intestinal helminthes (worms) to pupils in 75 rural primary schools phased in over 5 years.  The project found that deworming had significant health and nutritional impacts, as well as leading to dramatic gains in school attendance and enrollment. After five years, educational attainment was significantly higher among early treatment school children. Evidence from KLPS linking child health gains (from deworming) and adult human capital formation could be used to justify increased investment in child health and nutrition programs.

    The KLPS tracks individuals throughout Kenya using a rigorous two stage tracking system. During the first round of household survey data collection, IPA made direct contact with nearly 85% of target individuals. Round 2 survey data collection is currently underway.

    Decentralization: A Cautionary Tale - Public Finance in Kenya

    Kenya’s education system blends substantial centralization with elements of local control and school choice.  This project looks at the system of incentives created by elements of decentralization.

    Syndicate content
    Copyright 2014 Innovations for Poverty Action. All rights reserved.