Publications

We present a model of digital advertising with three key features: (1) advertisers can reach consumers on and off a platform, (2) additional data enhances the value of advertiser–consumer matches, and (3) the allocation of advertisements follows an auction-like mechanism. We contrast data-augmented auctions, which leverage the platform’s data advantage to improve match quality, with managed-campaign mechanisms that automate match formation and price-setting. The platform-optimal mechanism is a managed campaign that conditions the on-platform prices for sponsored products on the off-platform prices set by all advertisers. This mechanism yields the efficient on-platform allocation but inefficiently high off-platform product prices. It attains the vertical integration profit for the platform and the advertisers, and it increases off-platform product prices while decreasing consumer surplus, relative to data-augmented auctions.

The extent to which women participate in the labor market and have access to formal employment differs greatly across Indian states. In this paper we build on the methodology developed by Hsieh, Hurst, Jones, and Klenow (2019) to estimate the productivity consequences of such differences. Using rich microdata on occupational sorting and earnings, our theory allows to separately identify labor demand distortions (e.g., discrimination in hiring for formal jobs) from labor supply distortions (e.g., frictions that discourage women’s labor force participation). We find that both demand distortions and supply distortions are negatively related to state-level economic development. Equalizing distortions across Indian states could raise state-level productivity by up to 15%.

Innovations in big data and algorithms are enabling new approaches to target interventions at scale. We compare the accuracy of three different systems for identifying the poor to receive benefit transfers — proxy means-testing, nominations from community members, and an algorithmic approach using machine learning to predict poverty using mobile phone usage behavior — and study how their cost-effectiveness varies with the scale and scope of the program. We collect mobile phone records from all major telecom operators in Bangladesh and conduct community-based wealth rankings and detailed consumption surveys of 5,000 households, to select the 22,000 poorest households for $300 transfers from 106,000 listed households. While proxy-means testing is most accurate, algorithmic targeting becomes more cost-effective for national-scale programs where large numbers of households have to be screened. We explore the external validity of these insights using survey data and mobile phone records data from Togo, and cross-country information on benefit transfer programs from the World Bank.

We construct a new dataset to show that successful entrants in the consumer food sector build market share by adding new customers. Entrants reach new customers by entering more geographical markets, placing their product in more stores in these markets, and, for a positively selected subset of firms, advertising directly to customers. These activities are costly and are associated with persistent increases in quantities, but there are no differences in markups between new and mature markets. This confirms a central role for marketing and advertising in overcoming demand-side frictions that slow firm growth.

How does emigration of highly educated citizens of low-income countries to high-income countries affect the economies of the origin countries? The direct effect is “brain drain”—a decrease in the country’s human capital stock. However, there may also be indirect “brain gain” effects. This review summarizes evidence that uses causal inference methods to reveal mechanisms that may lead to brain drain, gain, or circulation. Collectively, the weight of the evidence suggests that migration opportunities often increase human capital stock in origin countries and produce downstream beneficial effects through remittances; foreign direct investment and trade linkages; transfers of knowledge, technology and norms; and return migration. We discuss conditions under which benefits from skilled migration may outweigh costs and also describe potential research paths to inform policy.

Heise et al. (2021) develop a model-based empirical measure—sellers per shipment (SPS)—to characterize how firms organize supply chains in response to a quality control problem. High SPS indicates spot-market purchasing with costly inspections, while low SPS suggests long-term relationships where buyers pay an incentive premium to prevent cheating. Here, we document intuitive variation in US importers' SPS across sectors and show that shipping characteristics such as average price, quantity shipped, and shipment frequency are in each sector consistent with the model of sourcing developed in Heise et al. (2021), providing further confidence in the measure.

We evaluate impacts heterogeneity of an Early Childhood Intervention, with respect to the Educational Attainment Polygenic Score (EA4 PGS) constructed from DNA data based on GWAS weights from a European population. We find that the EA4 PGS is predictive of several measures of child development, mother’s IQ and, to some extent, educational attainment. We also show that the impacts of the intervention are significantly greater in children with low PGS, to the point that the intervention eliminates the initial genetic disadvantage. Lastly, we find that children with high PGS attract more parental stimulation; however, the latter increases more strongly in children with low PGS.

Market-based environmental regulations are seldom used in low-income countries, where pollution is highest but state capacity is often low. We collaborated with the Gujarat Pollution Control Board (GPCB) to design and experimentally evaluate the world’s first particulate-matter emissions market, which covered industrial plants in a large Indian city. There are three main findings. First, the market functioned well. Treatment plants, randomly assigned to the emissions market, traded permits to become significant net sellers or buyers. After trading, treatment plants held enough permits to cover their emissions 99% of the time, compared with just 66% compliance with standards under the command-and-control status quo. Second, treatment plants reduced pollution emissions, relative to control plants, by 20%–30%. Third, the market reduced abatement costs by an estimated 11%, holding constant emissions. This cost-savings estimate is based on plant-specific marginal cost curves that we estimate from the universe of bids to buy and sell permits in the market. The combination of pollution reductions and low costs imply that the emissions market has mortality benefits that exceed its costs by at least 25 times.

