An increasing number of decisions are guided by machine learning algorithms. But when consequential decisions are encoded in algorithms, individuals may strategically alter their behavior to achieve desired outcomes. This paper develops an empirical approach that adjusts decision algorithms to anticipate manipulation. By explicitly modeling incentives to manipulate, our approach produces decision rules that are stable under manipulation, even when the rules are fully transparent. We stress-test this approach through a large field experiment in Kenya. When implemented, linear strategy-robust decision rules outperform standard linear models such as LASSO. (JEL C45, C93, D12, D91, G51, O12, O16)
Efficient risk sharing implies a simple factor structure for marginal utilities of expenditure (MUEs): Pareto weights divided by a common price. The standard approach infers MUEs from total expenditures, implicitly assuming homothetic preferences, unitary income elasticities, and identical price elasticities. Risk-sharing tests using total expenditures work for idiosyncratic shocks (budgets change, but not prices), but not “covariate” shocks (prices change). I describe all preferences that permit one to infer MUEs from expenditures and estimate nonhomothetic MUEs to test whether covariate shocks are shared efficiently in Uganda. This delivers sensible results; the standard approach suggests that droughts, floods, and pests are beneficial. (JEL D12, D81, O12, O13, Q12)
We study the relationship between corporate income taxation and carbon dioxide (CO 2 ) emissions in the United States. We show CO 2 -intensive firms benefit more from the tax advantage of debt and pay lower income taxes on their capital income. Building on these new facts, we provide evidence that a cut in the corporate income tax rate leads to a larger expansion of clean firms. We develop a multisector general equilibrium model that accounts for our evidence and quantify the impact of corporate tax reforms on aggregate emissions. A policy that eliminates the tax advantage of debt could reduce aggregate emissions without affecting GDP. (JEL D22, G32, H25, H32, Q52, Q54, Q58)
Experimental and quasi-experimental studies show that childhood neighborhoods have substantial causal impacts on children's adult earnings and other long-term socioeconomic outcomes. Several market failures are likely to lead to excessive residential segregation by parental income and to an undersupply of mixed-income neighborhoods. Missing markets for opportunity originate in capital market imperfections for borrowing on children's future earnings, neighbor spillovers, behavioral biases, and housing search frictions. Customized housing mobility services combined with subsidized housing vouchers can help low-income families move to higher-opportunity areas. Place-based policies that create mixed-income neighborhoods appear to increase long-term outcomes for less-advantaged children. (JEL I12, I32, I38, J13, J15, R23, R38)
Digital platforms reduce geographic frictions, enabling better matching between service providers and users. I quantify reallocation gains in Swedish online health care, using nationwide time-conditional random assignment between patients and physicians. Matching high-risk patients with doctors effective at reducing Emergency Room visits lowers such visits by 4.4 percent (SE 1.3); reallocations also reduce Counter-Guideline antibiotics by 3.1 percent (1.4). I find limited trade-offs in matching; horizontal differentiation among doctors and varied patient needs allow improvement in multiple outcomes simultaneously. Efficiency-enhancing reallocations also affect equity. The findings highlight the potential for care reorganization aligning provider heterogeneity with patient needs when geographic constraints are lifted. (JEL C78, H51, I11, I13, I14, I18)
