I. Introduction

Anemia among Primary School Age Children in China

Despite rapid economic development in China, the most recent national study (conducted in 2002) found that 25% of children in rural China are sufficiently iron deficient to be classified as anemic (MoH, MoST and NBS, 2004). Regional studies report even higher anemia rates in poorer parts of the country (40% of junior high school students in Gansu Province and 50-60% of students in Guizhou Province) (Wang, 2007, Chen et al., 2005).

The evidence indicates that anemia is quite common among school children in poor areas of rural China. According to a study by the Xi’an Jiaotong University School of Medicine (our collaborators for this project), as many as 39 percent of children in rural elementary schools in randomly selected poor areas have anemia. There are more than 700 poor rural counties in China with more than 300 million people (of which nearly 100 million are children). Our study areas are located in Northwest China, one of China’s poorest areas. And while we cannot be sure that these counties are representative of all poor areas in China, the sample schools were selected in a way that make them representative of rural Northwest China. If anemia is also present in the rest of China’s poor areas at rates similar to those found in Northwest China, it is possible that there are between 20 and 30 million children that are affected nationwide.

In light of the existence of effective technologies and sufficient resources, we hypothesize that the absence of programs to reduce anemia through schools (which are natural contact points with the entire child population) are largely attributable to informational deficiencies – and, perhaps more importantly, to weak supply-side incentives. Despite new information about the magnitude of anemia, school principals may still not be aware of the severity of the problem and its consequences for education (Shi et al., 2008). More critically, principals (who have executive decision-making authority in their schools) may also have weak incentives to address anemia in their student bodies. China’s primary school principals currently face incentives to prevent students from dropping out, to maximize school attendance, and to improve their students’ scores on annual standardized tests. Although anemia reduction may improve these performance metrics indirectly, principal effort undoubtedly focuses on bolstering them in more traditionally understood ways.

REAP has proposed to test these hypotheses by executing a large-scale policy experiment in rural primary schools in China. In doing so, we will provide more general insight into the full-potential of supply-side incentives for health improvement—and, in particular, the merits of directly rewarding the ultimate outcome of interest: health.

II. Baseline Survey

Sample Selection

To assess the performance of high-powered supply-side incentives in reducing anemia among primary school age children in China, we propose to conduct a cluster-randomized field experiment. The first step is to identify, through a canvas survey, a population of schools in our designated provinces (see below) in which we will be running the study. During the canvas survey enumerators will identify all rural elementary schools that meet our minimum size requirement. Schools are required to: a.) be six-grade schools (complete primary schools, or wanxiao); b.) have boarding facilities; c.) have a minimum of 400 students. We will use information conducted during the canvas survey to randomly choose the sample schools. We will use the canvas survey and a baseline survey before assigning the individual schools to their different “Treatments” (or assigning individual schools as control schools). As stated above, we will do a balance test to show if the 6 different sets of schools are statistically “identical” before the interventions.

We have chosen to run the study in three of China’s poorest provinces—Gansu, Ningxia and Qinghai. We have randomly chosen a group of 20 counties from a list of China’s nationally designated poverty counties (China County Economy Statistical Yearbook, 2007; about 1 in 4 counties in northwest China are officially designated as “poor”). The main implementing partner-coordinator in the three Northwest Provinces is Northwest University (Xi’an).

Cognitive Testing and Background Surveys of Primary School Children

Within our 75 sample schools in Gansu, Ningxia, and Qinghai provinces, our collaborators at the Chinese Academy of Sciences and Northwest University (of Xi’an) will administer a series of cognitive tests to all participating students in the study. The tests, developed by experts in Beijing, measure executive functions, math skills, attention span, concentration, and other cognitive skills, in addition to psychological, physical, and social health. The results of these tests will serve as baseline indicators to be compared with results of a similar battery of tests administered upon the completion of the study.

Our collaborators will also survey participating students and their families in their homes. These surveys will collect information about socio-economic characteristics and living characteristics; self-reported information about health status, health behaviors, health input use, and parental health knowledge; nutritional characteristics of household meals (consumption of meat, beans, fruits, and vegetables, for example) and parental knowledge about nutrition; and information about school diets, anemia-related school activities, and school characteristics generally (class size, infrastructure, etc.). All information gathered will be used to assess pre-intervention balance in observable characteristics across experimental conditions. Socio-economic characteristics of families and school characteristics will also be used to (possibly) improve the precision of our estimation.

