Abstract: We study the impacts of transportation infrastructure improvements on forest loss in Uganda between 2008 and 2021, leveraging a large-scale national road upgrade program. We combine high-resolution panel data on deforestation with changes in travel times to major markets resulting from road improvements. Exploiting quasi-random variation in the timing of project completion to isolate unanticipated market access shocks, we find that improved market access led to increased local forest loss. These effects are not driven by agriculture, mining, or construction. Instead, deforestation is driven by rural charcoal producers who clear trees to supply high-price urban markets. Our results highlight the important role that natural resource access and forest-product price differentials can play in shaping the general equilibrium effects of transportation infrastructure investments. 

Abstract: Sub-Saharan Africa is urbanizing, but a significant portion of the urban population still works in agriculture. We argue that this is because smaller cities are isolated from national markets and trade. We test this claim using individual panel data and Uganda's doubling of paved roads, which improved remote areas' market access. We find that market access causes workers to quit family farms for specialized paid employment. Effects concentrate in peripheral areas, households with comparative advantage in off-farm work, and reflect off-farm opportunities rather than a reduced demand for farm output. We also find that market access leads remote households to simplify farming techniques and scale back farming. Findings are consistent with reliable transport enabling trade with major markets, creating opportunities to specialize according to comparative advantage.

Abstract: Reconstructing the existing housing stock is a critical yet understudied approach to expanding affordable housing. This paper investigates the causal impact of invisible densification - the conversion of living rooms into additional bedrooms - facilitated by the largest PropTech rental platform in Beijing. Exploiting a natural experiment and employing an instrumental variable strategy, we find that a 1% increase in the share of remodeled units leads to a 0.9% decline in bedroom rents, a 0.5% decline in rents for non-remodeled rental units, and a 2.2% reduction in apartment sale prices. To evaluate welfare impacts, we develop a structural model that incorporates tenants' heterogeneous preferences for public amenities and analyzes the distributional effects of remodeling. Results reveal that disadvantaged groups exhibit stronger preferences for remodeled housing, whereas residents in buildings with such units experience negative externalities. Counterfactual analyses quantify the welfare consequences of these externalities and explore how alternative spatial allocations of remodeling could improve affordability.