What are the welfare gains from upgrading electric vehicle infrastructure? This paper develops a model of electric vehicle charging location decisions incorporating the transportation network structure, allocation of travel, and the effect on electric vehicle demand. We estimate the model with rich data on vehicle registrations, road segments, cell phone tracks, and charging locations and characteristics. We examine a counterfactual that adds level 3 (direct current fast) chargers optimally in the Connecticut transportation network. We find that adding chargers yields significant time savings and consumer welfare gains, while electric vehicle market shares are only modestly affected.

This paper examines creation and distribution of surplus from global value chains (GVCs) in low- and middle-income country (LMIC) domestic supply chains. While GVC participation can enhance growth and productivity, low prices paid to small in-put suppliers raise concerns that gains from GVC participation accrue to the large exporters (the buyers). Supply-chain transactions often occur in bargained agreements with non-price terms that increase small supplier surplus, such as quantity stability and other insurance-like terms. Therefore, low input prices reflect both buyers’ share of surplus generated by non-price terms and buyer capture. I enrich a Nash bargaining model to study how both i) value creation through insurance-like agreement terms that mitigate spot market frictions and ii) value capture from buyers threatening to replace external suppliers with in-house production affect prices paid to small, risk-averse suppliers. Using novel transaction data from an Indian garment manufacturer and its nearly 500 fabric suppliers, I estimate a structural model to decompose dis-counts into value creation and capture. Results illustrate that discounts reflect value creation rather than buyer capture; difference-in-differences estimates yield consistent findings. Counterfactual analyses highlight that increasing buyer competition has lim-ited effects on prices paid to small risk-averse suppliers, whereas introducing profit insurance substantially increases prices they receive.

We quantify how pollution affects aggregate productivity and welfare in spatial equilibrium. We show that skilled workers in China emigrate away from polluted cities. These patterns are evident under various empirical specifications, such as when instrumenting for pollution using upwind power plants, or thermal inversions. Pollution changes the spatial distribution of skilled and unskilled workers, and wage returns by location. We quantify the loss in aggregate productivity due to this re-sorting by estimating a spatial equilibrium model. Counterfactual simulations show that reducing pollution increases productivity through spatial re-sorting by approximately as much as the direct health benefits of clean air.

We study carbon offsets sold by firms in China under the Clean Development Mechanism (CDM). We find that offset-selling firms, meant to cut carbon emissions, instead increase them by 49% after starting an offset project. In a model of firm investment decisions and offset review, we estimate that CDM firms increase emissions due to both the selection of higher-growth firms into projects (35 pp) and because offset projects themselves boost firm growth and therefore emissions (14 pp). The CDM reduces global surplus by causing damages from increased emissions four times greater than private gains from trade in the offset market.

We exploit a unique event to study the extent to which popular attitudes towards trade are driven by economic fundamentals. In 2007, Costa Rica put a free trade agreement (FTA) to a national referendum. With a single question on the ballot, 59% of Costa Rican adult citizens cast a vote on whether they wanted an FTA with the U.S. to be ratified or not. We merge disaggregated referendum results, which break new ground on anonymity-compatible voting data, with employer–employee, customs, and firm-to-firm transactions data, and data on household composition and expenditures. We document that a firm’s exposure to the FTA, directly and via input–output linkages, significantly influences the voting behaviour of its employees. This effect dominates that of sector-level exposure and is greater for voters aligned with pro-FTA political candidates. We also show that citizens considered the expected decrease in consumer prices when exercising their vote. Overall, economic factors explain 7% of the variation in voting patterns, which cannot be accounted for by non-economic factors such as political ideology, and played a pivotal role in this vote.

We test whether payments for ecosystem services (PES) can curb the highly polluting practice of crop residue burning in India. Standard PES contracts pay participants after verification that they met a proenvironment condition (clearing fields without burning). We randomize paying a portion of the money up front and unconditionally to address liquidity constraints and farmer distrust, which may undermine the standard contract's effectiveness. Incorporating partial up-front payment into the contract increases compliance by 10 percentage points, which is corroborated by satellite-based burning measurements. The cost per life saved is $3,600–$5,400. The standard PES contract has no effect on burning.

We conduct a systematic re-analysis of intervention-based studies that promote hygienic latrines and evaluate via experimental methods. We impose systematic inclusion criteria to identify such studies and compile their microdata to harmonize outcome measures, covariates, and estimands across studies. We then re-analyze their data to report metrics that are consistently defined and measured across studies. We compare the relative effectiveness of different classes of interventions implemented in overlapping ways across four countries: community-level demand encouragement, sanitation subsidies, product information campaigns, and microcredit to finance product purchases. In the sample of studies meeting our inclusion criteria, interventions that offer financial benefits generally outperform information and education campaigns in increasing adoption of improved sanitation. Contrary to a policy concern about sustainability, financial incentives do not undermine usage of adopted latrines. Effects vary by share of women in the household, in both positive and negative directions, and differ little by poverty status.