Andrea Neyra-Nazarrett

Research

I work on labor, organizational, and personnel economics. I study how wages, managers, and firm practices shape worker effort and productivity, using large unstructured data from firms.

Job Market Paper

Working Papers

In my research I study what makes firms productive.

  • paper first page

    Training Within Firms

    with Brayan Diaz, Julian Ramirez, Raffaella Sadun, and Jorge Tamayo  ·  NBER Working Paper No. 33670 · HBS Working Paper 25-045

    Training investments are essential for improving worker and firm productivity, yet their implementation is often hindered by low participation rates and insufficient worker engagement. This study uses data from three firms (a car manufacturer, a quick-service restaurant chain, and a retail company) to show that variation in training participation among employees is closely tied to differences in middle managers’ behavior and practices. Middle managers who actively engage with their employees and emphasize their well-being and development are associated with significantly higher participation in training programs. These managerial differences significantly influence employee performance and absenteeism, especially during periods of organizational change. Together, these findings underscore the importance of middle managers in bridging the gap between centrally designed HR policies and their effective on-the-ground execution.

    Presented at the NBER Summer Institute (SI 2024, Personnel Economics).

  • paper first page

    What Do Good Managers Do and What Do They Leave Behind?

    with Achyuta Adhvaryu, Parker Howell, Anant Nyshadham, and Jorge Tamayo

    Presented at the NBER Summer Institute (SI 2025, Personnel Economics).

    Using administrative data from a large Colombian retailer and exploiting a manager rotation policy to estimate manager value-added, we study what good managers do and whether their impacts persist after they leave. Good managers increase sales by 12%, reducing stockouts while making inventory leaner, generating larger revenue gains from price changes without increasing the frequency of repricing, and shifting personnel attention toward back-office operations. When they depart, the staffing and inventory improvements persist, but pricing-related gains erode. These patterns reveal a fundamental distinction: some managerial contributions are embedded in organizational structure, while others depend on the manager’s ongoing, context-specific judgment.

  • paper first page

    VAT Holidays as Short-Term Economic Stimuli: Evidence from Colombia’s “Días sin IVA” during COVID-19

    with José Martinez-Carrasco, David Muñoz-Sanchez, Alejandro Rasteletti, and Diego Zamora  ·  IDB Working Paper

    Sales tax holidays (STH) are commonly used to spur short-term consumption, yet the evidence on their effectiveness in developing countries remains limited. This paper analyzes Colombia’s Días sin IVA (DSI), a nationwide VAT holiday held eight times between 2020 and 2022. The policy temporarily cut the 19% VAT to 0% on selected goods, covering about 40% of formal household consumption. Using high-frequency transaction data and a repeated regression discontinuity design, we find that DSI increased average sales in formal businesses by 15%, after accounting for intertemporal substitution. Unlike previous studies, we show evidence of substantial spillover effects across sectors. While businesses selling goods subject to the VAT reduction experienced a 17% average increase in sales, those selling goods with non-reduced VAT increased their sales by 14%, with widespread heterogeneous effects across diverse product categories.

  • Work in Progress

  • Trade Barriers Without Tariffs: Border Tightening and Pass-Through in Retail Markets

    Solo-authored  ·  Work in progress

    Do delays at the border spill over into domestic retail markets? When U.S. authorities tighten truck inspections at crossings such as Laredo and El Paso, trucks wait in line for a day or more before they can cross. Many Mexican suppliers sell both to U.S. buyers and to supermarkets in Mexico, so the delay reaches Mexican shelves. I study what happens on those shelves using scanner data from a large supermarket chain in Mexico: prices, quantities, stockouts, and purchase orders. Purchase orders are linked to supplier IDs, which lets me identify which suppliers export to the United States and which border crossings they use. An event-study design compares products from exporting suppliers with similar products in the same category from suppliers that do not export. Within exporters, it compares suppliers that use the affected crossings with those that use other crossings. The outcomes are temporary unavailability, delivery timing, prices, and product quality, measured by the share of perishables that arrive unsellable. The design separates a short-run effect, in which goods that cannot cross are sent to local shelves instead, from the longer-run strain that repeated delays put on supply.

  • AI Tools for Store Managers: Training, Adaptation, and Store Performance

    Field experiment with a large retailer in Mexico  ·  In preparation

    How do store managers put AI to work, and what changes in their stores when they do? In a field experiment with a large retailer in Mexico, I train store managers to use AI tools in two ways. In one arm, managers receive general training in using AI, with no instruction on how to apply it to specific tasks, and I observe which tasks they adapt it to on their own. In the other, managers are trained on specific tasks that can be done with AI, and I measure the impact on store performance. Comparing the two arms shows whether the gains from AI in the store come from the tool itself or from telling managers what to do with it.

  • Local Security Shocks and the Organization of Work Inside the Firm

    Solo-authored  ·  Work in progress

    Mexico's security crisis is usually studied through its aggregate consequences: homicides, displacement, investment, and employment measured by municipality or state. Yet violence is an event measured in hours. An armed confrontation occupies a street for an afternoon, and the people who work on that street decide that evening whether to return. I saw what that looks like from inside. When conflict between criminal factions broke open in one Mexican state in 2024, I was in the field with a large retailer, and the morning after the violence started the women and the youngest workers did not come in; the store ran on men who lived close enough to walk. A security shock does not simply subtract labor: it selects it, and the selection reorganizes the firm. I link the biometric time clocks, personnel files, transfers, promotions, and separations of the retailer's workers to armed confrontations and homicides in public space, geolocated to the place and day of occurrence and matched to stores by distance and date. Identification comes from the timing and location of events the store does not generate: an event at night falls on the closing shift and not on the morning one, a within-store, within-day comparison that demand cannot produce, and the firm's own hours and security decisions are observed, so its reaction is a mediator I can trace rather than a confounder. I measure absence and lateness, who is absent, how shifts are recomposed, who is moved and who is promoted, and how much of the store's deterioration in separations and shrinkage is compositional: violence damaging the store mainly by changing who is in it.

  • Electoral Business Cycles and Everyday Compliance

    Solo-authored  ·  Work in progress

    Inventory shrinkage as a measure of everyday compliance, comparing stores in municipalities where the 2019 mayoral race was decided narrowly against stores where it was not.

  • In the press