89124 S pe ci Af ResearchDigest al ric Is a World Bank su e on VOLUME 8 L NUMBER 3/4 L SPRING/SUMMER 2014 Tackling Malaria among Schoolchildren in Kenya Can intermittent screening and of infection status), delivered through IN THIS ISSUE treatment for malaria improve the schools, can improve the health and Tackling Malaria among Schoolchildren in health and education of children? cognitive function of schoolchildren in Kenya … page 1 intense and seasonal transmission set- What is the best way to control malaria I tings. But the recent withdrawal of the among school-age children? A recent study n many countries where malaria is primary drugs for IPT, sulphadoxine- evaluated a school-based strategy in Kenya endemic, successful control pro- pyrimethamine (SP) and amodiaquine grams have led to a decline in trans- (AQ), in many East African countries Are Small Farmers More Productive in mission. This has prompted a renewed precluded further investigation of IPT Rwanda? … page 2 emphasis on eliminating malaria, using SP+AQ. Where small farmers are more productive, leading to a shift in focus from case In Kenya the most recent National understanding the reasons why is important management of clinical malaria and Malaria Strategy (2009–17) identified for rural welfare policy interventions targeting traditionally testing and treating children with Postharvest Loss in Africa—What Do high-risk groups (children under five malaria parasitemia as a potential Farmers Say? … page 3 and pregnant women) to a more inclu- alternative strategy. Researchers A new study looks at farmers’ own reports sive approach aimed at interrupting from the London School of Hygiene of postharvest food losses—and at what they community-wide transmission by also & Tropical Medicine, Kenya Medical mean for ways to reduce the losses identifying and treating asymptomatic Research Institute, Harvard University, malaria parasitemia. and the Ministries of Health and Explaining Gender Gaps in Farm Productivity This shift has highlighted additional Education in Kenya evaluated this in Nigeria … page 4 vulnerable groups such as school-age policy through a study in a low- to Where female farmers have lower yields than children, who have some of the highest moderate-transmission setting on the male farmers, could equal access to inputs age-specific parasite rates and thus are south coast of Kenya in 2010–12. close the gender gap? important contributors to transmission. A recent paper by Halliday and co- In addition, success in lowering trans- authors presents the findings of the The Challenge of Measuring Hunger … page 5 mission levels will result in children study, a cluster randomized trial evalu- acquiring immunity later in life than in ating the impact of intermittent screen- Different ways of collecting data yield vastly the past, and the incidence of malaria ing and treatment (IST) for malaria on different estimates of the prevalence of hunger can be expected to increase in school- the health and education of schoolchil- Moving Up a Gear: Structural Change in age children. Yet school-age children dren. The school-based intervention Ethiopia … page 6 have persistently had the lowest cover- consisted of screening children for ma- An analysis finds emerging signs that age by interventions to prevent malaria, laria parasitemia using rapid diagnostic meaningful structural change is taking place and until recently there was limited tests (RDTs) once a term, and treating in Ethiopia’s economy evidence about the best approach to those found to be RDT-positive, with controlling malaria in this group. or without symptoms, with a full regi- Transactional Sex as Risk-Coping School health programs provide men of artemether-lumefantrine (AL). Behavior … page 7 a logical and affordable platform for The 24-month study enrolled 5,233 What are the health implications of tackling the malaria burden among children between ages 5 and 20, from transactional sex? And what conditions lead school-age children. Studies in West 101 government primary schools. Half women to enter the market for transactional and East Africa have shown that inter- the schools were randomly selected to sex? mittent preventive treatment (IPT) for receive the IST intervention; this was malaria (periodic mass administration of a therapeutic antimalarial regardless (continued on page 8) 2 World Bank ResearchDigest Are Small Farmers More Productive in Rwanda? Labor market imperfections seem needs with traditional technology. This controlling for land quality, yields, to be a key reason for the inverse prompted the government to put in labor intensity, and shadow profits per relationship between farm size and place a national land policy promot- hectare are all much higher on small ing land use planning, consolidation farms; and that profit per hectare (with productivity in Rwanda of land into “economic” plot sizes, and labor valued