Returns to Controlling a Neglected Tropical Disease: Schistosomiasis Control Programme and Education Outcomes in Nigeria

Returns to Controlling a Neglected Tropical Disease: Schistosomiasis Control Programme and... Abstract Using the rollout of the schistosomiasis campaign in Nigeria as a quasi-experiment, we examine the impact of the disease control programme on school-age children education outcomes. Schistosomiasis is a parasitic disease caused by infections from a small worm. Its most severe effects hamper growth and cognitive development of children. The mass campaign targeted four states that saw large reduction in the infectious disease afterwards. Using difference-in-differences strategy, we find that the cohort exposed to the treatment in rural areas accumulated an additional 0.6 years of education compared to cohort not exposed to the treatment. Moreover, the impact of the schistosomiasis treatment is mainly on girls residing in rural areas. 1. Introduction Schistosomiasis, a neglected tropical disease (NTD), is an acute and chronic disease caused by parasitic worms.1 According to the World Health Organization (WHO), at least 258 million people required treatment in 2014. The disease is particularly prevalent among school-age children. In general, repeated treatment helps reduce and prevent morbidity. We examine a deworming programme that provided regular treatment to at-risk population in one of the regions with the greatest burden of the disease. We do not find conclusive evidence that treatment increased school enrolment for 6 to 14-year-old children. However, we find that the cohort exposed to treatment in rural areas accumulated 0.6 years more of education. Moreover, most of the impact is driven by the impact on girl children. Resolving the problem of endemic disease is crucial to development in low-income countries. Studies in development economics provide evidence of the relationship between overall health environment and long-term changes in development outcomes. Although the role of disease reduction in affecting economic growth is found to be rather weak (Weil, 2010), global improvement in health conditions not only improved lives but also enhanced economic growth (Bloom et al., 2004; Weil, 2007). Albeit the debate on the empirical estimation is still unsettled, the implication for developing world appears to be of significance.2 Public policies directed toward the decline in prevalence, morbidity and mortality rates of infectious diseases not only have important regional implications but also implications at the individual level.3 Better health capital improves the individual return to investment in education and leads to higher future returns. For instance, Bleakley (2010) documents how and the extent to which exposure to endemic disease in early childhood development affects labour productivity in the adulthood. Cutler et al. (2010) review a malaria eradication programme in India and report a positive impact on the cohort’s expenditure and per capita household consumption. Venkataramani (2012) studies a nation-wide effort to eliminate malaria in Mexico in the 1950s. The author finds that, in the case for men, birth year exposure in treated states led to increases in cognitive test score performance and better schooling outcomes. The model, however, holds smaller and no significant estimates for women. Kazianga et al. (2014) find an increase in rural population and land transactions in Burkina Faso after a campaign against Onchoceriasis run by the WHO. Exposure to the Onchoceriasis control programme induced villages to develop and improve local institutions (e.g., land markets) and the provision of public goods (e.g., public markets, primary schools and telephone services). Bleakley (2007) provides evidence of the benefits of disease control on economic outcomes. He evaluates the impact of the eradication of hookworm, an infection similar to schistosomiasis, in the American South. Hookworm, an intestinal parasite, was responsible for half of the literacy rate gap between northern and southern states. After the programme was completed, it appeared that the disease accounted for as much as 20% of the difference in income between the north and the south. Other findings indicate that hookworm eradication had large positive effects on school enrolment and on years of education completed. Miguel and Kremer (2004) investigate the effects of deworming programmes in areas with high helminth infection rates of intestinal worms, including hookworm, roundworm and schistosomiasis in rural Kenya. The authors focus on school-age children and enrolment in school. They show evidence of substantial increases in school participation and reduced absenteeism in randomly selected treated schools. Although they found no evidence on pupils’ test scores, they argue that the deworming programme was extremely cost effective in terms of returns to education and human capital investments. Their cost-benefits analysis indicates that the programme has sizeable benefits. Our study exploits a large campaign of drug administration to school-age children that has reduced the prevalence of schistosomiasis in Nigeria. The disease causes anaemia, stunted growth, cognitive impairment and premature death. The drug distribution started in 1999 and initially covered two states (i.e., Plateau and Nasarawa). By the year 2010, the programme had expanded to include four states, and over a million children received the treatment each year. A substantial body of the medical literature has examined the effectiveness of the schistosomiasis mass treatment. Most of these studies that have so far investigated student performance response to the treatment have relied on cross sectional techniques comparing the outcomes of a control and a test group (Ekanem et al., 1994; Meremikwu et al., 2000; Ayoya et al., 2012). This paper contributes to and extends the literature on health and education externalities by reporting empirical results of the schistosomiasis control programme impact on children’s educational outcomes. Our estimation strategy uses difference-in-differences (DID) to show that efforts to eliminate the disease in developing region had positive impacts on children’s educational outcomes. By taking advantage of the expansion of the schistosomiasis control programme in four states of Nigeria from 1999 to 2013, we show that the reduction in the water-borne disease has contributed in improving the level of education of younger cohorts who were exposed to the programme. The remainder of the paper is organised as follows. Section 2 describes of the disease, the circumstances in which a child might get infected, and the symptoms of the infection. It also discusses the treatment campaign conducted by the Carter Center (CC) and the effectiveness of the programme. Section 3 describes the data employed. Section 4 sets out the model specification and the identification strategy used to capture the treatment impact. The findings of our study are presented in Section 5. The last section concludes. 2. Programme Description 2.1 Schistosomiasis Schistosomiasis is a water-borne disease that mostly affects children in tropical regions especially in developing countries of Africa, Asia and South America. The schistosome parasite, which is a worm, is acquired by contact with unprotected stagnant water. The classical sign of schistosomiasis infection is blood in the urine. Once in the blood vessels, the parasite attacks mainly the bladder and kidneys and causes fever and pain in the stomach and during urination. The most vulnerable group to the disease are school-age children. They are more likely to come in contact with the vector, and the disease exposes them to serious deficiency manifested through anaemia, inhibited growth, debility and high morbidity. Nigeria has one of the highest prevalence rates of schistosomiasis in the world. Abdulkadir et al. (2017) review a large number of articles in an attempt to estimate the prevalence of schistosomiasis in Nigeria. Among the analysed studies, they find the prevalence of urinary schistosomiasis to vary from 2 to 82.5%. Schistosomiasis is a concern for the developing world because it tends to be endemic in both infested rural places and densely populated urban areas as well. As with other NTDs, schistosomiasis can hinder development prospects if it is not adequately addressed. Although schistosomiasis is endemic in many regions, there are effective treatment regimens. The most common course of treatment consists of periodic intake of the drug praziquantel. The drug is highly effective as it can reverse up to 90% of the damage at a relatively low cost, between $0.15 and $0.20 per treatment (Hopkins et al., 2008). The WHO is active in raising awareness about schistosomiasis and recommends health education on access to safe water, improved sanitation and hygiene. Gutman et al. (2008) advocate for preventive mass treatment of all school-age children as the cheapest approach to control schistosomiasis. The authors note that the disease affects the ability of infected children ability to work, by weakening their bodies. Furthermore, infected children show signs of poor growth and learning difficulty at school. Agi and Okafor (2005) study the epidemiology of schistosomiasis and find a significant negative correlation between age and intensity of infection. Their results suggest that there is a progressive rise in prevalence for children aged 5–9, but the infection rate often peaks for children aged 10–14 and grows weaker as the individual gets older. Similarly, (Ezeadila, et al. 2015) find the difference in prevalence between children aged 6–9 and 10–13 to not be significant at the 5% level. The paper is connected to a strand of literature that relates the impact of a NTD to children’s educational outcomes. Early studies, such as (Ekanem et al., 1994), in a cross sectional analysis assess the effect of schistosoma infection in children aged 5–15 in southeastern Nigeria to find no significant impact on their physical growth and school performance. Meremikwu et al. (2000) also find no improvement in school attendance following repeated treatments of praziquantel to children aged 8–9 in Adim, Nigeria.4 Like Ekanem et al. (1994), Meremikwu et al. (2000) employ a simple difference comparing cohort pupils’ characteristics pre-treatment and post-treatment period without controlling for observed or unobserved factors within child, school or village that could influence the outcome. Ayoya et al. (2012) offer a stronger analysis, studying primary school children aged 7–12 in poor urban area in Bamako, Mali. Using a linear regression model, they find a significant increase in children’s attendance and school achievement, measured by the pupil’s passing rate. They acknowledge that they do not control for unobserved factors that could bias their estimates. By using DIDs, we can account for unobserved heterogeneity across time and space in estimating the effects of schistosomiasis control on children’s educational outcomes. 2.2 Programme intervention As part of the effort to control the disease, the CC has provided schistosomiasis health education and drug distribution in Nigeria since 1999 to children aged 5–14. Originally, the CC operated in Nigeria for the treatment of onchoceriasis. The CC eventually obtained a grant from Smithkline Beecham to begin the treatment of urinary schistosomiasis, the most severe form of the disease. In addition to that, the project was supported by a donation of 50,000 doses of praziquantel from Medochemie, Bayer and Shin Poong (Hopkins et al., 2002). First, the Federal Ministry of Health conducted a nation-wide postal survey in 1990, confirming the presence and the high prevalence of urinary schistosomiasis throughout the country. The rate of infection was particularly high among people living in poor rural areas with little or no access to a safe source of water. Of all age groups, children between 5 and 14 years who were traditionally responsible for water-related household chores were the most exposed and commonly affected. The CC mapping of the states of Plateau and Nasarawa confirmed the nation-wide mapping statements. For instance, in the village of Mungkohot in the Plateau state, a staggering 80% of school-age children were found infected with the disease. In the first year of the programme, the CC assessed the activity of 150 villages to not only determine where the infectious disease was located but also to identify communities’ habits and practices that were used in the preparation of health education material for the schistosomiasis campaign. (The Carter Center, 2000) The control programme at its pilot phase reached more than 8,000 people in highly affected villages in the states of Plateau and Nasarawa. The CC schistosomiasis programme, in association with the Federal Ministry of Health, initially launched its global assistance programme in two states, Plateau and Nasarawa. In 2004, the pilot programme expanded into all local government areas in these states and grew to include the state of Delta (2010) and Edo (2010). During the schistosomiasis campaign, the CC distributed treatment doses of praziquantel to the at-risk populations in the states of Plateau, Edo, Delta and Nasarawa (Figure 1). They received medicine for the severely debilitating form of the disease, the urinary schistosomiasis. In 2008, the CC received a donation of 1.1 million doses of praziquantel followed by another donation of 1.5 million doses in 2009 from the WHO and Merck. The contributions made in 2008 and 2009 to the programme surpassed the cumulative number of treatments from 1999 to 2007 (1.08 million treatments) to illustrate a significant expansion of the schistosomiasis control programme (Figure 2). Figure 1: View largeDownload slide Schistosomiasis Treatment States in Nigeria, 2013. Notes: Carter Center-Assisted Schistosomiasis Programme Treatment States in Nigeria. In 2014, the Treatment Initiative Was Extended to the States of Ebonyi and Enugu. Source: The Carter Center. Figure 1: View largeDownload slide Schistosomiasis Treatment States in Nigeria, 2013. Notes: Carter Center-Assisted Schistosomiasis Programme Treatment States in Nigeria. In 2014, the Treatment Initiative Was Extended to the States of Ebonyi and Enugu. Source: The Carter Center. Figure 2: View largeDownload slide Annual Praziquantel Treatments, 1999–2014. Notes: Carter Center-Assisted Schistosomiasis Programme Annual Praziquantel Treatments. Source: The Carter Center (Disease Data, updated in May 2015). Figure 2: View largeDownload slide Annual Praziquantel Treatments, 1999–2014. Notes: Carter Center-Assisted Schistosomiasis Programme Annual Praziquantel Treatments. Source: The Carter Center (Disease Data, updated in May 2015). The intervention as of today is believed to have considerably reduced schistosomiasis infection in the treated states. The CC reports that blood in school children’s urine has been reduced by approximately 94% in Plateau and Nasarawa states and approximately 88% in Delta state. The claim has been supported by several studies that documented a reduction in blood in urine in areas targeted by the CC (Agi and Okafor, 2005; Hopkins et al., 2002, 2008). This is pertinent to our study because we assume the treatment was effective in successfully controlling the disease. On the one hand, we have studies not related to the programme that collected urine samples among school-age children in the CC states and document a low prevalence of the disease (Ezeadila et al. 2015). On the other hand, there are findings of studies conducted in control states that concluded high prevalence of urinary schistosomiasis and the endemic state to the infection of many communities (Sam-Wobo et al., 2011; Babatunde et al., 2013). 