I am an economist studying how contracts and firm-to-firm relationships shape trade and development. My work uses a combination of quantitative models, administrative data, and field experiments.
We study search and trust frictions in international sourcing, and whether the growth of "social commerce" in lower-income countries can alleviate them. We document that firms use social media as sourcing infrastructure in ways related to these frictions. Guided by a relational-contracting model, we run a field experiment with 1,862 Senegalese garment retailers, matching them to suppliers in Türkiye (search) and cross-randomizing information about types (adverse selection) and incentives (moral hazard). New matches expand access to varieties and quality, but only develop into relationships when trust frictions are alleviated. Structural estimates imply trust frictions are complements and substantially limit trade.
Adverse selection and moral hazard in trade are severe and reinforce each other, but a large minority of firms in LMICs use "social commerce" as a decentralized, network-based way to mitigate them.
Modern quantitative trade models emphasize cross-country heterogeneity in productivity across goods as the source of trade and its benefits. In workhorse versions, this heterogeneity is assumed to follow specific functional forms, such as i.i.d. Frechet or Pareto. We provide a systematic exploration of how sensitive the gains from trade estimated from these models are to relaxing this assumption---first, with a new formula for sharp bounds on gains from trade across all possible distributions, and then with a numerical characterization for the case where distributions must be close to the workhorse form. Importantly, for each candidate distribution we ensure that its implied trade model is consistent with standard estimation methods such as the ability to match all bilateral trade data points and an exogeneity assumption. Our estimated bounds are wide, even for small departures from the canonical model: the maximal gain is about four times larger and the minimal gain is zero. Imposing high-level restrictions on the distributions considered---for example, maintaining Frechet or Pareto marginals while relaxing independence, or vice versa---does little to shrink these bounds.
The estimates of gains from trade produced by workhorse quantitative trade models are highly sensitive to deviations from the typical maintained assumptions, such as Frechet or Pareto productivity.
We embed a field experiment in a nationwide recruitment drive for a new health care position in Zambia to test whether career benefits attract talent at the expense of prosocial motivation. In line with common wisdom, offering career opportunities attracts less prosocial applicants. However, the trade-off exists only at low levels of talent; the marginal applicants in treatment are more talented and equally prosocial. These are hired, and perform better at every step of the causal chain: they provide more inputs, increase facility utilization, and improve health outcomes including a 25 percent decrease in child malnutrition.
Contract design does more than screen and incentivize a given pool of agents, as it also affects who selects into that pool in the first place: emphasizing future benefits for a healthcare job in Zambia attracted more talented applicants that substantially improved health outcomes.
Awarded the Arrow Award for the best paper published in health economics in English in 2020 by the International Health Economics Association.
Selected Work in Progress
Aggregate Consequences of Relationship Dynamics in Tradewith Pulak Ghosh
Most trade takes place in buyer–seller relationships that both expand gradually over time and use future surplus to sustain trade when formal enforcement is limited. We study how these relationship dynamics affect aggregate adjustment to trade shocks. Using rich firm-to-firm transactions data from India, we document that: (i) trade within a typical relationship doubles in value over its first two years, (ii) this gradual expansion is stable across the firm size, age, and product experience distributions, and (iii) relationships start meaningfully closer to maturity in states where contract enforcement is stronger. To study the aggregate consequences of these micro-level features, we develop a dynamic multi-state general-equilibrium model that embeds optimal relational contracts within a tractable Ricardian framework. Following a trade shock, adjustment is gradual because firms must build new supplier relationships, while changes in future surplus endogenously reshape the contracts supporting existing and new relationships. We estimate the model using the firm-to-firm data and use it to study the dynamics of the transition following a reduction in interstate trade costs calibrated to a major 2017 Indian economic integration reform. The short-run change in trade is only around half of the long-run change, despite the liberalization-induced loss of future surplus constraining existing own-state relationships.
Trade Credit and Digital Records: Evidence from Cote d’Ivoire with Deivy Houeix
Trade credit—suppliers allowing buyers to purchase inventory now and pay later—is central to firm-to-firm commerce but depends on trust. In settings with limited contract enforcement, suppliers often ration credit to unfamiliar buyers, limiting firm growth and creating barriers to entry. We study whether firms can use increasingly available mobile payments records to signal creditworthiness to potential suppliers or lenders---an open finance model---and how this affects relationship development and firm- and market-level outcomes. We collaborate with a large mobile money provider in Cote d'Ivoire to build and evaluate a tool for firms to provided verified disclosures, as well as a fintech specializing in trade credit to study how specialized intermediaries use such records. We run an RCT with 2,000 retailers in the packaged drinks sector, where we randomize whether firms can access the new disclosure tool and whether we connect them with the specialized trade credit lender. Pilot results show that the ability to do verified disclosures increases supplier willingness to extend credit by 20 percentage points. The full RCT is active, with outcomes of interest including credit access, supplier relationships, product varieties, prices, stockouts, and profits and sales.