at market rates) is virtu- W prohibitions on subdivision. ally identical across holding and plot hile an inverse relationship Such measures are not uncontro- sizes. Results thus point to labor mar- between a farm’s size and its versial and have proved to be difficult ket imperfections as a major reason for productivity is a recurrent to implement in other settings. In a the inverse relationship between farm empirical finding, different explana- recent paper Ali and Deininger under- size and productivity, but suggest that tions are proposed in the literature. take an empirical investigation of the with existing market imperfections, One is that small farmers apply more underlying assumptions using nation- small farms are able to gainfully ab- than the optimum amount of inputs, ally representative plot-level data that sorb large amounts of labor. possibly because of factor market allow them to control for household- As long as farmers’ labor use re- imperfections. Another is a failure to specific heterogeneity. sponds to price signals, land market adequately measure key factors, espe- Descriptive statistics by tercile of interventions such as restrictions on cially land area or quality. Empirical the farm size distribution reveal three subdivision or involuntary consolida- evidence from Africa remains particu- regularities. First, plot (and farm) size tion programs may thus yield few ben- larly ambiguous. is inversely related to land quality; that efits and could even be counterproduc- But as countries seek to modern- is, smaller farms and plots have higher tive. Efforts to reduce labor market ize agriculture and transition from land quality and are less likely to be imperfections, and nonagricultural a subsistence-based economy, the affected by crop shocks. Second, differ- growth that leads to higher wages and answer matters: If small farms are ef- ences in output per hectare and input nonagricultural employment opportu- ficient, policy should focus on attract- use intensity across farm size classes nities pulling labor out of agriculture, ing upstream investment (such as in are pronounced (figure 1): output value may be more effective tools to improve agroprocessing) and link smallholders per hectare for farms in the bottom ter- rural welfare. to markets. If they are not, a strategy cile ($860) is almost three times that aimed at leapfrogging to large-scale of those in the top tercile ($298), with farming may be more desirable, along differences even more pronounced at with a regulatory environment that the plot level (from $1,296 to $317). discourages further subdivision and But third, for profit per hectare based promotes land consolidation. on actual input costs and labor valued In Rwanda, Africa’s most densely at market wages, the inverse relation- populated country, fragmentation ship between size and productivity es- Daniel Ali and Klaus Deininger. 2014. “Is There and small farm sizes are considered sentially disappears (figure 2). a Farm-Size Productivity Relationship in African key policy issues. Average farm size is Empirical results suggest that Agriculture? Evidence from Rwanda.” Policy Re- only 0.72 hectares in four parcels, not technology is characterized by con- search Working Paper 6770, World Bank, Wash- enough to satisfy even subsistence stant returns to scale; that even after ington, DC. Figure 2. Net Profit Per Hectare at the Holding Level, with Family Figure 1. Yield Per Hectare at the Plot and Holding Level Labor Valued at Market and Shadow Wages ln(value of output per hectare) Net profit in US$ thousands per hectare 8 2 1 6 0 4 −1 2 −2 −8 −6 −4 −2 0 2 4 −6 −4 −2 0 2 4 ln(land size in hectares) ln(land size in hectares) Plot level Holding level 95% confidence interval Market wage Shadow wage 95% confidence interval World Bank Research Digest 3 Postharvest Loss in Africa—What Do Farmers Say? Global estimates of postharvest farmer-reported estimates. They obtain handling and storage loss estimate for food losses are alarmingly high. But these from recent nationally repre- cereals in Sub-Saharan Africa of 8 per- what do farmers say about their own sentative household surveys in three cent and much lower than the APHLIS African countries—Malawi, Tanzania estimates of 14–18 percent for maize losses? (two years), and Uganda. They focus on (all postharvest loss before marketing T the reported share of harvested maize but without processing). he Food and Agriculture lost. They also identify the key agro- Multivariate analysis of the 2008 Organization (FAO) estimates climatic and socioeconomic drivers Tanzanian experience further shows that 32 percent of global food of postharvest loss to better under- that economic incentives, especially production is lost after harvest—and stand the factors affecting adoption the seasonal price gap (and access up to 37 percent in Sub-Saharan of improved storage and postharvest to the