3. Data To examine the effect of the programme, we use five rounds of the Demographic Health Surveys (DHS) on Nigeria collected in the years 1990, 1999, 2003, 2008 and 2013.5 The surveys initiated by the National Population Commission (partly funded through the United States Agency for International Development (USAID)) were developed to provide accurate information on maternal and child health but also family planning. Samples of the DHS are drawn to be nationally representative. Administratively, Nigeria is divided into states, local government areas, localities and enumeration areas (EA). The DHS clusters, or primary sampling unit, were based on census EA. Some clusters combined several EA because the DHS requires a minimum of 80 households for each cluster. The sample includes all 37 states, including Plateau, Edo, Delta and Nasarawa where most of the efforts against schistosomiasis by the CC have been concentrated. The CC worked closely with the Ministry of Health and, to the best of our knowledge, there were no other similar programmes before or concurrently with the CC intervention. Since the CC programme started in 1999, we treat the 1990 and the 1999 survey as pre-programme, while the 2004, 2008 and 2013 surveys serve as post-programme data. Because the programme was concentrated in four states of Nigeria, we group these four states (Plateau, Nasarawa, Delta and Edo) as treated states, while the other 33 states serve as comparison states.6 Moreover, because the main beneficiaries were school-going children, we restrict the data to individuals in 6–14 age group.7 Table 1 provides descriptive statistics for some key variables for individuals in the 6–14 age group by treatment status. School enrolment is a dummy variable to indicate whether the member reported attending school during the current school year. Enrolment in the treated states was larger compared to the non-treated states in both the pre-programme and post-programme period.8 Table 1: Summary Statistics, Pre- and Post- Programme Intervention (6–14, All Surveys) All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 Notes: Standard deviations are displayed in parentheses. Data are restricted to 6–14 age group in each survey year. The before-treatment period regroups statistics for the 1990 and 1999 surveys, whereas the after-treatment period includes the 2003, 2008 and 2013 surveys. School enrolment is a 0–1 dummy that indicates that the member reported attending school. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). Table 1: Summary Statistics, Pre- and Post- Programme Intervention (6–14, All Surveys) All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 Notes: Standard deviations are displayed in parentheses. Data are restricted to 6–14 age group in each survey year. The before-treatment period regroups statistics for the 1990 and 1999 surveys, whereas the after-treatment period includes the 2003, 2008 and 2013 surveys. School enrolment is a 0–1 dummy that indicates that the member reported attending school. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). We are also interested in the years of education completed by individuals exposed to the programme. For this, we carry out a cohort-wise analysis using the 2013 data. For this part of the exercise, we are interested mainly in assessing the stock of education accumulated by the individual exposed to the treatment. Thus, we concentrate on individuals who were exposed to the CC treatment at a younger age and who have likely finished schooling at the time they were surveyed in 2013. This cohort corresponds to individuals born between 1985 and 1993, i.e., they were 6 to 14-years-old in 1999 when the programme started, and were 20 to 28 years old in 2013. We refer to this group as the ‘young cohort’. Individuals belonging to this cohort benefited from the treatment if they resided in a treated state. We defined a comparison group, referred to as the ‘old cohort’ that consists of individuals born between 1975 and 1983. These individuals were between 16 and 24-years-old in 1999 when the programme started and were 30 and 38-years-old in 2013. These individuals would not have benefited from the treatment regardless of their state of residence. Table 2 reports descriptive statistics for the data used in the cohort-wise analysis. The young cohort in the CC treatment states received 9.85 years of education, while the young cohort in the control states reported 8.79 years. The relative change in education attainment relative to the old cohort after the intervention is 0.54 years. Table 2: Summary Statistics, Young and Old Cohort by Treatment and Comparison States (2013 Survey) All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 Notes: Standard deviations are in parentheses. The table uses data from the 2013 survey. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). Table 2: Summary Statistics, Young and Old Cohort by Treatment and Comparison States (2013 Survey) All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 Notes: Standard deviations are in parentheses. The table uses data from the 2013 survey. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). 4. Methodology To identify the effect of the schistosomiasis treatment, we exploit the fact that the CC schistosomiasis programme was restricted to four states to implement DID strategy. We define the four states that benefitted from the programme as treated, while the rest of 33 states serve as control states. To examine the effect of the treatment of the disease, we first use the following model: Enrollijt=α+β(Postt×Treatedj)+δXijt+Agei+γj+θt+εijt, (1) where Enrollijt represents the school enrolment for 6–14 aged individuals i living in state j in period t (t = 1990, 1999, 2003, 2008, 2013). Postt is a dummy to indicate the time period after the treatment intervention.9Xijt represents a set of individual-specific controls–-indicator for gender of child, area of residence, household head age, head gender and father’s education. γj and θt are state and time fixed effects. Agei is age fixed effects. The interaction term Postt × Treatedj in the equation captures the DID treatment effect on school enrolment. The DID estimates provide a causal impact of the programme under the assumption that without the programme, both treated and control states would have followed a similar trend. The standard errors are clustered at the state level.10 Moreover, to identify differences in the impact of CC programme by year, we estimate the following model that allows for the impact to vary for each year: Enrollijt=α+∑t=19992013βt(Yeart×Treatedj)+δXijt+Agei+γj+θt+εijt, (2) where Yeart denotes an indicator for the year t. The parameters βt capture the effect of a child’s exposure to the campaign for each survey year. Since the treatment could potentially impact the amount of time spent in school that will be reflected in accumulated years of schooling. To capture the impact of the CC programme on the years of education, we carry out a cohort analysis using the following equation. Educijk=α+β(Youngi×Treatedj)+δXijk+γj+τk+νijk, (3) where Educijk represents the years of education for individual i residing in state j and born in year k; Youngi is an indicator for young defined as individuals who were in age group 6–14 in 1999; Treatedj is an indicator that takes the value of 1 for the treatment state and 0 otherwise; τk is the cohort fixed effects, and εijk is the error term. Equation (3) is estimated on the sample that is restricted to individuals who are identified either as young or old, where old is defined as individuals aged between 16 and 24 in 1999. The standard errors are clustered at the state level. The identification exploits the variation of cohort exposure to treatment across time and space. As stated, the old cohorts were never exposed to the programme, regardless of state of residence. The young cohorts were exposed to the programme if they lived in a treated state.11 The coefficient of the interaction term between Youngi and treated states Treatedj captures the impact of programme. Causal interpretation of our DID estimates hinges on the identifying assumption that in the absence of the CC intervention, the outcome would have had a similar trend in both treated and comparison groups. Although it is not possible to directly test this assumption because the same young cohort who were not exposed to the programme are not observed, we perform a falsification exercise by using old cohort as a placebo treatment group while using very old cohort-defined as individuals aged between 26 and 34 years in 1999 (or aged between 40 and 48 years in 2013)—as a comparison group. We basically estimated the following equation: Educijk=α+β(Oldi×Treatedj)+δXijk+γj+τk+νijk. (4) This specification is similar to equation (3) except now our estimation sample includes individuals who are defined as old along with the very old as the excludable group. Since both old and very old cohorts did not benefit from the CC programme, we expect the coefficient β in the above equation to be indistinguishable from zero if there were no pre-existing trends differences across the treatment and comparison groups. 5. Results Table 3 reports the results of the before and after regressions using equation (1). The outcome variable indicates whether the child aged 6–14 is currently enrolled in school at the time of the survey. The DID estimate, our parameter of interest, is the coefficient associated with the interaction variable Post-1999 × Treated.12 Column (1) shows estimates for a specification that does not control for any X covariates, but includes age dummies, year, and state fixed effects. Column (2) includes additional X covariates. The DID estimate does not change by inclusion of additional covariates (X), suggesting that our results are robust to controlling for additional covariates. While the point estimates are positive, they are statistically insignificant. Hence, we cannot preclude that the programme has no impact on the school enrolment of 6 to 14-year-old children. Table 3: Schistosomiasis Programme Impact on School Enrolment All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. ‘Post-1999’ indicates the period after treatment. See equation (1) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-value for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 3: Schistosomiasis Programme Impact on School Enrolment All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. ‘Post-1999’ indicates the period after treatment. See equation (1) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-value for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. We then estimate equation (1) on sub-samples of our data to investigate the heterogeneity of the programme’s impact by gender and area of residence. Columns (3) and (4) provide estimates for boys and girls separately. We do not find statistically significant impact on either gender. Similarly, we cannot rule out no impact both in either urban or rural areas (Columns (5) and (6)). Columns (7) and (8) present the results for boys and girls in rural areas, respectively. The DID estimates are statistically insignificant for both genders in rural areas. Hence, we cannot rule out no impact of the programme. Table 4 shows the results of equation (2). The year-by-year estimates for children of 6–14 are presented with and without controls in Columns (1) and (4), respectively. The estimate for the baseline 1999 year is zero. This suggests that there were no pre-existing trend differential between treated and non-treated states before the programme was implemented. However, the year-wise impact as captured by interaction terms is not statistically significant for either of the post-programme years.13 Table 4: Schistosomiasis Programme Impact on School Enrolment, Year-by-Year Results (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. See equation (2) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 4: Schistosomiasis Programme Impact on School Enrolment, Year-by-Year Results (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. See equation (2) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. The CC programme could potentially have affected the accumulated years of education even in the absence of impact on enrolment if the treatment enabled students to stay longer in school. Alternatively, if the drugs were distributed mainly in schools, it might not have affected enrolment per se but would have increased years of education completed.14 Treated children would be more likely to progress through the grades and less likely to drop out of school. Table 5 presents the results of our cohort-wise analysis using equation (3). The first column reports the point estimate for the entire sample. The coefficient on the interaction term is positive and statistically significant at the 1% level. In other words, relative to the control states, individuals in the younger cohort gained on average 0.45 additional years of education due to their exposure to the CC treatment. The estimated effect corresponds to an approximately 5.3% increase in education attainment with respect to the control group average education. Column (2) introduces additional X controls. The magnitude of the DID estimates declines but remains statistically significant. The point estimates in Columns (3) and (4) show the treatment impact for boys and girls, respectively. The coefficient attached to the treatment interaction with young male cohort is negative and not statistically different from zero. Comparatively, the coefficient on the interaction term for young female-cohort in Column (3) is positive and statistically significant. Table 5: Schistosomiasis Programme Impact on Years of Education, Young Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 5: Schistosomiasis Programme Impact on Years of Education, Young Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. The results in Columns (5) and (6) separate the programme’s impact into urban and rural areas. The point estimate in urban areas is small and not significantly different from zero, implying that the treatment has no detectable effect in urban locations. In contrast, the effect of the programme is positive and statistically significant in rural areas. This is not surprising given that the programme was concentrated in rural areas. The young cohort in the treatment rural villages gained about an additional 0.62 year of education relative to those in control villages. In Columns (7) and (8), we show the estimates for the young males and females residing in rural areas. The estimate in Column (7) indicates the years of education for the males increased because of the programme. However, the magnitude is small, and the estimate is not statistically significant. Importantly, the females exposed to the treatment in the rural areas gained about one more year of schooling compared to females not exposed to the programme residing in rural areas of non-treated states. A possible explanation for the heterogeneity in the treatment impact could be that girls were more involved in domestic chores and thus more likely to be in contact with contaminated water. Hence, girls will be more likely to benefit from the treatment. 