RESEARCH

WORKING PAPERS

Endogenous Sample Selection [paper]

This paper examines enumerator incentives in the production of survey data, a crucial input to research and policymaking. In theory, survey data is generated from a randomly selected, representative sample of the population. We provide causal evidence that in practice, enumerators respond to variation in effort cost across survey subjects by excluding high-cost subjects. To this end, we exploit the random assignment of individual questionnaires across 3.4 million households in 181 Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS), two prominent global household survey programs, as a source of exogenous variation in effort cost. We find that in 110 (39) of these surveys at least 5% (10%) of survey subjects eligible for individual questionnaires are missing from the sample. Missing individuals differ systematically from included individuals. As a result, survey samples are not representative of the population, causing bias in key aggregate statistics. In particular, we show that the fertility transition in Africa - where few alternative data sources on fertility exist - is likely to have been significantly faster than previously reported, leading to lower fertility levels today. Finally, we provide suggestive evidence that endogenous sample selection affects a wide range of surveys including living standards, labor force and firm surveys - thereby highlighting the broader relevance of the novel quantity-quality trade-off in data collection documented in this paper.

When Does Electrification Work? Evidence from Sub-Saharan Africa [paper]

(with Niclas Moneke)

Economists have long studied the underprovision of public goods. Electrification is highly positively correlated with economic development. However, causal evidence points to modest average effects of recent electrification programs in low- and middle-income countries at best. This paper unveils stark differences in the impact of grid access across locations underlying these average effects. First, we document that in rural Sub-Saharan Africa, the world’s least electrified region, most locations with grid access have either near zero or almost full electricity adoption. In other words, modest average connection rates mask a bimodal distribution across locations. Second, we empirically test for a potential determinant of successful electrification identified prominently in historical accounts and modern engineering plans: the productive use of electricity. Exploiting a large-scale, country-wide rural electrification program in Zambia, we provide three findings. First, the aggregate effects of rural electrification are mixed at best, confirming existing evidence. Second, locations with pre-existing productive uses experience almost full household electrification in contrast to near zero adoption elsewhere. Third, any development effects of rural electrification are confined to these same locations. We interpret the bimodal adoption and development pattern as reflective of the high fixed cost of the local electric network, and provide suggestive evidence that productive users help overcome the resulting ‘last mile problem’.

Misallocation in the Public Sector? Cross-Country Evidence from Two Million Primary Schools [paper]

This paper examines the allocation of teachers across public primary schools and assesses its relevance for aggregate learning levels. Harmonizing administrative data on the universe of public primary schools from 85 countries, I build a new global school-level data set that comprises nearly two million schools across 27,691 administrative units. I document that pupil-teacher ratios (PTRs) in high-income countries are low on aggregate and vary little between schools. In contrast, in poor countries aggregate PTRs are high and differences in PTRs between schools are large. Even at the local level, within second-tier administrative units, differences are substantial. Moreover, PTRs are only weakly negatively correlated with local levels of population density, wealth and education. To assess the implications of these facts for aggregate educational outcomes, I leverage data from an experiment in Kenya that assigned an extra teacher to randomly selected primary schools. I show that the effect of adding a teacher on test scores is larger at schools with higher baseline PTRs, thus suggesting that reallocating teachers from low- to high-PTR schools could not only reduce inequality of opportunity but also increase aggregate learning. [Old version available here: STICERD Working Paper No. EOPP 70]

Using Data for Development: Evidence from a Phone System for Agricultural Advice [paper] - R&R World Development

(with Michael Kremer, Ofir Reich, Zhengyun Sun, Sam van Herwaarden, Habtamu Yesigat, Elias Nure and Temesgen Gebeyehu)

We study the performance of a governmental phone-based agricultural extension system in a developing country which provides agricultural advice to farmers via interactive voice response (IVR). Through the analysis of administrative data generated by the system, we identify potential for system improvements. Based on this, we develop specific interventions to increase system performance and test these in randomized control trials. We find that small changes in the design of the interactive voice response system can have substantial effects on the amount of agricultural advice accessed by farmers.

Can Brick Phones Bridge the Digital Learning Divide? Evidence from SMS-based Math Practice [paper]

(with Guthrie Gray-Lobe and Sarah Kabay)

Hardware requirements are a barrier to widespread adoption of digital learning software among low-income populations. We investigate the demand among smallholder-farming households for a simple, adaptive math learning tool that can be accessed by “brick” phones, and its effect on educational outcomes. Over a quarter of invited households used the tool, with greater demand among households lacking electricity, radios, or televisions, and greater usage when schools were out of session. Engagement lapses without regular reminders to use the service. Using random variation in access to the service, we find some evidence that the platform increased test scores, school attendance, and grade attainment. Interpretation of these estimates is complicated by potentially endogenous outcome observation.

WORK IN PROGRESS

Seasonal Employment and Development (with Niclas Moneke)

Clean Air in the Classroom (with Nitin Bharti, David Blakeslee, Samreen Malik and Talha Naeem)