market), substantially reduce Africa. Why have farmers “tolerated” handling techniques. Success stories postharvest loss. Climatic factors such losses, and why has the inter- in promoting improved on-farm stor- (particularly the combination of heat national community not acted more age technologies have been rare in and humidity) substantially increase forcefully to reduce them? One answer Africa, with interventions too often it. Households’ wealth or poverty sta- holds that postharvest loss may not poorly aligned with farmers’ economic tus does not appear to be associated be that high. More than 30 years ago, incentives. with postharvest loss, and loss tends in the aftermath of the early 1970s The authors’ approach in esti- to be lower among female-headed world food crisis, Michael Lipton al- mating postharvest loss differs from households and those whose heads ready questioned the premise of high the FAO estimates as well as those have postprimary education. Some of postharvest loss and put on-farm from the African Postharvest Losses these factors probably work through grain losses among smallholders in Information System (APHLIS). The first the adoption of improved storage developing countries in the 5–8 per- are based on national food balances technologies, which remains limited. cent range. and loss fractions defined by experts, Between 0.6 percent (Uganda) and 11.5 Estimating postharvest loss is the second on national extrapolations percent (Tanzania) of maize farmers complex. The overall FAO estimates from purposively sampled (and often use improved storage technologies. cover all crops (including more per- older) in-depth case studies. The au- The prevalent postharvest treatment ishable roots, tubers, and fruits and thors’ approach has some advantages. method remains smoking or spraying. vegetables) and comprise losses dur- First, the use of nationally representa- The authors conclude that there ing all steps of the food supply chain tive samples avoids overestimation must be proper contextualization of (harvesting, on-farm handling and from sample selection bias. Second, the widely cited high and aggregate storage, processing, marketing, con- harmonization in the survey method- estimates of postharvest loss, with sumption). When confined to cereals ology facilitates comparison across farmer-reported estimates arguably and postharvest handling and storage countries. Third, while subjective, and more relevant indicators of demand loss only, the FAO estimate for Sub- thus prone to measurement error, self- for better storage and postharvest Saharan Africa is 8 percent. This esti- reported loss estimates are also more handling techniques. They argue that mate is definitionally more equivalent likely to reveal the losses that mat- interventions should be incentive com- and quantitatively closer to Lipton’s ter—and more likely to be incentive patible and carefully targeted, not “one numbers. compatible. Finally, the survey design size fits all.” And they call for wider use Nonetheless, the estimates are was exploited to obtain annualized of nationally representative surveys in necessarily based on many assump- loss estimates. studying postharvest loss. Nationally tions. In a new paper Kaminski and Among maize farmers, only be- representative estimates help eluci- Christiaensen complement the efforts tween 7 percent (Malawi) and 22 per- date the granularity in postharvest loss to quantify postharvest loss using cent (Uganda) report on-farm posthar- and storage behavior; they can also vest loss, averag- be used to fine-tune the algorithms Figure 1. Estimated Postharvest Loss for Maize in Three African Countries ing 21–27 percent underpinning postharvest loss infor- Percent of their harvest. mation systems such as APHLIS and to 25 This adds up to help update their annual estimates. Share of national harvest lost 20 Share of farmers reporting 1.4–5.9 percent of postharvest loss the national maize 15 harvest being lost 10 on-farm (figure 1). Jonathan Kaminski and Luc Christiaensen. 2014. 5 This is still quite a “Post-Harvest Loss in Sub-Saharan Africa— 0 bit lower than the What Do Farmers Say?” Policy Research Working Uganda (2009/10) Tanzania (2010/11) Malawi (2010/11) FAO postharvest Paper 6831, World Bank, Washington, DC. 4 World Bank ResearchDigest Explaining Gender Gaps in Farm Productivity in Nigeria In northern Nigeria women’s farm are able to decompose the gender gap were given similar levels of inputs, productivity is lower than men’s. into the part explained by differences the gender gap would diminish. Thus In southern Nigeria it isn’t. What in the level of inputs (for example, how providing additional inputs to female much differences in hours of labor farmers would both benefit the women accounts for the difference? contribute to the gap) and the part and increase overall productivity in the T explained by differences in returns to