6. Robustness Checks In this section, we report two robustness checks. First, we carried out a placebo test using equation (4). Table 6 shows the results. None of the point estimates is statistically different from zero at conventional levels, except for the urban sample. It is noteworthy that the sign of the coefficient for the urban sample is negative, and probably the DID estimate for urban areas in Table 5 underestimates the impact. However, a mean reversion in urban areas cannot be ruled out. Nonetheless, as reported earlier, the DID estimate for urban areas is marginally negative and statistically insignificant precluding any positive impact in the urban sample. While the results presented in Table 6 do not prove the existence of similar trends in the hypothetical case of no programme, they indicate that our conclusions are unlikely to be driven by the pre-existing trend differential between treatment and control groups. Table 6: Schistosomiasis Programme Impact on Years of Education, Old Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 6: Schistosomiasis Programme Impact on Years of Education, Old Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Second, the efforts against schistosomiasis coincided with the malaria-lymphatic filaris control programme, another large-scale health intervention. We demonstrate empirically that our findings are not driven by the malaria control programme. The malaria policy intervention to eliminate lymphatic filaris started with a pilot project in 2004 and operated mass drug administration of single dose treatment as well as the distribution of insecticide-treated bed nets to households in rural villages in Plateau and Nasarawa states (Blackburn et al., 2006). It could be possible that the programme has impacted child health and education and therefore introduced a bias in the estimates reported in Tables 5 and 6. Our data identified households that received treated bed nets. We thus estimate the effect of the malaria treatment programme on young children’s education. Table 7 shows the results. The estimates on the triple interaction term across all specifications are positive but not significantly different from zero at the conventional level. Hence, there is no strong evidence of any additional effects of the malaria campaigns on the schistosomiasis control programme. Table 7: Malaria Campaigns for Insecticidal Net and Schistosomiasis Treatment (Young Cohort) All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 7: Malaria Campaigns for Insecticidal Net and Schistosomiasis Treatment (Young Cohort) All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. 6.1 Possible channels We have argued that this large-scale health programme has improved health outcomes of eligible children in treated states, and this in turn has allowed these children to accumulate more education. We do not have access to direct measures of health outcomes, especially for the relevant cohorts we use. The DHS, however, collects height and weight for women aged 15–49 years. As we mentioned, one of the consequences of schistosomiasis if left untreated is stunted growth. Hence, we can use the DHS data to explore the programme’s effects on height for women. Height is largely determined in childhood and thus would not be sensitive to current economic circumstances. We use a specification identical to equation (3), except that the dependent variable is height (in cm). Panel A of Table 8 shows the results. The point estimates indicate that in rural areas, women belonging to the treated cohorts gained 0.48 cm relative to similar cohorts in the comparison areas. The point estimates are significant at the 1% level. Noticeably, none of the estimates in Columns 1–3 (especially in urban areas) is statistically different from 0, and the results are quite consistent with those reported in Table 5: the gains are essentially concentrated for females in rural areas. While 0.48 cm may appear small in absolute value, it is in range of secular gains in heights reported in the literature (e.g., Cole, 2000; Fudvoye and Parent, 2017).15 Table 8: Schistosomiasis Programme Impact on Adult Females’ Height All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 8: Schistosomiasis Programme Impact on Adult Females’ Height All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. In Panel B of Table 8, we report robustness check using a placebo test estimated using equation (4). The point estimates are smaller in magnitude (relative to Panel A), and are statistically not different from 0. Hence, it is unlikely that the increase in height that we detect in Panel A is due to pre-existing differences in trends between treated and comparison states. Eventually, because adult heights are largely predetermined by age 5, nutritionists would interpret the results in Table 8 as only revealing that the cohort in the treatment areas had the wind at their backs from conditions they experienced before entering school (Lundeen et al., 2014; Leroy et al., 2015). We argue nevertheless that children’s physical growth could be significantly impacted after the age of 5 for two reasons. First, schistosomiasis affects preadolescent children more than adults who are yet to reach their full potential maturity. As documented, the infection results in reduced physical growth development both in height and weight of school-age children, with evidence of gender-based differences (Parraga et al., 1996; Orsini et al., 2001). Second, as discussed in Moradi (2010) and Gurarie et al. (2011), it is possible in some cases for the secular trend on growth to be reversed and for the child to make up for the shortfall in height after the age of 5. In Panel C of Table 8, we display the results of the cohort group aged 0 to 5 in 1999, children stunted from shocks experienced in utero and before age 5. The estimates are not significantly different from zero.16 7. Conclusion This paper examines the impact on schooling outcomes for children exposed to the treatment of schistosomiasis, a disease that causes anaemia, poor growth, and impaired cognitive function. Schistosomiasis is one of the largest endemic disease around the world. It is estimated that at least 90% of those requiring treatment live in Africa (WHO, 2015). Large-scale distribution of drugs is the most effective and low-cost mechanism to control the parasitic transmission, and to reduce and prevent morbidity. Our research approach takes advantage of the exogeneous variation generated from the CC’s rollout of schistosomiasis treatment in four states in Nigeria to implement a DID strategy to identify a causal relationship between the schistosomiasis control programme and the child’s enrolment and educational attainment. We do not find evidence that the programme has an impact on the enrolment of school-going children. However, we find that the cohort exposed to the treatment in rural areas accumulated an additional 0.62 years of education compared to the cohort not exposed to the treatment. Moreover, the impact of the schistosomiasis treatment is mainly on girls residing in rural areas. Consistent with the impact of education, we find that height of females exposed to the treatment in rural areas increased by 0.48 cm. Overall, the findings of this research demonstrate substantial gains in health following the mass drug administration of schistosomiasis doses. These gains in health translated, in turn, into more accumulation of education. The gains are the most substantial for females in rural areas. Acknowledgements We would like to thank the editor and two anonymous referees for their helpful suggestions. This paper has benefited from the insightful comments offered by participants of seminars at Department of Economics and Legal Studies at Oklahoma State University, and workshop participants at the African Finance and Economic Association meetings. This paper is not endorsed by the Carter Center. All errors and omissions that may remain are the sole responsibility of the authors. Footnotes 1 The Centres for Disease Control and Prevention (CDC) categorises the NTDs in endemic infections in tropical regions that not only affect the world’s poorest people but also cause disabilities that make it more difficult to succeed in school, care for family or earn a living. NTDs also predominantly occur among populations that have little or no access to good housing, safe water supplies and sanitation, or formal health systems. 2 The decline in mortality from several diseases in the twentieth century positively impacted life expectancy and population growth in the developing countries (Acemoglu and Johnson, 2007). In sub-Saharan Africa, for example, infections due to the TseTse fly affected the continent’s precolonial prospects in developing agricultural technologies (Alsan, 2014). Regions where the parasite is endemic became less likely to use domesticated animals, to use the plough, and ultimately to intensify agriculture. 3 Implications of large endemic disease on the development of low-income countries include inhibition of growth of specific geographic regions, low investment in human capital and missing opportunities in transfer of technologies and agriculture (see, Strauss and Thomas, 1998; Gallup and Sachs, 2001; Glewwe et al., 2001; Becker et al. 2005). 4 Attendance is defined as being reported as enrolled in school at the time of the survey. 5 The 1999 survey is not distributed to the public. Although the 1999 data were collected for women aged 10–49, the indicators were calculated for women aged 15–49. We were able to access the 1999 data under the disclaimer that the DHS do not stand behind the quality of the dataset for that particular reason. 6 Ebonyi and Enugu states were included in the CC programme in 2014, but since our data are up to 2013, these two states are part of non-treated states. 7 The official entrance age for the lowest level of education in Nigeria is 6-years old. Therefore, children aged 6–14 were not only likely to be enrolled in school at the time of the intervention but also able to receive the treatment. 8 Table 1 shows that treatment areas have slightly higher school enrolment rates in the baseline years, 69% versus 63%. Several factors could explain the difference in education levels. For instance, northern states in Nigeria have education levels below the national average (Umar et al., 2014). The location of our treatment states, more in the south, could partly justify the education gap. These differences are cultural and historical and therefore likely to persist over time. 9 Bertrand et al. (2004) showed that estimates and inference of difference-in-differences are sensitive to serial correlation when the data are extended to several periods, and they recommend aggregating the data in two periods of pre- and post-intervention. 10 Given the concerns about number of clusters, we also report the p-values of zero null hypothesis derived through wild bootstrap clustered at the state level as proposed by Cameron et al. (2008). We use the Stata programme cgmwildboot.ado written by Judson Caskey. 11 Assignment of individuals to states is based on the state of residence at the time of survey, i.e., 2013, and not on the state of residence during school age, i.e., during 1999–2008. This may be problematic in the presence of migration; however, only interstate migration is a concern, not within state. The interstate migration in Nigeria (for all population) was about 10% in 2006 (National Population Commission, 2010). 12 The data collected at the baseline show differences in school enrolment rates in the treatment and control areas. However, the DID estimation does not require a balance at the baseline. As long as these differences are persistent over time, the DID consistently estimates the treatment impact of the programme. 13 School enrolment rates would likely remain high for grades that include children between 6–11 years of age, that is, for those at the level of education before secondary school (Glewwe and Kremer, 2006). We show the tests for children aged 6–9 and 10–14 in Columns (2) and (3), respectively. The treatment effect on school enrolment peaks in 2008 for the 6–9 group and in 2003 for the 10–14 group. We do not find any statistical significance in the estimates when breaking down the sample for these two age groups. 14 Although we do not have documentation of where the distribution of the drugs took place, almost all the reports we came across are centred around schools and/or school children. Thus, it is likely that schools played a central role in the distribution of the drugs. 15 Cole (2000) mentions height gains ranging from 3 mm per decade in Scandinavia to 30 mm/decade in parts of Southern and Eastern Europe during latter half of the 20th century. The data reported in Fudvoye and Parent (2017) imply that height gains ranged from 3.7 mm per decade in Portugal to 15.1mm per decade in the Netherlands, between 1880 and 1980. 