Jeffrey Clark

Selected work

1. 2026 Forthcoming Stata Journal

   autolabel: Automating variable and value labeling in Stata

   Jeffrey Clark · Jie Wen

   Register-based datasets from national statistical agencies arrive with cryptic variable names, unlabeled coded values, and documentation that lives separately from the data, often only in the national language. The autolabel command automates variable and value labeling by matching dataset variables to centralized, multilingual CSV metadata repositories and applying labels in a single call. A deferred execution pattern writes labeling commands to a temporary do-file during dataset inspection and applies them after the original state is restored, letting a single user-level command both inspect and apply. The domain-agnostic CSV schema lets any institution author its own metadata bundle.

   Show more
   Project Code
2. 2022 MSc thesis Stockholm School of Economics

   Rural Road Improvements and Local Agricultural Intensification: A Remote Sensing Evaluation in Mozambique

   Jeffrey Clark · Tamina Matti

   Transaction costs serve as an obstacle to competitive market exchanges in rural and remote areas around the world. Improvements to transportation infrastructure are hypothesized to lower these costs and help alleviate poverty among smallholder farmers. Yet, few empirical studies estimate the effect of improved rural infrastructure on agricultural output, especially in the sub-Saharan context. This thesis investigates whether rural road upgrades in northern Mozambique have any short-term effects on agricultural output; specifically, we evaluate the early effects of an ongoing World Bank project. By employing remote sensing and machine learning methods, we identify rural road upgrades that took place between 2018 and 2021. Using a differences-in-differences approach, we find that areas in immediate proximity to roads that received an upgrade did not experience changes in agricultural output, compared to areas that did not receive an upgrade. We restrict the sample and find a significant increase in agricultural output, although not robust. Future research should consider the medium- and long-term impact of rural road upgrades for the complete picture to emerge. While sole dependence on remote sensing data remains a challenge in economics, it is a promising avenue for future research, particularly in contexts where comprehensive survey data is lacking.

   Keywords: remote sensing · rural road improvements · agricultural output · Mozambique · economic development

   Show more
   Paper

   Original (SSE) Mirror (PDF)

   Code

   Main data contribution: a remote-sensing pipeline that detects rural road upgrades from Sentinel-2 imagery, useful for studying upgrade impacts where survey data is sparse. Below, six road segments our pipeline flagged as upgraded between 2020 and 2021.

   Pre · 2020

   Post · 2021

   Segment: 1 2 3 4 5 6