region. here is a general perception that the same inputs (for example, whether The findings suggest that when it across Sub-Saharan Africa female an hour of labor produces more or less comes to gender gaps in agricultural farmers have lower yields than on a woman’s farm than on a man’s). productivity, northern Nigeria looks a male farmers. But the reality is more The analysis is conducted separately lot like other countries in Sub-Saharan nuanced. While some studies find low- for the north and south because of the Africa while southern Nigeria is an ex- er average productivity among female socioeconomic diversity in the country. ception. This stark difference between farmers than among male farmers, In northern Nigeria analysis con- the two regions suggests that policies others find no significant differences trolling for input levels shows that need to take into account differences between the two groups. And in many female farmers produce 27 percent less in norms, markets, and institutions instances “back of the envelope” calcu- than their male counterparts (figure more broadly in order to increase lations show that if women had equal 1). This gender gap in productivity is the productivity of Nigeria’s female access to land and productive inputs, driven primarily by the fact that women farmers. the gender gap would almost always have lower levels of labor and fertil- disappear. izer and that they get lower returns to This suggests that women are not being older farmers and having more less capable farmers but instead that people in their household. Improving they face constraints that limit their access to fertilizer and hired labor and productivity. In Sub-Saharan Africa, promoting cash crops for women in where smallholders’ agricultural pro- the north would probably boost overall ductivity is low, reducing the gender agricultural growth, though it might gap by increasing female farmers’ not close the gender gap. Even if fe- productivity could increase overall pro- male farmers in the north were given ductivity, leading to higher incomes, the same level of inputs as their male lower poverty, and better food security counterparts, some differences in pro- and nutritional outcomes. ductivity might still persist because of In a new paper Oseni, Corral, the differences in returns. Goldstein, and Winters use a national- In the south, however, analysis con- ly representative data set for Nigeria to trolling for key characteristics and fac- investigate whether there is a gender tors of production (including inputs) Gbemisola Oseni, Paul Corral, Markus Goldstein, gap in agricultural productivity—and, finds no gender gap in productivity. and Paul Winters. 2014. “Explaining Gender if so, which factors in the production Male and female farmers in the re- Differentials in Agricultural Production in Nige- process appear to be driving it. Using gion appear to obtain similar returns ria.” Policy Research Working Paper 6809, World decomposition methods, the authors to factors of production—and if they Bank, Washington, DC. Figure 1. Difference between Male and Female Farmers’ Productivity in Selected Areas of Sub-Saharan Africa, 2009–12 Percent 30 27.4 26.8 20 16.6 14.3 10 4.5 0 0 Southern Nigeria Malawi Tanzania Uganda Niger Northern Nigeria Source: For Nigeria, Gbemisola Oseni and others, “Explaining Gender Differentials in Agricultural Production in Nigeria” (Policy Research Working Paper 6809, World Bank, Washington, DC, 2014); for other countries, World Bank and ONE Campaign, Levelling the Field: Improving Opportunities for Women Farmers in Africa (Washington, DC: World Bank Group, 2014). Note: Data are for the latest year available during the period shown. World Bank Research Digest 5 The Challenge of Measuring Hunger Estimates of the number of hungry commodities—which can range from differ as the hunger indicator interacts people in the world will remain a handful of aggregate commodities with household size, wealth, educa- unreliable until the design of to more than 400 relatively detailed tion, and other measures. For example, commodities. relative to the “gold standard,” recall household surveys is improved A recent paper by De Weerdt, surveys underestimate hunger as a T Beegle, Friedman, and Gibson studies household grows richer but overesti- here is much interest in counting the implications of survey design for mate hunger as the household increas- the number of hungry people in the measurement of hunger. One key es in size. the world and monitoring trends finding is that estimates of hunger are The global hunger estimates de- in hunger. Many observers follow even more sensitive to survey design rived from the FBS method are also global numbers on hunger prevalence, than are estimates of overall consump- subject to the vagaries of cross-country not least because halving the propor- tion (and therefore of poverty) because survey design. In this