16 We run additional tests for children of age 0–2 and then 3–5 (results not shown). The point estimates are not significantly different from zero and do not suggest a growth deficit at early childhood for these cohort groups. Using equation (1), we also compare the change in height-for-age Z-scores (HAZ) for children under 59 months. The pre-intervention period includes only year 1990; the DHS did not collect HAZ for children in 1999. The results are, therefore, presented in the Appendix Table 9 in Supplementary material. We do not detect any sizable difference in growth stunting in children under 5. References Abdulkadir A. , Ahmed M. , Abubakar B. M. , Suleiman I. E. , Yusuf I. , Imam I. M. , Sule A. A. , Tela U. M. , Dogo H. M. , Yakasai A. M. , Musa B. M. ( 2017 ) ‘ Prevalence of Urinary Schistosomiasis in Nigeria, 1994–2015: Systematic Review and Meta-Analysis ’, African Journal of Urology , 23 : 228 – 39 . Google Scholar Crossref Search ADS Acemoglu D. , Johnson S. ( 2007 ) ‘ Disease and Development: The Effect of Life Expectancy on Economic Growth ’, Journal of Political Economy , 115 ( 6 ): 925 – 85 . Google Scholar Crossref Search ADS Agi P. , Okafor E. ( 2005 ) ‘ The Epidemiology of Schistosoma Haematobium in Odau Community in the Niger Delta Area of Nigeria ’, Journal of Applied Sciences and Environmental Management , 9 ( 3 ): 37 – 43 . Alsan M. ( 2014 ) ‘ The Effect of the Tsetse Fly on African Development ’, The American Economic Review , 105 ( 1 ): 382 – 410 . Google Scholar Crossref Search ADS Ayoya M. A. , Spiekermann-Brouwer G. M. , Traoré A. K. , Garza C. ( 2012 ) ‘ Effect on School Attendance and Performance of Iron and Multiple Micronutrients as Adjunct to Drug Treatment of Schistosoma-Infected Anemic Schoolchildren ’, Food and Nutrition Bulletin , 33 ( 4 ): 235 – 41 . Google Scholar Crossref Search ADS PubMed Babatunde T. , Asaolu S. , Sowemimo O. ( 2013 ) ‘ Urinary Schistosomiasis Among Pre-School and School Aged Children in Two Peri-Urban Communities in Southwest Nigeria ’, Journal of Parasitology and Vector Biology , 5 ( 7 ): 96 – 101 . Becker G. S. , Philipson T. J. , Soares R. R. ( 2005 ) ‘ The Quantity and Quality of Life and the Evolution of World Inequality ’, The American Economic Review , 95 ( 1 ): 277 – 91 . Google Scholar Crossref Search ADS PubMed Bertrand M. , Duflo E. , Mullainathan S. ( 2004 ) ‘ How Much Should We Trust Differences-in-Differences Estimates? ’, The Quarterly Journal of Economics , 119 ( 1 ): 249 – 75 . Google Scholar Crossref Search ADS Blackburn B. G. , Eigege A. , Gotau H. , Gerlong G. , Miri E. , Hawley W. A. , Mathieu E. , Richards F. ( 2006 ) ‘ Successful Integration of Insecticide-Treated Bed Net Distribution with Mass Drug Administration in Central Nigeria ’, The American Journal of Tropical Medicine and Hygiene , 75 ( 4 ): 650 – 5 . Google Scholar Crossref Search ADS PubMed Bleakley H. ( 2007 ) ‘ Disease and Development: Evidence from Hookworm Eradication in the American South ’, The Quarterly Journal of Economics , 122 ( 1 ): 73 – 117 . Google Scholar Crossref Search ADS PubMed Bleakley H. ( 2010 ) ‘ Malaria Eradication in the Americas: A Retrospective Analysis of Childhood Exposure ’, American Economic Journal: Applied Economics , 2 ( 2 ): 1 – 45 . Google Scholar Crossref Search ADS Bloom D. E. , Canning D. , Sevilla J. ( 2004 ) ‘ The Effect of Health on Economic Growth: A Production Function Approach ’, World Development , 32 ( 1 ): 1 – 13 . Google Scholar Crossref Search ADS Cameron A. C. , Gelbach J. B. , Miller D. L. ( 2008 ) ‘ Bootstrap-Based Improvements for Inference with Clustered Errors ’, The Review of Economics and Statistics , 90 ( 3 ): 414 – 27 . Google Scholar Crossref Search ADS Cole T. J. ( 2000 ) ‘ Secular Trends in Growth ’, Proceedings of the Nutrition Society , 59 ( 2 ): 317 – 24 . Google Scholar Crossref Search ADS PubMed Cutler D. , Fung W. , Kremer M. , Singhal M. , Vogl T. ( 2010 ) ‘ Early-Life Malaria Exposure and Adult Outcomes: Evidence from Malaria Eradication in India ’, American Economic Journal: Applied Economics , 2 ( 2 ): 72 – 94 . Google Scholar Crossref Search ADS Ekanem E. , Asindi A. , Ejezie G. , Antia-Obong O. ( 1994 ) ‘ Effect of Schistosoma Haematobium Infection on the Physical Growth and School Performance of Nigerian Children ’, The Central African Journal of Medicine , 40 ( 2 ): 38 – 44 . Google Scholar PubMed Ezeadila J. O. , Okoli I. , Agomuo M. , Aneke F. C. , Egbuche C. M. ( 2015 ) ‘ Prevalence of Urinary Schistosomiasis Among Community Primary School Pupils in Amagunze, Enugu State, Nigeria ’, World Rural Observations , 7 ( 3 ): 46 – 51 . Fudvoye J. , Parent A.-S. ( 2017 ) ‘Secular trends in growth’, in Annales d’Endocrinologie , Vol. 78 . Issy-les-Moulineaux, France : Elsevier , pp. 88 – 91 . Gallup J. L. , Sachs J. D. ( 2001 ) ‘ The Economic Burden of Malaria ’, The American Journal of Tropical Medicine and Hygiene , 64 ( 1 ): 85 – 96 . Google Scholar Crossref Search ADS PubMed Glewwe P. , Jacoby H. G. , King E. M. ( 2001 ) ‘ Early Childhood Nutrition and Academic Achievement: A Longitudinal Analysis ’, Journal of Public Economics , 81 ( 3 ): 345 – 68 . Google Scholar Crossref Search ADS Glewwe P. , Kremer M. ( 2006 ) ‘ Schools, Teachers, and Education Outcomes in Developing Countries ’, Handbook of the Economics of Education , 2 : 945 – 1017 . Google Scholar Crossref Search ADS Gurarie D. , Wang X. , Bustinduy A. L. , King C. H. ( 2011 ) ‘ Modeling the Effect of Chronic Schistosomiasis on Childhood Development and the Potential for Catch-up Growth with Different Drug Treatment Strategies Promoted for Control of Endemic Schistosomiasis ’, The American Journal of Tropical Medicine and Hygiene , 84 ( 5 ): 773 – 81 . Google Scholar Crossref Search ADS PubMed Gutman J. , Fagbemi A. , Alphonsus K. , Eigege A. , Miri E. , Richards F. Jr ( 2008 ) ‘ Missed Treatment Opportunities for Schistosomiasis Mansoni, in an Active Programme for the Treatment of Urinary Schistosomiasis in Plateau and Nasarawa States, Nigeria ’, Annals of Tropical Medicine and Parasitology , 102 ( 4 ): 335 – 46 . Google Scholar Crossref Search ADS PubMed Hopkins D. R. , Eigege A. , Miri E. S. , Gontor I. , Ogah G. , Umaru J. , Gwomkudu C. C. , Mathai W. , Amadiegwu S. , Oyenekan O. K. , et al. ( 2002 ) ‘ Lymphatic Filariasis Elimination and Schistosomiasis Control in Combination with Onchocerciasis Control in Nigeria ’, The American Journal of Tropical Medicine and Hygiene , 67 ( 3 ): 266 – 72 . Google Scholar Crossref Search ADS PubMed Hopkins D. R. , Richards F. O. Jr , Ruiz-Tiben E. , Emerson P. , Withers C. Jr , et al. ( 2008 ) ‘ Dracunculiasis, Onchocerciasis, Schistosomiasis, and Trachoma ’, Annals of the New York Academy of Sciences , 1136 ( 1 ): 45 – 52 . Google Scholar Crossref Search ADS PubMed Kazianga H. , Masters W. A. , McMillan M. S. ( 2014 ) ‘ Disease control, demographic change and institutional development in Africa ’, Journal of Development Economics , 110 : 313 – 26 . Google Scholar Crossref Search ADS Leroy J. L. , Ruel M. , Habicht J.-P. , Frongillo E. A. ( 2015 ) ‘ Using Height-for-Age Differences (HAD) Instead of Height-for-Age Z-Scores (HAZ) for the Meaningful Measurement of Population-Level Catch-up in Linear Growth in Children Less than 5 Years of Age ’, BMC Pediatrics , 15 ( 1 ): 145 – 56 . Google Scholar Crossref Search ADS PubMed Lundeen E. A. , Stein A. D. , Adair L. S. , Behrman J. R. , Bhargava S. K. , Dearden K. A. , Gigante D. , Norris S. A. , Richter L. M. , Fall C. H. , et al. ( 2014 ) ‘ Height-for-Age Z Scores Increase Despite Increasing Height Deficits Among Children in 5 Developing Countries ’, The American Journal of Clinical Nutrition , 100 ( 3 ): 821 – 5 . Google Scholar Crossref Search ADS PubMed Meremikwu M. , Asuquo P. , Ejezie G. , Useh M. , Udoh A. ( 2000 ) ‘ Treatment of Schistosomiasis Haematobium with Praziquantel in Children: Its Effect on Educational Performance in Rural Nigeria ’, Tropical Medicine , 42 ( 1 ): 39 – 45 . Miguel E. , Kremer M. ( 2004 ) ‘ Worms: Identifying Impacts on Education and Health in the Presence of Treatment Externalities ’, Econometrica , 72 ( 1 ): 159 – 217 . Google Scholar Crossref Search ADS Moradi A. ( 2010 ). Selective mortality or growth after childhood? What really is key to understand the puzzlingly tall adult heights in sub-SaharanAfrica. CSAE Working Paper Series 2010–17, Centre for the Study of African Economies, University of Oxford. National Population Commission ( 2010 ) Population and Housing Census of the Federal Republic of Nigeria 2006, Government Publication . Abuja, Nigeria : National Population Commission . Orsini M. , Rocha R. S. , Disch J. , Katz N. , Rabello A. ( 2001 ) ‘ The role of Nutritional Status and Insulin-Like Growth Factor in Reduced Physical Growth in Hepatosplenic Schistosoma Mansoni Infection ’, Transactions of the Royal Society of Tropical Medicine and Hygiene , 95 ( 4 ): 453 – 6 . Google Scholar Crossref Search ADS PubMed Parraga I. M. , Assis A. M. O. , Prado M. S. , Barreto M. L. , Reis M. G. , King C. H. , Blanton R. E. ( 1996 ) ‘ Gender Differences in Growth of School-Aged Children with Schistosomiasis and Geohelminth Infection ’, The American Journal of Tropical Medicine and Hygiene , 55 ( 2 ): 150 – 6 . Google Scholar Crossref Search ADS PubMed Sam-Wobo S. , Idowu J. , Adeleke M. , et al. ( 2011 ) ‘ Urinary Schistosomiasis Among Children and Teenagers near Oyan Dam, Abeokuta, Nigeria ’, Journal of Rural and Tropical Public Health , 10 : 57 – 60 . Strauss J. , Thomas D. ( 1998 ) ‘ Health, Nutrition, and Economic Development ’, Journal of Economic Literature , 36 ( 2 ): 766 – 817 . The Carter Center ( 2000 ). The Carter Center Annual Report 1998–1999. https://www.cartercenter.org/documents/520.pdf. Umar H. M. , Ismail R. , Abdul-Hakim R. ( 2014 ) ‘ Regional Inequality of Educational Attainment in Nigeria ’, British Journal of Economics, Management and Trade , 4 ( 3 ): 420 – 30 . Google Scholar Crossref Search ADS Venkataramani A. S. ( 2012 ) ‘ Early Life Exposure to Malaria and Cognition in Adulthood: Evidence from Mexico ’, Journal of Health Economics , 31 ( 5 ): 767 – 80 . Google Scholar Crossref Search ADS PubMed Weil D. N. ( 2007 ) ‘ Accounting for the Effect of Health on Economic Growth ’, The Quarterly Journal of Economics , 122 ( 3 ): 1265 – 1306 . Google Scholar Crossref Search ADS Weil D. N. ( 2010 ) ‘ Endemic Diseases and African Economic Growth: Challenges and Policy Responses ’, Journal of African Economies , 19 ( 3 ): 81 – 109 . World Health Organization ‘ Schistosomiasis: Number of People Treated Worldwide in 2013 ’, Weekly Epidemiological Record , 90 ( 5 ): 25 – 32 . © The Author(s) 2018. Published by Oxford University Press on behalf of the Centre for the Study of African Economies, all rights reserved. For Permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of African Economies Oxford University Press

Returns to Controlling a Neglected Tropical Disease: Schistosomiasis Control Programme and Education Outcomes in Nigeria

Loading next page...
 
/lp/ou_press/returns-to-controlling-a-neglected-tropical-disease-schistosomiasis-0iGBKuRe33
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Centre for the Study of African Economies, all rights reserved. For Permissions, please email: journals.permissions@oup.com.
ISSN
0963-8024
eISSN
1464-3723
D.O.I.
10.1093/jae/ejy003
Publisher site
See Article on Publisher Site

Abstract

Abstract Using the rollout of the schistosomiasis campaign in Nigeria as a quasi-experiment, we examine the impact of the disease control programme on school-age children education outcomes. Schistosomiasis is a parasitic disease caused by infections from a small worm. Its most severe effects hamper growth and cognitive development of children. The mass campaign targeted four states that saw large reduction in the infectious disease afterwards. Using difference-in-differences strategy, we find that the cohort exposed to the treatment in rural areas accumulated an additional 0.6 years of education compared to cohort not exposed to the treatment. Moreover, the impact of the schistosomiasis treatment is mainly on girls residing in rural areas. 1. Introduction Schistosomiasis, a neglected tropical disease (NTD), is an acute and chronic disease caused by parasitic worms.1 According to the World Health Organization (WHO), at least 258 million people required treatment in 2014. The disease is particularly prevalent among school-age children. In general, repeated treatment helps reduce and prevent morbidity. We examine a deworming programme that provided regular treatment to at-risk population in one of the regions with the greatest burden of the disease. We do not find conclusive evidence that treatment increased school enrolment for 6 to 14-year-old children. However, we find that the cohort exposed to treatment in rural areas accumulated 0.6 years more of education. Moreover, most of the impact is driven by the impact on girl children. Resolving the problem of endemic disease is crucial to development in low-income countries. Studies in development economics provide evidence of the relationship between overall health environment and long-term changes in development outcomes. Although the role of disease reduction in affecting economic growth is found to be rather weak (Weil, 2010), global improvement in health conditions not only improved lives but also enhanced economic growth (Bloom et al., 2004; Weil, 2007). Albeit the debate on the empirical estimation is still unsettled, the implication for developing world appears to be of significance.2 Public policies directed toward the decline in prevalence, morbidity and mortality rates of infectious diseases not only have important regional implications but also implications at the individual level.3 Better health capital improves the individual return to investment in education and leads to higher future returns. For instance, Bleakley (2010) documents how and the extent to which exposure to endemic disease in early childhood development affects labour productivity in the adulthood. Cutler et al. (2010) review a malaria eradication programme in India and report a positive impact on the cohort’s expenditure and per capita household consumption. Venkataramani (2012) studies a nation-wide effort to eliminate malaria in Mexico in the 1950s. The author finds that, in the case for men, birth year exposure in treated states led to increases in cognitive test score performance and better schooling outcomes. The model, however, holds