Research

ABSTRACT: We study how transportation infrastructure improvements affect forest loss in Uganda, leveraging a large-scale national road upgrade program from 2008 to 2021. We combine high-resolution deforestation panel data with changes in travel times to major markets. Using quasi-random variation in project completion timing to isolate unanticipated market-access shocks, we find that improved market access increased local forest loss. The effects are not driven by agriculture, mining, or construction. Instead, they are consistent with rural charcoal producers clearing trees to supply high-price urban markets, highlighting how forest-product price differentials shape the environmental effects of infrastructure investments.

ABSTRACT: We study the impact of child labor standards in Regional Trade Agreements (RTAs) on a variety of child labor market outcomes. Using harmonized survey microdata from 101 developing countries, we find that RTAs without provisions targeting child labor reduce child employment and increase secondary school enrollment. Child labor provisions in RTAs perversely increase child employment and decrease school enrollment, with especially strong effects for older children. Our findings are consistent with theoretical predictions from a stylized general equilibrium model of child labor in an economy open to international trade.

ABSTRACT: We consider the impact of linking trade liberalization to international environmental agreements. We develop a stylized model of optimal environmental regulation and show that conditioning trade liberalization on environmental treaties can improve compliance. We study this problem in the context of the Montreal Protocol, the watershed environmental agreement regulating ozone depleting substances (ODS). We find that trade agreements with ODS-related commitments reverse the rise in ODS emissions that is typically observed in trade agreements lacking such commitments. Our results highlight an important policy tool in addressing global environmental challenges.

ABSTRACT: Trade liberalization can increase schooling while reducing what children learn. We study this tension in Peru, where large unilateral tariff reductions between 2007 and 2016 generated substantial local variation in import competition. Combining household surveys with administrative records on enrollment and test performance, we examine effects on children's schooling, work, learning, and later outcomes. We document a trade-off between schooling quantity and learning quality: greater exposure increased enrollment and attendance but reduced relative gains in math achievement. This pattern extends from primary into secondary school, where exposed districts saw even larger participation gains, yet the relative math learning gap continued to widen as affected cohorts aged. We trace these effects to countervailing income and substitution channels. Declining protected-sector opportunities lowered the opportunity cost of schooling, while negative income shocks increased children's work-school combination and shifted students from private to public schools. Public schools absorbed this demand without proportional increases in teachers, further limiting learning. In the longer run, affected cohorts completed more secondary school but shifted away from university and toward vocational programs, and by young adulthood were less likely to be employed and earned lower wages. Trade reforms can thus expand school access while weakening skill formation, revealing human-capital costs that enrollment metrics may miss.

ABSTRACT: Since 2000, foreign investors have acquired vast tracts of land in developing countries for agriculture and natural-resource development, part of a global land rush whose cumulative area exceeds the landmass of Myanmar. Whether trade liberalization encourages such acquisitions is theoretically ambiguous. Lower trade costs weaken the case for producing abroad to serve a local market but strengthen the case for allocating production toward locations with a fixed factor advantage. Land, immobile and tied to site-specific agroecological endowments, is the clearest instance of the latter. Combining bilateral data on large-scale land transactions from 1980 to 2018 with the universe of regional trade agreements (RTAs), we find that entry into force of a first RTA raises bilateral land acquisitions to roughly 1.7 times the unconditional sample mean. Effects are concentrated in agriculture and forestry, where they are about six times larger than in mining, and are driven by investors headquartered in upper-middle-income countries. Consistent with a vertical motive, the response is larger for agreements containing investment-protection provisions, for target countries with stronger access to third markets, and for investors based in land-scarce countries. Market access and origin land scarcity contribute in equal measure and with opposite signs, as the vertical motive predicts.

ABSTRACT: Frontier agricultural settlement is a major driver of land-use change in Latin America, but causal evidence on its environmental costs and local development effects remains limited by the absence of spatially explicit settlement data. We study Low German Mennonite colonies, a form of migrant-led agrarian expansion characterized by large contiguous land acquisitions and rapid conversion to commercial agriculture. We combine mapped boundaries and establishment years for 105 colonies founded after 2000 with annual satellite measures of forest loss, biomass carbon, species richness, and local nighttime luminosity. Comparing colony polygons to adjacent untreated buffer landscapes in a staggered difference-in-differences design, we find that colony establishment substantially increases annual forest loss, with associated carbon and biodiversity exposure. In contrast, we find no detectable increase in nighttime luminosity in nearby areas. The results suggest that migrant-led frontier settlement can generate large environmental externalities without measurable local economic spillovers. Counterfactual exercises find that coordinated location choice may dramatically reduce the ecological impact of colony establishments without sacrificing land productivity for agriculture.

ABSTRACT: Roughly two billion people worldwide rely on solid biomass for cooking, with demand for firewood and charcoal a leading driver of forest degradation. We study whether expanding access to liquefied petroleum gas (LPG) as an alternative cooking fuel relieves this pressure, leveraging Ghana's expansion of retail filling stations. We identify local LPG access expansions from sharp changes in household LPG use and stove ownership around station locations and use these access shocks to estimate effects on household fuel use and satellite-derived measures of forest disturbance. Improved LPG access significantly reduces firewood use and the share of charcoal that households produce themselves, with no detectable change in total charcoal use. Local forest disturbance also falls, with the largest reductions in degradation and in the loss of already-degraded forest. Valuing the avoided disturbance using Ghana-specific forest-carbon densities implies substantial avoided emissions, worth several million dollars annually at conservative carbon prices. Energy infrastructure investments can therefore deliver meaningful environmental co-benefits alongside their household-level returns.