method the vari- tion of hungry people between 1990 most of the design differences in glob- ance of calorie consumption is mea- and 2015 is among the prominent al practice are due to differences in the sured through HCES while the mean Millennium Development Goals (and measurement of food consumption. is taken from the national FBS. There one on which achievements appear to One review of household surveys from are many reasons to question the ac- be slight). more than 100 countries found great curacy of the balance sheets, including But measuring the number of hun- variation in the survey mode (diary or the degree to which they capture na- gry people is not easy. Current global recall), in the length of the food item tional postharvest stores and losses as counts rely on combining statistics on list, and in the recall period used. The well as the accuracy of their root crop each country’s food availability from measurement of nonfood consumption yields. While the paper does not ad- national food balance sheets (FBS) varies much less around the world. dress these issues, it is clear that vari- with information on consumption pat- Using a survey experiment in ation in survey design greatly affects terns from household consumption Tanzania, the paper explores the net the estimates of hunger from the FBS expenditure surveys (HCES). Recent effect of reporting error arising from method: estimates of the prevalence of research has advocated estimating seven different and very common con- hunger range from 17 to 28 percent. hunger directly from these house- sumption survey designs, including The study shows that changing the hold surveys. In both the standard one taken as a proxy for the “gold stan- design of the survey questionnaire approach (the FBS method) and the dard”—an intensively supervised per- (that is, how questions are asked) approach relying solely on household sonal diary with detailed cross-checks greatly affects the estimate of hunger surveys (the HCES method), the use of across household members. The study prevalence—in Tanzania, by millions of household survey data is fundamental. is a randomized within-village experi- people—regardless of which of the two The FBS method uses the variance in ment with good covariate balance that approaches is used. As a consequence, calorie consumption estimated across extends over an entire calendar year, comparable and valid estimates of households through the survey, while so the results should not be subject the number of hungry people in the the HCES method measures hunger to seasonal or community-level con- world will remain out of reach until directly, as a function of the observed founders. The only differences in the improvements are made in the design total household food consumption in numbers of hungry people should of household surveys, especially the relation to estimated household calor- be due to the particular survey de- consistency of questionnaires. ic need. So the quality and consistency sign administered. And these differ- of household surveys matter in both ences are large, both statistically and approaches. economically. One challenge to these approaches, The estimated daily per capita however, is that the design of house- kilocalorie intake ranges from 1,793 to hold consumption surveys varies 2,677, depending on the survey mod- widely around the world in a number ule. As a consequence, estimates of of dimensions, including these: the prevalence of hunger range from 19 • The method of data capture— to 68 percent. This wide range trans- which is typically either diary or a re- lates into a difference of more than 23 call questionnaire. million people in Tanzania, a country • The reference period over which with a population of 45 million. consumption is measured—which can And survey design influences not Joachim De Weerdt, Kathleen Beegle, Jed Fried- vary from one day to one week to one only prevalence estimates but also the man, and John Gibson. 2014. “The Challenge of month. targeting of the hungry. The hunger Measuring Hunger.” Policy Research Working • The degree of detail in profiles suggested by each module Paper 6736, World Bank, Washington, DC. 6 World Bank ResearchDigest Moving Up a Gear: Structural Change in Ethiopia What is driving Ethiopia’s strong and demographic change to growth in result of within-sector improvements. growth? Mostly labor productivity output per capita. Labor productivity Changes in employment accounted within sectors, but structural change is measured as output per worker, and for about 10 percent of the growth, its impact is further decomposed into while the contribution of demographic is accelerating within-sector and intersectoral contri- change was negligible. I butions, with the latter representing Nonetheless, recent trends are t is a historical inevitability. All structural change. promising. The relative contribution successful