smaller and no significant estimates for women. Kazianga et al. (2014) find an increase in rural population and land transactions in Burkina Faso after a campaign against Onchoceriasis run by the WHO. Exposure to the Onchoceriasis control programme induced villages to develop and improve local institutions (e.g., land markets) and the provision of public goods (e.g., public markets, primary schools and telephone services). Bleakley (2007) provides evidence of the benefits of disease control on economic outcomes. He evaluates the impact of the eradication of hookworm, an infection similar to schistosomiasis, in the American South. Hookworm, an intestinal parasite, was responsible for half of the literacy rate gap between northern and southern states. After the programme was completed, it appeared that the disease accounted for as much as 20% of the difference in income between the north and the south. Other findings indicate that hookworm eradication had large positive effects on school enrolment and on years of education completed. Miguel and Kremer (2004) investigate the effects of deworming programmes in areas with high helminth infection rates of intestinal worms, including hookworm, roundworm and schistosomiasis in rural Kenya. The authors focus on school-age children and enrolment in school. They show evidence of substantial increases in school participation and reduced absenteeism in randomly selected treated schools. Although they found no evidence on pupils’ test scores, they argue that the deworming programme was extremely cost effective in terms of returns to education and human capital investments. Their cost-benefits analysis indicates that the programme has sizeable benefits. Our study exploits a large campaign of drug administration to school-age children that has reduced the prevalence of schistosomiasis in Nigeria. The disease causes anaemia, stunted growth, cognitive impairment and premature death. The drug distribution started in 1999 and initially covered two states (i.e., Plateau and Nasarawa). By the year 2010, the programme had expanded to include four states, and over a million children received the treatment each year. A substantial body of the medical literature has examined the effectiveness of the schistosomiasis mass treatment. Most of these studies that have so far investigated student performance response to the treatment have relied on cross sectional techniques comparing the outcomes of a control and a test group (Ekanem et al., 1994; Meremikwu et al., 2000; Ayoya et al., 2012). This paper contributes to and extends the literature on health and education externalities by reporting empirical results of the schistosomiasis control programme impact on children’s educational outcomes. Our estimation strategy uses difference-in-differences (DID) to show that efforts to eliminate the disease in developing region had positive impacts on children’s educational outcomes. By taking advantage of the expansion of the schistosomiasis control programme in four states of Nigeria from 1999 to 2013, we show that the reduction in the water-borne disease has contributed in improving the level of education of younger cohorts who were exposed to the programme. The remainder of the paper is organised as follows. Section 2 describes of the disease, the circumstances in which a child might get infected, and the symptoms of the infection. It also discusses the treatment campaign conducted by the Carter Center (CC) and the effectiveness of the programme. Section 3 describes the data employed. Section 4 sets out the model specification and the identification strategy used to capture the treatment impact. The findings of our study are presented in Section 5. The last section concludes. 2. Programme Description 2.1 Schistosomiasis Schistosomiasis is a water-borne disease that mostly affects children in tropical regions especially in developing countries of Africa, Asia and South America. The schistosome parasite, which is a worm, is acquired by contact with unprotected stagnant water. The classical sign of schistosomiasis infection is blood in the urine. Once in the blood vessels, the parasite attacks mainly the bladder and kidneys and causes fever and pain in the stomach and during urination. The most vulnerable group to the disease are school-age children. They are more likely to come in contact with the vector, and the disease exposes them to serious deficiency manifested through anaemia, inhibited growth, debility and high morbidity. Nigeria has one of the highest prevalence rates of schistosomiasis in the world. Abdulkadir et al. (2017) review a large number of articles in an attempt to estimate the prevalence of schistosomiasis in Nigeria. Among the analysed studies, they find the prevalence of urinary schistosomiasis to vary from 2 to 82.5%. Schistosomiasis is a concern for the developing world because it tends to be endemic in both infested rural places and densely populated urban areas as well. As with other NTDs, schistosomiasis can hinder development prospects if it is not adequately addressed. Although schistosomiasis is endemic in many regions, there are effective treatment regimens. The most common course of treatment consists of periodic intake of the drug praziquantel. The drug is highly effective as it can reverse up to 90% of the damage at a relatively low cost, between $0.15 and $0.20 per treatment (Hopkins et al., 2008). The WHO is active in raising awareness about schistosomiasis and recommends health education on access to safe water, improved sanitation and hygiene. Gutman et al. (2008) advocate for preventive mass treatment of all school-age children as the cheapest approach to control schistosomiasis. The authors note that the disease affects the ability of infected children ability to work, by weakening their bodies. Furthermore, infected children show signs of poor growth and learning difficulty at school. Agi and Okafor (2005) study the epidemiology of schistosomiasis and find a significant negative correlation between age and intensity of infection. Their results suggest that there is a progressive rise in prevalence for children aged 5–9, but the infection rate often peaks for children aged 10–14 and grows weaker as the individual gets older. Similarly, (Ezeadila, et al. 2015) find the difference in prevalence between children aged 6–9 and 10–13 to not be significant at the 5% level. The paper is connected to a strand of literature that relates the impact of a NTD to children’s educational outcomes. Early studies, such as (Ekanem et al., 1994), in a cross sectional analysis assess the effect of schistosoma infection in children aged 5–15 in southeastern Nigeria to find no significant impact on their physical growth and school performance. Meremikwu et al. (2000) also find no improvement in school attendance following repeated treatments of praziquantel to children aged 8–9 in Adim, Nigeria.4 Like Ekanem et al. (1994), Meremikwu et al. (2000) employ a simple difference comparing cohort pupils’ characteristics pre-treatment and post-treatment period without controlling for observed or unobserved factors within child, school or village that could influence the outcome. Ayoya et al. (2012) offer a stronger analysis, studying primary school children aged 7–12 in poor urban area in Bamako, Mali. Using a linear regression model, they find a significant increase in children’s attendance and school achievement, measured by the pupil’s passing rate. They acknowledge that they do not control for unobserved factors that could bias their estimates. By using DIDs, we can account for unobserved heterogeneity across time and space in estimating the effects of schistosomiasis control on children’s educational outcomes. 2.2 Programme intervention As part of the effort to control the disease, the CC has provided schistosomiasis health education and drug distribution in Nigeria since 1999 to children aged 5–14. Originally, the CC operated in Nigeria for the treatment of onchoceriasis. The CC eventually obtained a grant from Smithkline Beecham to begin the treatment of urinary schistosomiasis, the most severe form of the disease. In addition to that, the project was supported by a donation of 50,000 doses of praziquantel from Medochemie, Bayer and Shin Poong (Hopkins et al., 2002). First, the Federal Ministry of Health conducted a nation-wide postal survey in 1990, confirming the presence and the high prevalence of urinary schistosomiasis throughout the country. The rate of infection was particularly high among people living in poor rural areas with little or no access to a safe source of water. Of all age groups, children between 5 and 14 years who were traditionally responsible for water-related household chores were the most exposed and commonly affected. The CC mapping of the states of Plateau and Nasarawa confirmed the nation-wide mapping statements. For instance, in the village of Mungkohot in the Plateau state, a staggering 80% of school-age children were found infected with the disease. In the first year of the programme, the CC assessed the activity of 150 villages to not only determine where the infectious disease was located but also to identify communities’ habits and practices that were used in the preparation of health education material for the schistosomiasis campaign. (The Carter Center, 2000) The control programme at its pilot phase reached more than 8,000 people in highly affected villages in the states of Plateau and Nasarawa. The CC schistosomiasis programme, in association with the Federal Ministry of Health, initially launched its global assistance programme in two states, Plateau and Nasarawa. In 2004, the pilot programme expanded into all local government areas in these states and grew to include the state of Delta (2010) and Edo (2010). During the schistosomiasis campaign, the CC distributed treatment doses of praziquantel to the at-risk populations in the states of Plateau, Edo, Delta and Nasarawa (Figure 1). They received medicine for the severely debilitating form of the disease, the urinary schistosomiasis. In 2008, the CC received a donation of 1.1 million doses of praziquantel followed by another donation of 1.5 million doses in 2009 from the WHO and Merck. The contributions made in 2008 and 2009 to the programme surpassed the cumulative number of treatments from 1999 to 2007 (1.08 million treatments) to illustrate a significant expansion of the schistosomiasis control programme (Figure 2). Figure 1: View largeDownload slide Schistosomiasis Treatment States in Nigeria, 2013. Notes: Carter Center-Assisted Schistosomiasis Programme Treatment States in Nigeria. In 2014, the Treatment Initiative Was Extended to the States of Ebonyi and Enugu. Source: The Carter Center. Figure 1: View largeDownload slide Schistosomiasis Treatment States in Nigeria, 2013. Notes: Carter Center-Assisted Schistosomiasis Programme Treatment States in Nigeria. In 2014, the Treatment Initiative Was Extended to the States of Ebonyi and Enugu. Source: The Carter Center. Figure 2: View largeDownload slide Annual Praziquantel Treatments, 1999–2014. Notes: Carter Center-Assisted Schistosomiasis Programme Annual Praziquantel Treatments. Source: The Carter Center (Disease Data, updated in May 2015). Figure 2: View largeDownload slide Annual Praziquantel Treatments, 1999–2014. Notes: Carter Center-Assisted Schistosomiasis Programme Annual Praziquantel Treatments. Source: The Carter Center (Disease Data, updated in May 2015). The intervention as of today is believed to have considerably reduced schistosomiasis infection in the treated states. The CC reports that blood in school children’s urine has been reduced by approximately 94% in Plateau and Nasarawa states and approximately 88% in Delta state. The claim has been supported by several studies that documented a reduction in blood in urine in areas targeted by the CC (Agi and Okafor, 2005; Hopkins et al., 2002, 2008). This is pertinent to our study because we assume the treatment was effective in successfully controlling the disease. On the one hand, we have studies not related to the programme that collected urine samples among school-age children in the CC states and document a low prevalence of the disease (Ezeadila et al. 2015). On the other hand, there are findings of studies conducted in control states that concluded high prevalence of urinary schistosomiasis and the endemic state to the infection of many communities (Sam-Wobo et al., 2011; Babatunde et al., 2013). 3. Data To examine the effect of the programme, we use five rounds of the Demographic Health Surveys (DHS) on Nigeria collected in the years 1990, 1999, 2003, 2008 and 2013.5 The surveys initiated by the National Population Commission (partly funded through the United States Agency for International Development (USAID)) were developed to provide accurate information on maternal and child health but also family planning. Samples of the DHS are drawn to be nationally representative. Administratively, Nigeria is divided into states, local government areas, localities and enumeration areas (EA). The DHS clusters, or primary sampling unit, were based on census EA. Some clusters combined several EA because the DHS requires a minimum of 80 households for each cluster. The sample includes all 37 states, including Plateau, Edo, Delta and Nasarawa where most of the efforts against schistosomiasis by the CC have been concentrated. The CC worked closely with the Ministry of Health and, to the best of our knowledge, there were no other similar programmes before or concurrently with the CC intervention. Since the CC programme started in 1999, we treat the 1990 and the 1999 survey as pre-programme, while the 2004, 2008 and 2013 surveys serve as post-programme data. Because the programme was concentrated in four states of Nigeria, we group these four states (Plateau, Nasarawa, Delta and Edo) as treated states, while the other 33 states serve as comparison states.6 Moreover, because the main beneficiaries were school-going children, we restrict the data to individuals in 6–14 age group.7 Table 1 provides descriptive statistics for some key variables for individuals in the 6–14 age group by treatment status. School enrolment is a dummy variable to indicate whether the member reported attending school during the current school year. Enrolment in the treated states was larger compared to the non-treated states in both the pre-programme and post-programme period.8 Table 1: Summary Statistics, Pre- and Post- Programme Intervention (6–14, All Surveys) All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 Notes: Standard deviations are displayed in parentheses. Data are restricted to 6–14 age group in each survey year. The before-treatment period regroups statistics for the 1990 and 1999 surveys, whereas the after-treatment period includes the 2003, 2008 and 2013 surveys. School enrolment is a 0–1 dummy that indicates that the member reported attending school. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). Table 1: Summary Statistics, Pre- and Post- Programme Intervention (6–14, All Surveys) All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 All sample Before treatment After treatment Difference in means Treated states Non-treated states Treated states Non-treated states (4) − (2) (5) − (3) (1) (2) (3) (4) (5) (6) (7) School enrolment 0.69 0.69 0.63 0.87 0.69 0.18 0.06 (0.46) (0.46) (0.48) (0.33) (0.46) [0.00] [0.00] Age (years) 9.56 9.80 9.58 9.66 9.54 −0.14 −0.04 (2.55) (2.62) (2.55) (2.56) (2.54) [0.02] [0.02] Gender (male = 1) 0.51 0.53 0.51 0.51 0.51 −0.02 0.00 (0.50) (0.50) (0.50) (0.50) (0.50) [0.22] [0.67] Age of head (years) 48.39 51.24 49.14 48.09 48.17 −3.15 −0.97 (13.05) (14.54) (13.70) (13.72) (12.75) [0.00] [0.00] Household size 8.02 10.11 8.32 7.87 7.92 −2.24 −0.40 (4.08) (7.57) (4.59) (3.89) (3.82) [0.00] [0.00] Father education (years) 2.45 2.64 2.11 3.14 2.45 0.50 0.34 (3.04) (2.63) (3.08) (3.16) (3.01) [0.00] [0.00] Mother education (years) 0.35 0.28 0.26 0.49 0.36 0.21 0.10 (0.48) (0.38) (0.43) (0.49) (0.48) [0.00] [0.00] Gender of head (male = 1) 0.87 0.85 0.88 0.83 0.87 −0.02 −0.01 (0.34) (0.36) (0.33) (0.38) (0.34) [0.03] [0.00] Observations 112,788 2103 19,485 9002 82,198 Notes: Standard deviations are displayed in parentheses. Data are restricted to 6–14 age group in each survey year. The before-treatment period regroups statistics for the 1990 and 1999 surveys, whereas the after-treatment period includes the 2003, 2008 and 2013 surveys. School enrolment is a 0–1 dummy that indicates that the member reported attending school. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). We are also interested in the years of education completed by individuals exposed to the programme. For this, we carry out a cohort-wise analysis using the 2013 data. For this part of the exercise, we are interested mainly in assessing the stock of education accumulated by the individual exposed to the treatment. Thus, we concentrate on individuals who were exposed to the CC treatment at a younger age and who have likely finished schooling at the time they were surveyed in 2013. This cohort corresponds to individuals born between 1985 and 1993, i.e., they were 6 to 14-years-old in 1999 when the programme started, and were 20 to 28 years old in 2013. We refer to this group as the ‘young cohort’. Individuals belonging to this cohort benefited from the treatment if they resided in a treated state. We defined a comparison group, referred to as the ‘old cohort’ that consists of individuals born between 1975 and 1983. These individuals were between 16 and 24-years-old in 1999 when the programme started and were 30 and 38-years-old in 2013. These individuals would not have benefited from the treatment regardless of their state of residence. Table 2 reports descriptive statistics for the data used in the cohort-wise analysis. The young cohort in the CC treatment states received 9.85 years of education, while the young cohort in the control states reported 8.79 years. The relative change in education attainment relative to the old cohort after the intervention is 0.54 years. Table 2: Summary Statistics, Young and Old Cohort by Treatment and Comparison States (2013 Survey) All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 Notes: Standard deviations are in parentheses. The table uses data from the 2013 survey. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). Table 2: Summary Statistics, Young and Old Cohort by Treatment and Comparison States (2013 Survey) All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 All sample Treatment states Control states Difference in means Age 6–14 in 1999 Age 16–24 in 1999 Age 6–14 in 1999 Age 16–24 in 1999 (2) − (4) (3) − (5) (1) (2) (3) (4) (5) (6) (7) Years of education 7.53 9.85 7.53 8.79 7.01 1.06 0.52 (5.61) (4.34) (5.56) (5.00) (5.82) [0.00] [0.00] Age (years) 28.01 23.85 23.80 33.52 33.40 −9.67 −9.6 (5.49) (2.69) (2.70) (2.69) (2.73) [0.00] [0.00] Gender (male = 1) 0.46 0.48 0.44 0.50 0.48 −0.02 −0.04 (0.50) (0.50) (0.50) (0.50) (0.50) [0.09] [0.00] Age of head (years) 41.49 44.12 41.27 42.51 41.23 1.61 0.04 (14.55) (17.35) (16.27) (13.06) (11.55) [0.00] [0.80] Household size 5.82 5.76 5.69 5.77 5.99 −0.01 −0.3 (3.66) (3.84) (3.83) (3.38) (3.44) [0.95] [0.00] Father education (years) 2.90 3.37 2.68 3.60 3.01 −0.23 −0.33 (2.97) (4.46) (2.98) (2.44) (2.66) [0.04] [0.00] Mother education (years) 0.58 0.66 0.56 0.70 0.59 −0.04 −0.03 (0.82) (0.84) (0.84) (0.76) (0.79) [0.18] [0.00] Gender of head (male = 1) 0.87 0.82 0.85 0.88 0.90 −0.06 −0.05 (0.34) (0.38) (0.36) (0.33) (0.30) [0.00] [0.00] Observations 43,492 2772 21,665 2050 17,005 Notes: Standard deviations are in parentheses. The table uses data from the 2013 survey. The P-values for the test in mean difference are reported in brackets in Columns (6) and (7). 4. Methodology To identify the effect of the schistosomiasis treatment, we exploit the fact that the CC schistosomiasis programme was restricted to four states to implement DID strategy. We define the four states that benefitted from the programme as treated, while the rest of 33 states serve as control states. To examine the effect of the treatment of the disease, we first use the following model: Enrollijt=α+β(Postt×Treatedj)+δXijt+Agei+γj+θt+εijt, (1) where Enrollijt represents the school enrolment for 6–14 aged individuals i living in state j in period t (t = 1990, 1999, 2003, 2008, 2013). Postt is a dummy to indicate the time period after the treatment intervention.9Xijt represents a set of individual-specific controls–-indicator for gender of child, area of residence, household head age, head gender and father’s education. γj and θt are state and time fixed effects. Agei is age fixed effects. The interaction term Postt × Treatedj in the equation captures the DID treatment effect on school enrolment. The DID estimates provide a causal impact of the programme under the assumption that without the programme, both treated and control states would have followed a similar trend. The standard errors are clustered at the state level.10 Moreover, to identify differences in the impact of CC programme by year, we estimate the following model that allows for the impact to vary for each year: Enrollijt=α+∑t=19992013βt(Yeart×Treatedj)+δXijt+Agei+γj+θt+εijt, (2) where Yeart denotes an indicator for the year t. The parameters βt capture the effect of a child’s exposure to the campaign for each survey year. Since the treatment could potentially impact the amount of time spent in school that will be reflected in accumulated years of schooling. To capture the impact of the CC programme on the years of education, we carry out a cohort analysis using the following equation. Educijk=α+β(Youngi×Treatedj)+δXijk+γj+τk+νijk, (3) where Educijk represents the years of education for individual i residing in state j and born in year k; Youngi is an indicator for young defined as individuals who were in age group 6–14 in 1999; Treatedj is an indicator that takes the value of 1 for the treatment state and 0 otherwise; τk is the cohort fixed effects, and εijk is the error term. Equation (3) is estimated on the sample that is restricted to individuals who are identified either as young or old, where old is defined as individuals aged between 16 and 24 in 1999. The standard errors are clustered at the state level. The identification exploits the variation of cohort exposure to treatment across time and space. As stated, the old cohorts were never exposed to the programme, regardless of state of residence. The young cohorts were exposed to the programme if they lived in a treated state.11 The coefficient of the interaction term between Youngi and treated states Treatedj captures the impact of programme. Causal interpretation of our DID estimates hinges on the identifying assumption that in the absence of the CC intervention, the outcome would have had a similar trend in both treated and comparison groups. Although it is not possible to directly test this assumption because the same young cohort who were not exposed to the programme are not observed, we perform a falsification exercise by using old cohort as a placebo treatment group while using very old cohort-defined as individuals aged between 26 and 34 years in 1999 (or aged between 40 and 48 years in 2013)—as a comparison group. We basically estimated the following equation: Educijk=α+β(Oldi×Treatedj)+δXijk+γj+τk+νijk. (4) This specification is similar to equation (3) except now our estimation sample includes individuals who are defined as old along with the very old as the excludable group. Since both old and very old cohorts did not benefit from the CC programme, we expect the coefficient β in the above equation to be indistinguishable from zero if there were no pre-existing trends differences across the treatment and comparison groups. 5. Results Table 3 reports the results of the before and after regressions using equation (1). The outcome variable indicates whether the child aged 6–14 is currently enrolled in school at the time of the survey. The DID estimate, our parameter of interest, is the coefficient associated with the interaction variable Post-1999 × Treated.12 Column (1) shows estimates for a specification that does not control for any X covariates, but includes age dummies, year, and state fixed effects. Column (2) includes additional X covariates. The DID estimate does not change by inclusion of additional covariates (X), suggesting that our results are robust to controlling for additional covariates. While the point estimates are positive, they are statistically insignificant. Hence, we cannot preclude that the programme has no impact on the school enrolment of 6 to 14-year-old children. Table 3: Schistosomiasis Programme Impact on School Enrolment All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. ‘Post-1999’ indicates the period after treatment. See equation (1) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-value for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 3: Schistosomiasis Programme Impact on School Enrolment All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 All sample All sample Boys only Girls only Urban areas Rural areas Boys in rural Girls in rural (1) (2) (3) (4) (5) (6) (7) (8) Post-1999 × treated states 0.06 0.06 0.06 0.05 −0.01 0.08 0.08 0.08 (0.10) (0.09) (0.08) (0.10) (0.07) (0.08) (0.08) (0.09) [0.45] [0.46] [0.44] [0.46] [0.88] [0.42] [0.33] [0.45] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 112,943 112,788 57,373 55,415 37,324 75,464 38,635 36,829 R2 0.30 0.35 0.32 0.38 0.19 0.36 0.33 0.39 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. ‘Post-1999’ indicates the period after treatment. See equation (1) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-value for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. We then estimate equation (1) on sub-samples of our data to investigate the heterogeneity of the programme’s impact by gender and area of residence. Columns (3) and (4) provide estimates for boys and girls separately. We do not find statistically significant impact on either gender. Similarly, we cannot rule out no impact both in either urban or rural areas (Columns (5) and (6)). Columns (7) and (8) present the results for boys and girls in rural areas, respectively. The DID estimates are statistically insignificant for both genders in rural areas. Hence, we cannot rule out no impact of the programme. Table 4 shows the results of equation (2). The year-by-year estimates for children of 6–14 are presented with and without controls in Columns (1) and (4), respectively. The estimate for the baseline 1999 year is zero. This suggests that there were no pre-existing trend differential between treated and non-treated states before the programme was implemented. However, the year-wise impact as captured by interaction terms is not statistically significant for either of the post-programme years.13 Table 4: Schistosomiasis Programme Impact on School Enrolment, Year-by-Year Results (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. See equation (2) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 4: Schistosomiasis Programme Impact on School Enrolment, Year-by-Year Results (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 (1) (2) (3) (4) School enrolment School enrolment School enrolment School enrolment Age 6–14 Age 6–9 Age 10–14 Age 6–14 Year 1999 × treated states −0.01 0.03 −0.01 0.00 (0.11) (0.10) (0.07) (0.08) [0.89] [0.77] [0.68] [0.93] Year 2003 × treated states 0.09 0.09 0.06 0.07 (0.16) (0.15) (0.13) (0.14) [0.45] [0.59] [0.53] [0.49] Year 2008 × treated states 0.08 0.10 0.05 0.07 (0.15) (0.13) (0.11) (0.12) [0.45] [0.47] [0.56] [0.45] Year 2013 × treated states 0.03 0.06 0.02 0.04 (0.15) (0.14) (0.11) (0.12) [0.60] [0.51] [0.86] [0.53] Controls No Yes Yes Yes Observations 112,943 57,671 55,117 112,788 R2 0.30 0.35 0.34 0.35 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. ‘Treated states’ indicate the Carter Center-assisted states of Delta, Nasarawa, Edo and Plateau. See equation (2) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. The CC programme could potentially have affected the accumulated years of education even in the absence of impact on enrolment if the treatment enabled students to stay longer in school. Alternatively, if the drugs were distributed mainly in schools, it might not have affected enrolment per se but would have increased years of education completed.14 Treated children would be more likely to progress through the grades and less likely to drop out of school. Table 5 presents the results of our cohort-wise analysis using equation (3). The first column reports the point estimate for the entire sample. The coefficient on the interaction term is positive and statistically significant at the 1% level. In other words, relative to the control states, individuals in the younger cohort gained on average 0.45 additional years of education due to their exposure to the CC treatment. The estimated effect corresponds to an approximately 5.3% increase in education attainment with respect to the control group average education. Column (2) introduces additional X controls. The magnitude of the DID estimates declines but remains