Deivy Houeix Profile Picture

My name is pronounced: Day-vee Wex

Welcome! I received my Ph.D. in Economics from MIT in May 2025. I am currently a Prize Fellow at the Center for History and Economics at Harvard University. In 2027, I will join Columbia Business School as an Assistant Professor of Economics.

My primary field is Development Economics, with secondary interests in Organizational Economics.

My research focuses on technology and firms in lower-income countries, particularly in West Africa. I explore how digital technologies reshape economic relationships and contract structures within and between firms, uncovering some key drivers and barriers to their adoption.

Over the past eight years, I have conducted research projects in Côte d'Ivoire, Ethiopia, Senegal, and Togo.

[Access my CV here] – deivyhoueix@fas.harvard.edu

Working Papers

  • Asymmetric Information and Digital Technology Adoption: Evidence from Senegal, [New version!]
    [2025 Daniel Cohen Award] [Ronald H. Coase Best Dissertation Award Honorable Mention, SIOE], Media Coverage: RFI, World Bank Blog, Jeune Afrique
    Abstract

    Should workers control who can access the data digital tools generate? Across two field experiments in Senegal's taxi industry, I randomize the information architecture of a digital payment technology that roughly halves drivers' cash-handling costs-varying whether taxi owners observe the transaction trail. Visibility raises driver effort and retention, but deters adoption by threatening informational rents: ensuring transaction privacy nearly doubles take-up, especially among the poorest, lowest-performing drivers. In an estimated relational-contracting model, the level and distribution of the gains from digitalization hinge on the architecture. Driver-controlled disclosure Pareto-dominates the alternatives and is what the partner company implemented at scale.

  • Relational Frictions Along the Supply Chain: Evidence from Senegalese Traders (with Edward Wiles)
    Media Coverage: World Bank Blog
    Abstract

    Search and trust frictions have historically made it hard for small firms in lower-income countries to buy inputs from foreign markets. The growth in smartphone ownership and social media usage has the potential to alleviate these barriers. Informed by a dynamic model of relational contracting, we run a field experiment leveraging these technological tools to provide exogenous variation in (1) search frictions and (2) trust frictions (adverse selection and moral hazard) in a large international import market. In the search treatment, we connect a randomly selected 80\% of 1,862 small garment firms in Senegal to new suppliers in Türkiye. We then cross-randomize two trust treatments that provide additional information about the types (adverse selection) and incentives (moral hazard) of these new suppliers. Reducing search frictions is sufficient to increase access to foreign markets: in all treated groups, firms are 26% more likely to have the varieties a mystery shopper requests, and the goods sold are 30% more likely to be high quality. However, the trust treatments are necessary for longer-term effects: using both transaction-level mobile payments data and a follow-up survey, we show that these groups are significantly more likely to develop the connections into relationships that persist beyond the endline survey. These new relationships lead to increases in medium-run profit and sales. Finally, we use the treatment effects to estimate the model and counterfactually lower the trust frictions among the whole supplier pool for a given firm, finding that the largest gains come from alleviating adverse selection.

  • Payment Infrastructure and Policy Effectiveness: Evidence from Public Programs in Togo (with Paul Brimble, Axel Eizmendi Larrinaga, and Toni Oki) [New version!]
    Abstract

    Public programs in lower-income countries are increasingly delivered through digital payments, but rural economic life still runs on cash. Households must convert cash into digital money to pay, or a digital transfer into cash to spend it, and either conversion requires a trip to an agent. We show that this last-mile problem is costly enough to weaken policy effectiveness. In rural Togo, the transport cost alone of reaching an agent amounts to 19% of what the median customer pays each month for pay-as-you-go solar electricity. Exploiting the staggered rollout of a nationwide solar subsidy, we find its effect on adoption is nearly three times larger where an agent is within walking distance. A subsequent government-led agent expansion initiative raises adoption by as much as this differential, consistent with complementarity and isolating the role of payment access from potential confounders. The same friction, now on the cash-out side, attenuates the effects of Togo's emergency cash transfers, delivered digitally through the same agents. Since operators place agents by transaction volume and public programs often target underserved, low-volume areas, this wedge between private and social returns suggests a role for public investment in payment infrastructure.

  • Nationwide Diffusion of Technology: Experimental Evidence from Multiple Networks [Coming soon!]
    Media Coverage: Liberation

  • Aggregating Partial Rankings from Neighbors: Methodology and Empirical Evidence (with Pascaline Dupas and Marcel Fafchamps), R&R Quantitative Economics
    Abstract

    Many decisions require ordering alternatives: for example, the selection of top candidates for a competitive academic program or the selection of the poorest individuals for a cash transfer program. One common approach consists in aggregating orderings reported by different observers (e.g., committee or community members), but those orderings are typically partial: not all observers rank all applicants. We introduce a novel type of approach, based on pairwise rankings, to (i) aggregate partial orderings reported by multiple observers and (ii) construct confidence intervals for the resulting aggregate ordering. We identify, both theoretically and using simulations, the conditions under which a pairwise approach dominates rank averaging: when reporting error is low, reported orderings are partial, and observers rank alternatives that are close to each other in their true latent ordering. We introduce improvements to rank averaging and pairwise methods and illustrate them using several datasets. We find that, with partial reported orderings, Borda counts (i.e., simple rank averages) are dominated by the averaging of normalized ranks and should never be used in practice.

Non-Refereed Publications