developing countries Results show that the composi- of intersectoral shifts has been grow- have undergone a fundamental and tion of Ethiopia’s output has changed ing in recent years and thus playing an potent process of structural change. significantly over the past 20 years. increasingly important role in boosting This transformation entails a realloca- Agriculture remains a very large pro- output per capita. As workers move tion of labor from low-productivity duction sector, but its share in total out of agriculture, the sector with the sectors to more dynamic economic output has steadily declined, from 66 lowest productivity, this creates a posi- activities—typically from subsistence percent in 1991 to 45 percent in 2011. tive momentum for growth-enhancing agriculture to industry and modern Services have expanded considerably, structural change. While productivity services. Through its impact on labor especially trade and real estate. Yet growth in agriculture and trade has productivity, structural change plays changes in the sectoral distribution of provided a considerable boost to total an instrumental role in sustaining employment have lagged behind. The labor productivity, structural change is economic growth, generating produc- share of agriculture in total employ- becoming another key source of eco- tive employment, and raising living ment has declined in recent years, but nomic dynamism (figure 1). standards. remains very high at about 78 percent. The recent decline in the employ- In many African economies, how- Trade currently accounts for 9 percent ment rate has adversely affected out- ever, shifts in the structure of output of total employment—and “other ser- put per capita growth. But it is due in have rarely been followed by commen- vices” (which includes public adminis- part to younger people staying longer surate changes in sectoral employ- tration, education, financial interme- in school—which can be seen as a ment. This undermines the impact and diation, and real estate) for 8 percent. positive development in the long run sustainability of the growth process. Labor productivity growth has been as students acquire skills that can What has been the recent experience strong across most sectors—though support future growth. There is also in Ethiopia, where growth has aver- with substantial variations—as total evidence that Ethiopia is starting to aged about 10 percent a year since output per worker doubled in the past benefit from a demographic dividend, 2003? decade. But large productivity gaps re- which accounted for 10 percent of out- A recent paper by Martins inves- main. The output per worker in mining put per capita growth in the 2005–11 tigates the key drivers of this remark- and quarrying, electricity and water, period. The country’s dependency ratio able performance. It examines how and transport and communications is has been declining, which provides a the composition of output and em- more than 10 times that in agriculture. boost to economic activity. ployment has changed through time The decomposition analysis shows The analysis thus finds emerg- and assesses trends in sectoral labor that labor productivity has been the ing signs that meaningful structural productivity. More importantly, it es- main contributor to output per capita change is taking place in Ethiopia. timates the relative contributions of growth: it accounted for about 90 per- While these changes have not reached labor productivity, employment rates, cent during 1996–2011, mainly as a the levels of transformation experi- enced by many East Asian countries, Figure 1. Sector Contribution to Labor Productivity Growth in Ethiopia, 1999–2011 they are certainly motivating. The key Percent challenge for policy makers is to de- 50 vise and implement policies that can 1999–2005 40 further encourage the modernization 2005–11 and transformation of the Ethiopian 30 economy. 20 10 0 Pedro Martins. 2014. “Structural Change in −10 Agriculture Mining Manufactur- Electricity Construction Trade Transport & Other Intersectoral Ethiopia: An Employment Perspective.” Policy & quarrying ing & water communications services shift Research Working Paper 6749, World Bank, −20 Washington, DC. World Bank Research Digest 7 Transactional Sex as Risk-Coping Behavior Transactional sex rises after shocks, A recent paper by de Walque, Dow, shocks and STIs is strongest, they find suggesting that access to insurance and Gong uses a unique set of data that shocks lead to an almost threefold or savings may have public health that measure sexual behavior, STIs, and increase in paid sex (sex in exchange transfers in a sample of women who are for cash or gifts). Finally, using reported implications not sex workers and are representative income data, the authors estimate that H of women in rural Tanzania. The analy- as income goes down the risk of STIs ow poor people in low-income sis is based