statistically significant. The point estimates in Columns (3) and (4) show the treatment impact for boys and girls, respectively. The coefficient attached to the treatment interaction with young male cohort is negative and not statistically different from zero. Comparatively, the coefficient on the interaction term for young female-cohort in Column (3) is positive and statistically significant. Table 5: Schistosomiasis Programme Impact on Years of Education, Young Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 5: Schistosomiasis Programme Impact on Years of Education, Young Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 6–14 in 1999 0.45*** 0.34*** −0.06 0.70*** −0.07 0.62*** 0.08 1.11*** (0.22) (0.16) (0.15) (0.24) (0.37) (0.16) (0.17) (0.24) [0.01] [0.00] [0.76] [0.00] [0.78] [0.00] [0.13] [0.00] Age 6–14 in 1999 0.38* 0.61*** 0.03 1.15*** 0.58* 0.70*** 0.32 1.11*** (0.20) (0.16) (0.21) (0.23) (0.29) (0.18) (0.25) (0.27) [0.07] [0.00] [0.18] [0.00] [0.09] [0.01] [0.87] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. The results in Columns (5) and (6) separate the programme’s impact into urban and rural areas. The point estimate in urban areas is small and not significantly different from zero, implying that the treatment has no detectable effect in urban locations. In contrast, the effect of the programme is positive and statistically significant in rural areas. This is not surprising given that the programme was concentrated in rural areas. The young cohort in the treatment rural villages gained about an additional 0.62 year of education relative to those in control villages. In Columns (7) and (8), we show the estimates for the young males and females residing in rural areas. The estimate in Column (7) indicates the years of education for the males increased because of the programme. However, the magnitude is small, and the estimate is not statistically significant. Importantly, the females exposed to the treatment in the rural areas gained about one more year of schooling compared to females not exposed to the programme residing in rural areas of non-treated states. A possible explanation for the heterogeneity in the treatment impact could be that girls were more involved in domestic chores and thus more likely to be in contact with contaminated water. Hence, girls will be more likely to benefit from the treatment. 6. Robustness Checks In this section, we report two robustness checks. First, we carried out a placebo test using equation (4). Table 6 shows the results. None of the point estimates is statistically different from zero at conventional levels, except for the urban sample. It is noteworthy that the sign of the coefficient for the urban sample is negative, and probably the DID estimate for urban areas in Table 5 underestimates the impact. However, a mean reversion in urban areas cannot be ruled out. Nonetheless, as reported earlier, the DID estimate for urban areas is marginally negative and statistically insignificant precluding any positive impact in the urban sample. While the results presented in Table 6 do not prove the existence of similar trends in the hypothetical case of no programme, they indicate that our conclusions are unlikely to be driven by the pre-existing trend differential between treatment and control groups. Table 6: Schistosomiasis Programme Impact on Years of Education, Old Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 6: Schistosomiasis Programme Impact on Years of Education, Old Cohort All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × age 16–24 in 1999 0.07 0.12 −0.19 0.21 −0.47** 0.46 0.08 0.54 (0.23) (0.22) (0.32) (0.27) (0.23) (0.30) (0.37) (0.38) [0.81] [0.96] [0.67] [0.83] [0.03] [0.42] [0.73] [0.49] Age 16–24 in 1999 0.86*** 1.07*** 1.16** 1.23*** 1.35*** 1.01*** 1.08** 1.13*** (0.23) (0.21) (0.29) (0.28) (0.36) (0.21) (0.30) (0.27) [0.00] [0.00] [0.04] [0.00] [0.00] [0.00] [0.00] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 32,232 32,232 15,943 16,289 13,060 19,172 9451 9721 R2 0.31 0.47 0.49 0.46 0.28 0.47 0.53 0.43 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Second, the efforts against schistosomiasis coincided with the malaria-lymphatic filaris control programme, another large-scale health intervention. We demonstrate empirically that our findings are not driven by the malaria control programme. The malaria policy intervention to eliminate lymphatic filaris started with a pilot project in 2004 and operated mass drug administration of single dose treatment as well as the distribution of insecticide-treated bed nets to households in rural villages in Plateau and Nasarawa states (Blackburn et al., 2006). It could be possible that the programme has impacted child health and education and therefore introduced a bias in the estimates reported in Tables 5 and 6. Our data identified households that received treated bed nets. We thus estimate the effect of the malaria treatment programme on young children’s education. Table 7 shows the results. The estimates on the triple interaction term across all specifications are positive but not significantly different from zero at the conventional level. Hence, there is no strong evidence of any additional effects of the malaria campaigns on the schistosomiasis control programme. Table 7: Malaria Campaigns for Insecticidal Net and Schistosomiasis Treatment (Young Cohort) All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 7: Malaria Campaigns for Insecticidal Net and Schistosomiasis Treatment (Young Cohort) All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 All sample All sample Male only Female only Urban areas Rural areas Male in rural Female in rural (1) (2) (3) (4) (5) (6) (7) (8) Treated states × malaria × age 6–14 in 1999 0.07 0.30 0.07 0.26 −0.31 0.55 0.31 0.61 (0.52) (0.49) (0.46) (0.52) (0.39) (0.68) (0.55) (0.72) [0.86] [0.58] [0.92] [0.58] [0.34] [0.52] [0.60] [0.50] Malaria × age 6–14 in 1999 −0.15 0.17 0.19 0.17 0.12 0.18* 0.19 0.17* (0.20) (0.14) (0.18) (0.15) (0.18) (0.14) (0.22) (0.13) [0.41] [0.13] [0.27] [0.17] [0.39] [0.09] [0.29] [0.09] Treated states × age 6–14 in 1999 0.44** 0.31*** −0.06 0.66*** −0.04 0.56*** 0.06* 1.02*** (0.23) (0.17) (0.13) (0.29) (0.34) (0.17) (0.16) (0.29) [0.03] [0.00] [0.70] [0.00] [0.86] [0.00] [0.10] [0.00] Controls No Yes Yes Yes Yes Yes Yes Yes Observations 43,492 43,492 20,182 23,310 17,584 25,908 11,852 14,056 R2 0.34 0.47 0.45 0.50 0.23 0.48 0.47 0.49 Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. 6.1 Possible channels We have argued that this large-scale health programme has improved health outcomes of eligible children in treated states, and this in turn has allowed these children to accumulate more education. We do not have access to direct measures of health outcomes, especially for the relevant cohorts we use. The DHS, however, collects height and weight for women aged 15–49 years. As we mentioned, one of the consequences of schistosomiasis if left untreated is stunted growth. Hence, we can use the DHS data to explore the programme’s effects on height for women. Height is largely determined in childhood and thus would not be sensitive to current economic circumstances. We use a specification identical to equation (3), except that the dependent variable is height (in cm). Panel A of Table 8 shows the results. The point estimates indicate that in rural areas, women belonging to the treated cohorts gained 0.48 cm relative to similar cohorts in the comparison areas. The point estimates are significant at the 1% level. Noticeably, none of the estimates in Columns 1–3 (especially in urban areas) is statistically different from 0, and the results are quite consistent with those reported in Table 5: the gains are essentially concentrated for females in rural areas. While 0.48 cm may appear small in absolute value, it is in range of secular gains in heights reported in the literature (e.g., Cole, 2000; Fudvoye and Parent, 2017).15 Table 8: Schistosomiasis Programme Impact on Adult Females’ Height All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. Table 8: Schistosomiasis Programme Impact on Adult Females’ Height All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes All sample All sample Urban areas Rural areas (1) (2) (3) (4) Panel A: Young Cohort Treated states × age 6–14 in 1999 0.15 0.13 −0.33 0.48*** (0.23) (0.21) (0.50) (0.19) [0.53] [0.57] [0.61] [0.00] Age 6–14 in 1999 −1.94*** −1.92*** −1.69*** −2.04*** (0.29) (0.28) (0.37) (0.38) [0.00] [0.00] [0.00] [0.00] Observations 22,592 22,592 8930 13,662 R2 0.06 0.07 0.05 0.05 Panel B: Old Cohort Treated states × age 16–24 in 1999 −0.30 −0.24 −0.22 −0.25 (0.26) (0.26) (0.44) (0.38) [0.31] [0.37] [0.65] [0.50] Age 16–24 in 1999 −0.53* −0.54* −0.20 −0.71** (0.32) (0.31) (0.54) (0.30) [0.09] [0.09] [0.76] [0.03] Observations 15,777 15,777 6320 9457 R2 0.05 0.06 0.05 0.05 Panel C: Younger Cohort Treated states × age 0–5 in 1999 −0.54 −0.57 −0.44 −0.72 (0.51) (0.51) (0.52) (0.74) [0.35] [0.34] [0.44] [0.47] Age 0–5 in 1999 −4.64*** −4.71*** −3.59*** −5.36*** (0.44) (0.43) (0.68) (0.41) [0.00] [0.00] [0.00] [0.00] Observations 16,505 16,505 6633 9872 R2 0.11 0.12 0.11 0.11 Controls No Yes Yes Yes Notes: All the models include fixed effects for age, year and state. The additional control variables (X) include areas of residence, gender, household head age, head’s gender and father’s education. See equation (3) for details. Standard errors in parentheses are clustered at the state level. The brackets report P-values for zero null hypothesis derived through wild bootstrap clustered at the state level with 700 repetitions. *, **, and *** denote significance at the 10%, the 5%, and the 1% level, respectively. In Panel B of Table 8, we report robustness check using a placebo test estimated using equation (4). The point estimates are smaller in magnitude (relative to Panel A), and are statistically not different from 0. Hence, it is unlikely that the increase in height that we detect in Panel A is due to pre-existing differences in trends between treated and comparison states. Eventually, because adult heights are largely predetermined by age 5, nutritionists would interpret the results in Table 8 as only revealing that the cohort in the treatment areas had the wind at their backs from conditions they experienced before entering school (Lundeen et al., 2014; Leroy et al., 2015). We argue nevertheless that children’s physical growth could be significantly impacted after the age of 5 for two reasons. First, schistosomiasis affects preadolescent children more than adults who are yet to reach their full potential maturity. As documented, the infection results in reduced physical growth development both in height and weight of school-age children, with evidence of gender-based differences (Parraga et al., 1996; Orsini et al., 2001). Second, as discussed in Moradi (2010) and Gurarie et al. (2011), it is possible in some cases for the secular trend on growth to be reversed and for the child to make up for the shortfall in height after the age of 5. In Panel C of Table 8, we display the results of the cohort group aged 0 to 5 in 1999, children stunted from shocks experienced in utero and before age 5. The estimates are not significantly different from zero.16 7. Conclusion This paper examines the impact on schooling outcomes for children exposed to the treatment of schistosomiasis, a disease that causes anaemia, poor growth, and impaired cognitive function. Schistosomiasis is one of the largest endemic disease around the world. It is estimated that at least 90% of those requiring treatment live in Africa (WHO, 2015). Large-scale distribution of drugs is the most effective and low-cost mechanism to control the parasitic transmission, and to reduce and prevent morbidity. Our research approach takes advantage of the exogeneous variation generated from the CC’s rollout of schistosomiasis treatment in four states in Nigeria to implement a DID strategy to identify a causal relationship between the schistosomiasis control programme and the child’s enrolment and educational attainment. We do not find evidence that the programme has an impact on the enrolment of school-going children. However, we find that the cohort exposed to the treatment in rural areas accumulated an additional 0.62 years of education compared to the cohort not exposed to the treatment. Moreover, the impact of the schistosomiasis treatment is mainly on girls residing in rural areas. Consistent with the impact of education, we find that height of females exposed to the treatment in rural areas increased by 0.48 cm. Overall, the findings of this research demonstrate substantial gains in health following the mass drug administration of schistosomiasis doses. These gains in health translated, in turn, into more accumulation of education. The gains are the most substantial for females in rural areas. Acknowledgements We would like to thank the editor and two anonymous referees for their helpful suggestions. This paper has benefited from the insightful comments offered by participants of seminars at Department of Economics and Legal Studies at Oklahoma State University, and workshop participants at the African Finance and Economic Association meetings. This paper is not endorsed by the Carter Center. All errors and omissions that may remain are the sole responsibility of the authors. Footnotes 1 The Centres for Disease Control and Prevention (CDC) categorises the NTDs in endemic infections in tropical regions that not only affect the world’s poorest people but also cause disabilities that make it more difficult to succeed in school, care for family or earn a living. NTDs also predominantly occur among populations that have little or no access to good housing, safe water supplies and sanitation, or formal health systems. 2 The decline in mortality from several diseases in the twentieth century positively impacted life expectancy and population growth in the developing countries (Acemoglu and Johnson, 2007). In sub-Saharan Africa, for example, infections due to the TseTse fly affected the continent’s precolonial prospects in developing agricultural technologies (Alsan, 2014). Regions where the parasite is endemic became less likely to use domesticated animals, to use the plough, and ultimately to intensify agriculture. 3 Implications of large endemic disease on the development of low-income countries include inhibition of growth of specific geographic regions, low investment in human capital and missing opportunities in transfer of technologies and agriculture (see, Strauss and Thomas, 1998; Gallup and Sachs, 2001; Glewwe et al., 2001; Becker et al. 2005). 