In addition to shedding light on private behavioral responses to our interventions, survey data collected from school children (about school diet and other anemia-related school activities such as the provision of vitamins, for example) will also help us to assess precisely how principals and primary schools respond to our interventions. These child reports will provide evidence that compliments our other measures of principal and school behavioral responses (such as school budgets and school expenditure records as well as principal surveys) to reduce anemia.

Hemocue Testing and Anthropometric Measurement of Children

In conjunction with our surveys of children and their households, a collaborating nursing team from Xi’an Jiaotong University School of Medicine will also collect physiological measures of health. Specifically, these will include hemocue finger-stick tests of hemoglobin counts to measure anemia. In addition to providing primary measures of the effectiveness of our various interventions, the hemocue tests will also provide the basis for our performance incentive rewards to principals. Anthropometric measures including height and weight will also be taken by the nursing team (allowing us to construct standard body-mass index measures and to standardize growth).

School Principal Surveys and School Budget and Expenditure Records

Finally, our collaborators at the Chinese Academy of Sciences and Northwest University (of Xi’an) will conduct a survey of all sample school principals. This survey will serve a variety of purposes. First, it will provide measures of principal’s knowledge and understanding of anemia and how it can be treated and prevented. Second, it will collect detailed information about principals’ contractual terms of employment, including base salaries and the size of the bonuses (both available and received) in preceding years. Third, in the follow-up wave of our survey of principals, the enumerators will ask a variety of direct questions about actions taken by principals to reduce anemia (both open- and closed-ended questions). Fourth, our survey teams will collect information about other general school characteristics (such as size, number of classes, educational attainment of teachers, etc.). Finally, at the time of the principal survey, we will also collect information about each school’s budgetary revenues and expenditures (including line items related to diet and anemia-related activities).

III. Intervention

In this section, we present each experimental condition and its rationale.

Control Condition. For the sake of comparison, we will randomly assign some participating schools to a control condition. In these schools, we will measure anemia among students but will not otherwise intervene.

1. Information Only. Our first randomly-assigned intervention is the provision of information to primary school principals about the share of enrolled students who are anemic and about efficacious methods for reducing anemia (including vitamin supplementation and other dietary changes). In addition, we will also teach principals about the close link between anemia, school attendance and educational performance according to findings published in the academic literature.

2. Information + Earmarked Operating Budget Subsidy. Because purchasing inputs to reduce anemia may not be a priority (or may not be possible) for school principals given current operating budgets, we will randomly assign some schools to receive earmarked operating budget subsidies for the purchase of relevant inputs to alleviate anemia (i.e., vitamin supplements and school lunch ingredients). These earmarked subsidies will be provided in addition to the information described above.

3. Information + Earmarked Operating Budget Subsidy + Anemia Reduction Incentive. Finally, we will randomly assign some schools to receive information, operating subsidies and explicit incentive payments to principals for reducing anemia among enrolled students (in addition to providing information about anemia rates among students and methods of reducing them). Given the governance structure of primary schools in China described earlier (school principals generally make executive decisions about school functions), anemia incentive payments will be incremental monetary rewards paid directly to principals for each percentage point reduction in the anemia rate among enrolled students.

We base our preliminary performance incentive reward calculations on average anemia rates in Gansu, Ningxia and Qinghai provinces and average bonuses currently received by principals for good attendance and test score performance (the equivalent of 1.5 months of salary). Our tentative incremental payments will provide the equivalent of 3 months of salary for a 10 percentage point anemia reduction. Implied performance payments are therefore 0.30 months of salary per percentage point reduction in anemia prevalence.

IV. Evaluation

At the conclusion of the study period, we will conduct a repeat of the baseline testing and surveys. The impacts on the anemia rate among each group of schools will be evaluated and compared between the four sub-groups. We also will document the correlation between changes in the anemia levels and changes in educational outcomes (e.g., grades, attendance, cognitive skills), school-related behaviors (such as changes in attention span and concentration) and psychological, social and physical health. The evaluation will also delve deeper into the relationship between improved attention span following reduced anemia and the impact on education attainment. Importantly, the evaluation will provide evidence on the effectiveness (including cost-effectiveness) of using Pay for Performance incentive mechanisms (in lieu of direct monitoring) to deliver health inputs, to realize health outcomes and to achieve educational impacts.

V. Results

This study is classified as ongoing.