ABSTRACT: We develop a model of comparative advantage with quality-differentiated inputs that predicts trade liberalization will reallocate high-quality labor toward agriculture in developing countries, raising productivity and reducing input intensity, while generating the opposite pattern in advanced economies. Leveraging the exogenous timing of entry into force of regional trade agreements (RTAs) in a stacked synthetic triple difference-in-differences design, we estimate causal impacts of trade liberalization on agricultural intensification and agricultural emissions. We find significant reductions in agricultural use of nitrogen, phosphorus, and potash following trade liberalization with corresponding reductions in agricultural emissions of nitrous oxide, methane, and carbon dioxide. Both developing and advanced economies reduce emissions and nutrient use following liberalization, but advanced economies reduce more---consistent with specializing away from agriculture. We find increases in total factor productivity for developing countries and flat-to-negative responses for advanced economies, suggesting that trade liberalization induces asymmetric changes in production technologies across the development spectrum. Using harmonized labor force microdata, we document narrowing sectoral education and wage gaps in developing countries following liberalization, consistent with a comparative-advantage-driven reallocation of higher-quality labor into agriculture that can improve the environmental footprint of agricultural production.

ABSTRACT: This paper develops a spatial general equilibrium model to study how failures in legal contracting institutions shape the distribution of economic activity within a developing economy. Relational contracts endogenously emerge as substitutes for legal enforcement, generating three documented patterns in Sub-Saharan African development: excess urban concentration, entry barriers at the periphery, and weak bilateral trade. Infrastructure investment raises welfare but increases rather than decreases the deadweight loss from contracting frictions; institutional reform reduces this loss directly. The two policy instruments are complements whose effects vary systematically across space.

ABSTRACT: This paper studies environmental regulation in a two-sector endogenous growth economy in which production and innovation compete for a common input. When research draws on the same resources as production, policies that reduce the profitability of production raise the relative return to innovation. As a result, environmental regulation can increase innovation within the very sectors it targets—a regulatory paradox. I show that tighter regulation reallocates labor from production to research within the regulated sector, raising its innovation rate, while general equilibrium price effects may offset this force by shifting activity across sectors. A quantitative calibration to European carbon regulation demonstrates that the elasticity of substitution required for reallocation to overturn the within-sector innovation response lies well above empirically estimated values. As a result, the paradox is likely to arise under most empirically plausible conditions. I then characterize optimal environmental policy and show that when regulation is restricted to output-based taxes, optimal Pigouvian taxes are attenuated because they also reshape innovation incentives. With additional instruments that directly target research, the first best can be restored.

ABSTRACT: Agricultural contracts are important tools in risk management for both producers and contractors.  Critically, contracts can eliminate commodity price uncertainty, with output prices and delivery locations agreed to in advance.  However, contracts also introduce a degree of inflexibility in producers’ ability to respond to changes in input prices.  In particular, by committing to delivery locations and output quantities in advance, producers expose themselves to uncertainty in transportation costs.  We develop a simple model of endogenous contracting where producers face both output price risk and transportation cost risk.  Our model predicts that contracting intensity is a decreasing function of distance-to-market and transportation costs more generally.  These findings have important implications for contracting as a development tool, particularly as it relates to transportation infrastructure in developing countries.

ABSTRACT: Local democratic institutions have important implications for natural resource management, especially when extraction is associated with environmental externalities.  We study the effects of a transition to direct local democratic institutions on natural resource management using a political agency model with electoral competition. Our model of political decentralization suggests that this transition will lead to an increase in local extraction with greater negative externalities. In localities with weak electoral competition, agency costs are higher and the transition leads to resource rent capture by corrupt officials with illegal extraction in protected areas.  In localities with strong competition, increased extraction is through legal channels and is associated with local increases in welfare.  We use Indonesia's transition from locally-appointed leaders to locally-elected leaders as a natural experiment to study forest management. The timing of local elections varied district-to-district both across and within provinces and was determined by the local appointment cycles before the transition. Using high-resolution satellite data on forest cover loss, we apply a difference-in-differences approach and find significant evidence that district-level deforestation increased following direct local elections. In districts with weak electoral competition, this deforestation appears consistent with resource rent capture by corrupt officials driven by illegal logging in protected areas and no commensurate increase in economic development. In contrast, deforestation in districts with high electoral competition appears to be driven by local development.

ABSTRACT: I develop a new class of time series models that explicitly identify horizon-based economic relationships in a unified linear regression framework. Horizon-based effects are identified using an additive decomposition of regressors in the time domain that corresponds to an orthogonal partition in the frequency domain.  This additive decomposition is accomplished with a multiresolution analysis associated with a maximal overlap discrete wavelet transform.  I propose a simple OLS estimator for the class of models that allows for heterogeneity of marginal effects across horizons and establish asymptotic properties of the estimator.  The new framework is robust to horizon-based misspecification—meaning that when the model is inappropriately applied to time series regressions, valid inference is still possible.  Inference is also largely robust to the stationarity (or more generally, persistence) of decomposed regressor variables.  Furthermore, in contrast to existing methods like long-horizon regressions, time horizons are disentangled from predictive horizons in this new framework.  The new class of model also offers improved inference on seasonal relationships between regressand and regressors.