4 Attendance is defined as being reported as enrolled in school at the time of the survey. 5 The 1999 survey is not distributed to the public. Although the 1999 data were collected for women aged 10–49, the indicators were calculated for women aged 15–49. We were able to access the 1999 data under the disclaimer that the DHS do not stand behind the quality of the dataset for that particular reason. 6 Ebonyi and Enugu states were included in the CC programme in 2014, but since our data are up to 2013, these two states are part of non-treated states. 7 The official entrance age for the lowest level of education in Nigeria is 6-years old. Therefore, children aged 6–14 were not only likely to be enrolled in school at the time of the intervention but also able to receive the treatment. 8 Table 1 shows that treatment areas have slightly higher school enrolment rates in the baseline years, 69% versus 63%. Several factors could explain the difference in education levels. For instance, northern states in Nigeria have education levels below the national average (Umar et al., 2014). The location of our treatment states, more in the south, could partly justify the education gap. These differences are cultural and historical and therefore likely to persist over time. 9 Bertrand et al. (2004) showed that estimates and inference of difference-in-differences are sensitive to serial correlation when the data are extended to several periods, and they recommend aggregating the data in two periods of pre- and post-intervention. 10 Given the concerns about number of clusters, we also report the p-values of zero null hypothesis derived through wild bootstrap clustered at the state level as proposed by Cameron et al. (2008). We use the Stata programme cgmwildboot.ado written by Judson Caskey. 11 Assignment of individuals to states is based on the state of residence at the time of survey, i.e., 2013, and not on the state of residence during school age, i.e., during 1999–2008. This may be problematic in the presence of migration; however, only interstate migration is a concern, not within state. The interstate migration in Nigeria (for all population) was about 10% in 2006 (National Population Commission, 2010). 12 The data collected at the baseline show differences in school enrolment rates in the treatment and control areas. However, the DID estimation does not require a balance at the baseline. As long as these differences are persistent over time, the DID consistently estimates the treatment impact of the programme. 13 School enrolment rates would likely remain high for grades that include children between 6–11 years of age, that is, for those at the level of education before secondary school (Glewwe and Kremer, 2006). We show the tests for children aged 6–9 and 10–14 in Columns (2) and (3), respectively. The treatment effect on school enrolment peaks in 2008 for the 6–9 group and in 2003 for the 10–14 group. We do not find any statistical significance in the estimates when breaking down the sample for these two age groups. 14 Although we do not have documentation of where the distribution of the drugs took place, almost all the reports we came across are centred around schools and/or school children. Thus, it is likely that schools played a central role in the distribution of the drugs. 15 Cole (2000) mentions height gains ranging from 3 mm per decade in Scandinavia to 30 mm/decade in parts of Southern and Eastern Europe during latter half of the 20th century. The data reported in Fudvoye and Parent (2017) imply that height gains ranged from 3.7 mm per decade in Portugal to 15.1mm per decade in the Netherlands, between 1880 and 1980. 16 We run additional tests for children of age 0–2 and then 3–5 (results not shown). The point estimates are not significantly different from zero and do not suggest a growth deficit at early childhood for these cohort groups. Using equation (1), we also compare the change in height-for-age Z-scores (HAZ) for children under 59 months. The pre-intervention period includes only year 1990; the DHS did not collect HAZ for children in 1999. The results are, therefore, presented in the Appendix Table 9 in Supplementary material. We do not detect any sizable difference in growth stunting in children under 5. References Abdulkadir A. , Ahmed M. , Abubakar B. M. , Suleiman I. E. , Yusuf I. , Imam I. M. , Sule A. A. , Tela U. M. , Dogo H. M. , Yakasai A. M. , Musa B. M. ( 2017 ) ‘ Prevalence of Urinary Schistosomiasis in Nigeria, 1994–2015: Systematic Review and Meta-Analysis ’, African Journal of Urology , 23 : 228 – 39 . Google Scholar Crossref Search ADS Acemoglu D. , Johnson S. ( 2007 ) ‘ Disease and Development: The Effect of Life Expectancy on Economic Growth ’, Journal of Political Economy , 115 ( 6 ): 925 – 85 . Google Scholar Crossref Search ADS Agi P. , Okafor E. ( 2005 ) ‘ The Epidemiology of Schistosoma Haematobium in Odau Community in the Niger Delta Area of Nigeria ’, Journal of Applied Sciences and Environmental Management , 9 ( 3 ): 37 – 43 . Alsan M. ( 2014 ) ‘ The Effect of the Tsetse Fly on African Development ’, The American Economic Review , 105 ( 1 ): 382 – 410 . Google Scholar Crossref Search ADS Ayoya M. A. , Spiekermann-Brouwer G. M. , Traoré A. K. , Garza C. ( 2012 ) ‘ Effect on School Attendance and Performance of Iron and Multiple Micronutrients as Adjunct to Drug Treatment of Schistosoma-Infected Anemic Schoolchildren ’, Food and Nutrition Bulletin , 33 ( 4 ): 235 – 41 . Google Scholar Crossref Search ADS PubMed Babatunde T. , Asaolu S. , Sowemimo O. ( 2013 ) ‘ Urinary Schistosomiasis Among Pre-School and School Aged Children in Two Peri-Urban Communities in Southwest Nigeria ’, Journal of Parasitology and Vector Biology , 5 ( 7 ): 96 – 101 . Becker G. S. , Philipson T. J. , Soares R. R. ( 2005 ) ‘ The Quantity and Quality of Life and the Evolution of World Inequality ’, The American Economic Review , 95 ( 1 ): 277 – 91 . Google Scholar Crossref Search ADS PubMed Bertrand M. , Duflo E. , Mullainathan S. ( 2004 ) ‘ How Much Should We Trust Differences-in-Differences Estimates? ’, The Quarterly Journal of Economics , 119 ( 1 ): 249 – 75 . Google Scholar Crossref Search ADS Blackburn B. G. , Eigege A. , Gotau H. , Gerlong G. , Miri E. , Hawley W. A. , Mathieu E. , Richards F. ( 2006 ) ‘ Successful Integration of Insecticide-Treated Bed Net Distribution with Mass Drug Administration in Central Nigeria ’, The American Journal of Tropical Medicine and Hygiene , 75 ( 4 ): 650 – 5 . Google Scholar Crossref Search ADS PubMed Bleakley H. ( 2007 ) ‘ Disease and Development: Evidence from Hookworm Eradication in the American South ’, The Quarterly Journal of Economics , 122 ( 1 ): 73 – 117 . Google Scholar Crossref Search ADS PubMed Bleakley H. ( 2010 ) ‘ Malaria Eradication in the Americas: A Retrospective Analysis of Childhood Exposure ’, American Economic Journal: Applied Economics , 2 ( 2 ): 1 – 45 . Google Scholar Crossref Search ADS Bloom D. E. , Canning D. , Sevilla J. ( 2004 ) ‘ The Effect of Health on Economic Growth: A Production Function Approach ’, World Development , 32 ( 1 ): 1 – 13 . Google Scholar Crossref Search ADS Cameron A. C. , Gelbach J. B. , Miller D. L. ( 2008 ) ‘ Bootstrap-Based Improvements for Inference with Clustered Errors ’, The Review of Economics and Statistics , 90 ( 3 ): 414 – 27 . Google Scholar Crossref Search ADS Cole T. J. ( 2000 ) ‘ Secular Trends in Growth ’, Proceedings of the Nutrition Society , 59 ( 2 ): 317 – 24 . Google Scholar Crossref Search ADS PubMed Cutler D. , Fung W. , Kremer M. , Singhal M. , Vogl T. ( 2010 ) ‘ Early-Life Malaria Exposure and Adult Outcomes: Evidence from Malaria Eradication in India ’, American Economic Journal: Applied Economics , 2 ( 2 ): 72 – 94 . Google Scholar Crossref Search ADS Ekanem E. , Asindi A. , Ejezie G. , Antia-Obong O. ( 1994 ) ‘ Effect of Schistosoma Haematobium Infection on the Physical Growth and School Performance of Nigerian Children ’, The Central African Journal of Medicine , 40 ( 2 ): 38 – 44 . Google Scholar PubMed Ezeadila J. O. , Okoli I. , Agomuo M. , Aneke F. C. , Egbuche C. M. ( 2015 ) ‘ Prevalence of Urinary Schistosomiasis Among Community Primary School Pupils in Amagunze, Enugu State, Nigeria ’, World Rural Observations , 7 ( 3 ): 46 – 51 . Fudvoye J. , Parent A.-S. ( 2017 ) ‘Secular trends in growth’, in Annales d’Endocrinologie , Vol. 78 . Issy-les-Moulineaux, France : Elsevier , pp. 88 – 91 . Gallup J. L. , Sachs J. D. ( 2001 ) ‘ The Economic Burden of Malaria ’, The American Journal of Tropical Medicine and Hygiene , 64 ( 1 ): 85 – 96 . Google Scholar Crossref Search ADS PubMed Glewwe P. , Jacoby H. G. , King E. M. ( 2001 ) ‘ Early Childhood Nutrition and Academic Achievement: A Longitudinal Analysis ’, Journal of Public Economics , 81 ( 3 ): 345 – 68 . Google Scholar Crossref Search ADS Glewwe P. , Kremer M. ( 2006 ) ‘ Schools, Teachers, and Education Outcomes in Developing Countries ’, Handbook of the Economics of Education , 2 : 945 – 1017 . Google Scholar Crossref Search ADS Gurarie D. , Wang X. , Bustinduy A. L. , King C. H. ( 2011 ) ‘ Modeling the Effect of Chronic Schistosomiasis on Childhood Development and the Potential for Catch-up Growth with Different Drug Treatment Strategies Promoted for Control of Endemic Schistosomiasis ’, The American Journal of Tropical Medicine and Hygiene , 84 ( 5 ): 773 – 81 . Google Scholar Crossref Search ADS PubMed Gutman J. , Fagbemi A. , Alphonsus K. , Eigege A. , Miri E. , Richards F. Jr ( 2008 ) ‘ Missed Treatment Opportunities for Schistosomiasis Mansoni, in an Active Programme for the Treatment of Urinary Schistosomiasis in Plateau and Nasarawa States, Nigeria ’, Annals of Tropical Medicine and Parasitology , 102 ( 4 ): 335 – 46 . Google Scholar Crossref Search ADS PubMed Hopkins D. R. , Eigege A. , Miri E. S. , Gontor I. , Ogah G. , Umaru J. , Gwomkudu C. C. , Mathai W. , Amadiegwu S. , Oyenekan O. K. , et al. ( 2002 ) ‘ Lymphatic Filariasis Elimination and Schistosomiasis Control in Combination with Onchocerciasis Control in Nigeria ’, The American Journal of Tropical Medicine and Hygiene , 67 ( 3 ): 266 – 72 . Google Scholar Crossref Search ADS PubMed Hopkins D. R. , Richards F. O. Jr , Ruiz-Tiben E. , Emerson P. , Withers C. Jr , et al. ( 2008 ) ‘ Dracunculiasis, Onchocerciasis, Schistosomiasis, and Trachoma ’, Annals of the New York Academy of Sciences , 1136 ( 1 ): 45 – 52 . Google Scholar Crossref Search ADS PubMed Kazianga H. , Masters W. A. , McMillan M. S. ( 2014 ) ‘ Disease control, demographic change and institutional development in Africa ’, Journal of Development Economics , 110 : 313 – 26 . Google Scholar Crossref Search ADS Leroy J. L. , Ruel M. , Habicht J.-P. , Frongillo E. A. ( 2015 ) ‘ Using Height-for-Age Differences (HAD) Instead of Height-for-Age Z-Scores (HAZ) for the Meaningful Measurement of Population-Level Catch-up in Linear Growth in Children Less than 5 Years of Age ’, BMC Pediatrics , 15 ( 1 ): 145 – 56 . Google Scholar Crossref Search ADS PubMed Lundeen E. A. , Stein A. D. , Adair L. S. , Behrman J. R. , Bhargava S. K. , Dearden K. A. , Gigante D. , Norris S. A. , Richter L. M. , Fall C. H. , et al. ( 2014 ) ‘ Height-for-Age Z Scores Increase Despite Increasing Height Deficits Among Children in 5 Developing Countries ’, The American Journal of Clinical Nutrition , 100 ( 3 ): 821 – 5 . Google Scholar Crossref Search ADS PubMed Meremikwu M. , Asuquo P. , Ejezie G. , Useh M. , Udoh A. ( 2000 ) ‘ Treatment of Schistosomiasis Haematobium with Praziquantel in Children: Its Effect on Educational Performance in Rural Nigeria ’, Tropical Medicine , 42 ( 1 ): 39 – 45 . Miguel E. , Kremer M. ( 2004 ) ‘ Worms: Identifying Impacts on Education and Health in the Presence of Treatment Externalities ’, Econometrica , 72 ( 1 ): 159 – 217 . Google Scholar Crossref Search ADS Moradi A. ( 2010 ). Selective mortality or growth after childhood? What really is key to understand the puzzlingly tall adult heights in sub-SaharanAfrica. CSAE Working Paper Series 2010–17, Centre for the Study of African Economies, University of Oxford. National Population Commission ( 2010 ) Population and Housing Census of the Federal Republic of Nigeria 2006, Government Publication . Abuja, Nigeria : National Population Commission . Orsini M. , Rocha R. S. , Disch J. , Katz N. , Rabello A. ( 2001 ) ‘ The role of Nutritional Status and Insulin-Like Growth Factor in Reduced Physical Growth in Hepatosplenic Schistosoma Mansoni Infection ’, Transactions of the Royal Society of Tropical Medicine and Hygiene , 95 ( 4 ): 453 – 6 . Google Scholar Crossref Search ADS PubMed Parraga I. M. , Assis A. M. O. , Prado M. S. , Barreto M. L. , Reis M. G. , King C. H. , Blanton R. E. ( 1996 ) ‘ Gender Differences in Growth of School-Aged Children with Schistosomiasis and Geohelminth Infection ’, The American Journal of Tropical Medicine and Hygiene , 55 ( 2 ): 150 – 6 . Google Scholar Crossref Search ADS PubMed Sam-Wobo S. , Idowu J. , Adeleke M. , et al. ( 2011 ) ‘ Urinary Schistosomiasis Among Children and Teenagers near Oyan Dam, Abeokuta, Nigeria ’, Journal of Rural and Tropical Public Health , 10 : 57 – 60 . Strauss J. , Thomas D. ( 1998 ) ‘ Health, Nutrition, and Economic Development ’, Journal of Economic Literature , 36 ( 2 ): 766 – 817 . The Carter Center ( 2000 ). The Carter Center Annual Report 1998–1999. https://www.cartercenter.org/documents/520.pdf. Umar H. M. , Ismail R. , Abdul-Hakim R. ( 2014 ) ‘ Regional Inequality of Educational Attainment in Nigeria ’, British Journal of Economics, Management and Trade , 4 ( 3 ): 420 – 30 . Google Scholar Crossref Search ADS Venkataramani A. S. ( 2012 ) ‘ Early Life Exposure to Malaria and Cognition in Adulthood: Evidence from Mexico ’, Journal of Health Economics , 31 ( 5 ): 767 – 80 . Google Scholar Crossref Search ADS PubMed Weil D. N. ( 2007 ) ‘ Accounting for the Effect of Health on Economic Growth ’, The Quarterly Journal of Economics , 122 ( 3 ): 1265 – 1306 . Google Scholar Crossref Search ADS Weil D. N. ( 2010 ) ‘ Endemic Diseases and African Economic Growth: Challenges and Policy Responses ’, Journal of African Economies , 19 ( 3 ): 81 – 109 . World Health Organization ‘ Schistosomiasis: Number of People Treated Worldwide in 2013 ’, Weekly Epidemiological Record , 90 ( 5 ): 25 – 32 . © The Author(s) 2018. Published by Oxford University Press on behalf of the Centre for the Study of African Economies, all rights reserved. For Permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Journal of African EconomiesOxford University Press

Published: Nov 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off