on a panel of women in goes up. For unmarried women, they countries respond to the myriad rural Tanzania involved in a conditional also find that as income goes down risks they face is a crucial ques- cash transfer study aimed at the pre- paid sex goes up. tion. Their risk-coping behaviors can vention of HIV and STIs. Four rounds This work contributes to the large have long-term adverse consequences of data, spaced four months apart, body of literature documenting the for both human and health capital, were collected. At each round individu- effects of negative shocks on health which has implications for the persis- als were tested for four curable STIs outcomes and the relationships be- tence of poverty. Transactional sex, the (chlamydia, gonorrhea, trichomoniasis, tween income and disease. The study exchange of money or gifts for sexual and Mycoplasma genitalium), which are also provides additional evidence that relations, is believed to be one means used as the main outcome of interest. transactional sex is not limited to sex that women use to cope with risk; Compared with self-reported sexual workers but may be seen as a common it also leads to greater exposure to behaviors, these biomarkers have the risk-coping mechanism for a much sexually transmitted infections (STIs), advantage of not being subject to self- larger population. Understanding the including HIV. While prostitution and reporting bias. circumstances in which transactional sugar daddy relationships (relation- The authors estimate a relation- sex occurs and the scope of this behav- ships between younger women and ship between household-level negative ior has important policy implications. wealthier older men) are most com- shocks and sexual behavior incorporat- If transactional sex is being used as a monly associated with transactional ing individual and time fixed effects. risk-coping mechanism, then providing sex, these types of relationships have They find that women experiencing a women with access to formal insurance also been documented among a negative shock are 5 percentage points or savings may have important public broader population, including older more likely to be infected with an STI. health implications. married women. The magnitude of this increase is both Two questions have arisen from the significant and large, corresponding to literature: What are the health implica- a 36 percent increase in STI risk over tions of transactional sex? And what a four-month period. In addition, they conditions lead women to enter the find suggestive evidence that this effect market for transactional sex? Because is stronger among unmarried women transactional sex is believed to be a and those with the lowest socioeco- leading contributor to the HIV/AIDS nomic status. epidemic in Sub-Saharan Africa, an- The relationship established be- swers to these questions are of both tween shocks and STIs does not neces- scientific and policy interest. sarily mean that women are responding Documenting transactional sex is to shocks by engaging in transactional difficult. It involves accurate measures sex. For example, women experiencing of both sexual behavior and transfers shocks may be unable to afford medi- made from partners. Many of the lead- cal treatment for STIs. They may also ing studies focus on commercial sex have compromised immune systems workers, who are more comfortable that make them more susceptible to discussing their sexual behavior; other STIs. In both cases it is possible that women may be less forthcoming. In ad- a woman’s likelihood of having an STI dition, while transfers between clients could increase even if her sexual activ- and sex workers are typically made at ity remains unchanged. the time of the sexual act, for women The authors therefore present cor- who are not sex workers the timing of roborating evidence using data on self- transfers may not correspond with the reported sexual behaviors. They find Damien de Walque, William H. Dow, and Erick act. For example, women may have a that women experiencing shocks are 12 Gong. 2014. “Coping with Risk: The Effects of sexual occurrence with a friend, but percent more likely to have unprotected Shocks on Reproductive Health and Transactional receive a transfer from this partner days sex. In addition, for unmarried women, Sex in Rural Tanzania.” Policy Research Working or weeks later. for whom the relationship between Paper 6751, World Bank, Washington, DC. 8 World Bank ResearchDigest (continued from page 1) delivered to randomly selected children infections may also have contributed to Recent Policy Research from grades 1 and 5. Anemia, malaria the lack of impact. Working Papers parasitemia, classroom attention, and The results suggest that school- educational achievement were mea- based IST for malaria should not 6739 Does Migration Foster Exports? Evidence sured at baseline and at two follow-up be introduced