Research

WORKING PAPERS

The politics of public service reform: Experimental evidence from Liberia – Under review
This paper provides experimental evidence on the electoral effect of a large education reform in a developing democracy. Despite significantly improving school quality, the policy reduced the incumbent party’s presidential vote share by 3 percentage points (10%). This does not imply that voters fundamentally oppose service improvements: household surveys showed strong support for the policy, and variation in school-pair-level treatment effects shows that the more the policy raised test scores, the more it increased incumbent vote share. Instead, the negative average electoral effect was driven by opposition from teachers. The policy reduced teachers’ job satisfaction, their support for the incumbent government, and their political engagement. The more the policy reduced teacher political engagement, the more it reduced incumbent vote share. Counterfactual simulations suggest that relatively small improvements in effectiveness and/or teacher engagement could have made the policy a net vote winner. This paper empirically demonstrates the importance of political feasibility in the design of public service reforms.
VoxDev column – Medium post – CESifo Working Paper 10633 – SSRN working paper

Secondary school access raises primary school achievement
I study the effect of Tanzania’s Free Secondary Education (FSE) program on primary student learning. Using the universe of grade 7 (primary) and 9 (secondary) standardized test scores, as well as household survey data, I confirm that the reform increased secondary school access: secondary enrollments rose, household education spending fell, primary scores predicted secondary transition better, and the transition rate advantage for local politicians’ children disappeared. To estimate the causal effect of increased secondary access on primary student learning, I exploit variation in family financial constraints using microdata on pre-reform grade 9 exam fee payments. This difference-in-difference estimation shows that the reform increased grade 7 exam pass rates and transition rates by about 3 percentage points each (passing the grade 7 exam is required to enter public secondary school). Effects along the distribution of test scores for each exam subject suggest that students responded by optimally allocating effort to clear the passing threshold. Using ward-level variation in treatment exposure, I show that the results are not driven by school entry or changes to the size of primary education cohorts. This shows that expansions in secondary-school supply affect primary-school demand, and that free secondary education programs may have wider benefits than previously understood.

Priming the pump: Can paying interest upfront increase savings?
(with Peter Carroll, Flora Myamba, Daniel Nielson, Joseph Price, and Phillip Roessler)
We experimentally test whether people are more likely to reach a savings target when interest is paid as a conditional upfront bonus, rather than after the savings target has been reached. We partner with a mobile network operator (MNO) and an NGO which provides cash transfers to women in rural Tanzania. We invited participants to keep at least 10% of their cash transfers in their mobile wallet until a certain date, 5 months later. We randomized whether doing so would let them keep an “upfront” bonus provided at the start, earn an equivalent sum of “interest” at the end, or receive no bonus but only “info” about saving. We measure outcomes using daily mobile money wallet balances provided by the MNO. The “upfront” group were more likely to reach the savings target (net of the bonus), not just by the end but also throughout the study period. They were less likely to withdraw the entire cash transfer immediately, and more likely to make deposits during the study period. The traditional “interest” treatment, despite offering an above-market interest rate, had no effect on meeting the target or on deposits and withdrawals. Neither treatment had any effect on noisy average mobile wallet balances, but the “upfront” group carried higher median net mobile wallet balances one month into the program. However, after regulatory disputes prevented th NGO from making an expected cash transfer during the study period, many participants withdrew all money – including the upfront bonus – such that the upfront group’s median net balance at the study’s end was negative, and lower than the other groups’. We show that the design of financial products can affect savings behavior.
SSRN working paper – IPA summary

The road to reelection: Political returns to highway construction
(with Daniel Leff and Alejandro Nakab)

publications

Outsourcing Education: Experimental Evidence from Liberia
(with Mauricio Romero and Justin Sandefur)
American Economic Review, Volume 110, No. 2, 2020
In 2016, the Liberian government delegated management of 93 randomly selected public schools to private providers. Providers received US$50 per pupil, on top of US$50 per pupil annual expenditure in control schools. After one academic year, students in outsourced schools scored 0.18σ higher in English and mathematics. We do not find heterogeneity in learning gains or enrollment by student characteristics, but there is significant heterogeneity across providers. While outsourcing appears to be a cost-effective way to use new resources to improve test scores, some providers engaged in unforeseen and potentially harmful behavior, complicating any assessment of welfare gains.
AEA registry

Work in progress