This paper studies variance estimators in panel data settings. There has been a recent surge in research on panel data models with a number of new estimators proposed. However, there has been less attention paid to the quantification of the precision of these estimators. Of the variance estimators that have been proposed, their relative merits are not well understood. In this paper we develop a common framework for comparing some of the proposed variance estimators for generic point estimators. We reinterpret three commonly used approaches as targeting different conditional variances under an exchangeability assumption. We find that the estimators we consider are all valid on average, but that their performance in terms of power differs substantially depending on the heteroskedasticity structure of the data. Building on these insights, we propose a new variance estimator that flexibly accounts for heteroskedasticity in both the unit and time dimensions, and delivers superior statistical power in realistic panel data settings.

This paper studies the long-term impacts of dual apprenticeships on youth employment in a high-informality labor market. A Randomized Controlled Trial in Côte d'Ivoire with four follow-up surveys collected over seven years shows that dual apprenticeships have sustained impacts: youth earnings increase by 14 to 20 percent two to five years after program completion. Gains are observed across the earnings distribution, and the share of youth in working poverty—with earnings below the minimum wage—decreases by 11 percent. Importantly, results highlight a distinct pathway whereby training raises earnings through self-employment, with no impact on access to wage employment. Youth perform more complex, non-routine tasks, consistent with improved technical skills and productivity. In a setting where formal wage jobs are rare, these findings show that dual apprenticeships constitute an effective and inclusive skilling model that raises earnings in self-employment and reduces working poverty.

Can small competing firms share procurement scale, and does cooperation among direct rivals change downstream market outcomes? We randomly reduce the cost of cooperation among clustered retailers in urban Côte d'Ivoire. Experimental evidence demonstrates that correcting pessimistic second-order beliefs about peers increases participation by 14 percentage points. Facilitating a meeting for participants to discuss terms and initiate group operations further doubles this effect, resulting in a 30 pp increase. The introduction of an additional third-party monitor provides no additional effect. These effects persist for 12 months, during which the practice of collective sourcing also diffuses to the control group. Drawing on a hand-collected retailer × product panel, treated retailers experience a 10.1 percent reduction in wholesale unit costs, a 5.8 percent decrease in retail prices, and a 22 percent expansion in two-week assortment. Finally, a two-stage structural model with endogenous activation sheds light on how newly formed group members internalize group members' quantity decisions. The model also quantifies the cost-sharing mechanisms through which both retailers and consumers benefit from group formation.

In early March 2020, a few cases of COVID-19 were diagnosed in Abidjan, the capital city of Côte d'Ivoire. To combat the spread of the disease, major restrictions to mobility and gatherings were introduced between mid-March and late May 2020. We collected panel survey data on over 2,500 individuals from the Greater Abidjan area over the period immediately before and after the start of the pandemic. We document striking drops in employment, hours worked, income, and food consumption in the first months after the onset of COVID-19, when lockdown was in place. We also find that, in response, survey respondents received more private transfers from other parts of the country, at a time when remittances from abroad fell — and that some respondents moved either temporarily or permanently. In terms of recovery, we find that subjective well-being was lower on average in December 2020 than at baseline. Yet, despite schools being closed between mid-March and July 2020, school enrollment suffered little: by December 2020, enrollment rates had returned to baseline. Our results indicate that government policies aimed at alleviating the worst effects of lockdown only reached a few people, and not necessarily those most in need.

This report provides summary statistics on the 2019 household listing and the 2019-2020 individual survey conducted under the auspices of AUDRI in the Greater Abidjan region of Côte d'Ivoire. The household listing and the individual surveys were undertaken in collaboration with IPA-Côte d'Ivoire and with funding from the Stanford King Center on Global Development.