in low- to moderate- from Africa visits. transmission settings and that more Hélène Ehrhart, Maëlan Le Goff, Emmanuel Rocher, and Raju Jan Singh On average 17.5 percent of the chil- research is needed to identify interven- 6741 What Explains Rwanda’s Drop in Fertility dren in the intervention schools were tions that can be included in school between 2005 and 2010? Tom Bundervoet RDT-positive over the five screening health programs in different transmis- 6750 Economic Growth in Ghana: Determinants and rounds. The prevalences of anemia and sion settings. Encouragingly, however, Prospect Anna K. Raggl malaria parasitemia were similar in the the results highlight a potential role for 6764 Decomposition of Gender Differentials in intervention and control groups at the periodic screening of schoolchildren to Agricultural Productivity in Ethiopia Arturo Aguilar, Eliana Carranza, Markus 12- and 24-month follow-ups, and there identify pockets of high transmission Goldstein, and Talip Kilic was no difference between the two that can be targeted with focal control 6769 Credit Constraints, Agricultural Productivity, groups in classroom attention scores measures. and Rural Nonfarm Participation: Evidence from Rwanda at the 9- and 24-month follow-ups. The Daniel Ayalew Ali, Klaus Deininger, and Marguerite Duponchel IST intervention also had no effect on 6772 International Interventions to Build Social educational achievement in the older Capital: Evidence from a Field Experiment in class but, surprisingly, had an apparent Sudan Alexandra Avdeenko and Michael J. Gilligan negative effect on spelling and arithme- 6786 Mobile Money Services Development: The tic scores in the younger class. Cases of the Republic of Korea and Uganda Eva Gutierrez and Tony Choi These findings indicate that in this 6788 Customs, Brokers, and Informal Sectors: A low- to moderate-transmission setting, Katherine E. Halliday, George Okello, Elizabeth Cameroon Case Study Thomas Cantens, Jonathan Kaminski, Gaël IST as implemented in the study pro- L. Turner, Kiambo Njagi, Carlos Mcharo, Juddy Raballand, and Tchouawou Tchapa vided no health or education benefits Kengo, Elizabeth Allen, Margaret M. Dubeck, 6815 Implementation of REDD+ Mechanisms in Tanzania to schoolchildren. The reason may be Matthew C. H. Jukes, and Simon J. Brooker. Paula Cordero Salas that in this setting most of the children 2014. “Impact of Intermittent Screening and 6824 Designing Experiments to Measure Spillover screened did not require treatment Treatment for Malaria among School Children in Effects Kenya: A Cluster Randomized Trial.” Policy Re- Sarah Baird, Aislinn Bohren, Craig McIntosh, and those who did lived in focal, high- and Berk Özler search Working Paper 6791, World Bank, Wash- 6825 HIV Testing, Behavior Change, and the transmission regions where reinfection ington, DC. Also published in PLoS Medicine Transition to Adulthood in Malawi occurred between screening rounds. 11 (1). Kathleen Beegle, Michelle Poulin, and Gil Moreover, although children found to Shapira The work was supported by grants from the Ma- 6832 The Impact of an Adolescent Girls be infected were treated, a substantial Employment Program: The EPAG Project in laria Impact Evaluation Program under the World Liberia share of the school population and Bank’s Development Impact Evaluation Unit and Franck Adoho, Shubha Chakravarty, Dala T. wider community were untested and the World Bank–administered Spanish Impact Korkoyah Jr., Mattias Lundberg, and Afia untreated, contributing to reinfection. Tasneem Evaluation Fund and Bank-Netherlands Part- The limited post-treatment protection nership Program; the International Initiative for Working Papers can be downloaded at http://econ.worldbank.org period of AL and variable performance Impact Evaluation; and the Partnership for Child To download the World Bank Research E-Newsletter, of the RDTs in detecting low-intensity Development. go to http://econ.worldbank.org/research_newsletter The World Bank Research Digest is a quarterly publica- The Research Digest is financed by the Bank’s Editorial Committee: Indermit S. Gill (managing editor), tion disseminating findings of World Bank research. Research Committee and managed by DECRS, the Aslı Demirgüç-Kunt, and Shiva S. Makki. Editor: Alison The views and interpretations in the articles are those research support unit of the Development Economics Strong; production: Roula Yazigi. For information or of the authors and do not necessarily represent the Senior Vice Presidency (DEC). The Research Digest is free subscriptions, send email to research@worldbank. views of the World Bank, its Executive Directors, or the not copyrighted and may be reproduced with appropri- org or visit http://econ.worldbank.org/research_digest. countries they represent. ate source attribution. The World Bank 1818 H Street, NW Washington, DC 20433, USA Printed on Recycled Paper