Noor Sethi

Selected Work

"Hit or Miss: Targeting the Poor (Better)"

Developing country governments use proxy means testing (PMT) to target cash transfers when income is unobserved, traditionally relying on ordinary least squares regression to predict household welfare from observable characteristics. This paper evaluates whether machine learning methods can improve targeting accuracy and social welfare relative to conventional approaches. Using household survey data from India, I compare six prediction methods: OLS, elastic net, k-nearest neighbors, decision trees, random forests, and gradient boosting. Machine learning methods reduce targeting errors by 2-3 percentage points and increase social welfare by 4-5 percent relative to OLS. k-Nearest neighbors achieves the best performance, followed by ensemble methods. Critically, I demonstrate that error minimization and welfare maximization require different targeting thresholds. Error-minimizing thresholds approach universal coverage, while welfare-maximizing thresholds target approximately half the population (48th-55th percentile). This divergence reflects the asymmetric welfare costs of targeting errors: exclusion errors impose larger welfare losses than inclusion errors due to diminishing marginal utility. These findings suggest governments can substantially improve both accuracy and welfare outcomes by adopting machine learning for poverty targeting, though optimal threshold selection depends on whether policymakers prioritize minimizing fiscal costs or maximizing social welfare.

"Hit or Miss(ing) Assets: Optimal Targeting Thresholds Under Strategic Behavior"

Developing country governments use proxy means testing (PMT) to identify households eligible for cash transfers, collecting asset data to predict income. However, PMT creates incentives for households to hide assets to qualify for transfers. This paper develops a theoretical framework to determine optimal targeting thresholds when households behave strategically. Using household data from India, I model the government as choosing a targeting threshold that maximizes social welfare under a fixed budget constraint, where eligibility is determined by predicted rather than actual income. I extend this to account for strategic behavior, where households compare the expected benefit of hiding assets against the cost of concealment. Simulations reveal that the optimal threshold when income is unobserved (Rs. 13,200) is higher than when perfectly observed (Rs. 12,000), corresponding to 65% versus 30% of the population receiving transfers. Higher thresholds reduce misreporting incentives through three mechanisms: smaller per-capita transfers, lower marginal utility for wealthier households, and harder-to-hide assets among the wealthy. These findings inform ongoing policy debates about transfer program scope and suggest governments can improve targeting accuracy by strategically choosing thresholds that account for household behavioral responses.

"Progress or Backsliding? Changes in the Gender Wage Gap for Business Professionals" (with Ann Harrison and Laura Kray)

In the United States, much of the gap in earnings between men and women is due to the persistent gap for high wage earners. This paper explores changes in the gender wage gap for MBAs graduating from a large public university over 30 years. We document large gender wage gaps on average, which grow in the course of men's and women's careers. Comparing graduates at identical career stages across time periods to address composition concerns, we show that the raw gender wage gap has shrunk by 33 to 50 percent over the last two decades. Additionally, the temporal pattern of the gap has fundamentally shifted: while gaps only emerged over time in earlier decades, significant gaps now emerge immediately. Convergence in labor supply factors, particularly hours worked, explains much of the narrowing gap, alongside shifts in industry composition. However, unexplained wage gaps persist for recent graduates from the very start of their careers, suggesting different underlying mechanisms across cohorts. These findings highlight both progress in gender wage equity among business professionals and concerning patterns that emerge earlier in careers than in previous decades.

"Statistics for Critical Thinkers" (with Gautam Sethi)

Co-authoring innovative statistics textbook that communicates complex analytical methods through step-by-step explanations and real-world policy applications. Uniquely designed to improve statistical literacy by relating foundational concepts like t-tests and ANOVA to demonstrate their unified logic.

Field Experience

In other words:  the people, places, and chance adventures that bend and sway our minds, shape what we study and how we write about it, but rarely make it past the acknowledgements page.

• Busia, Kenya | Innovations for Poverty Action

• Dharamshala, India | Chinmaya Organization for Rural Development

• Córdoba, Argentina | Ciudadanos 365

Research Assistantships

• Ann Harrison & Laura Kray | Haas School of Business

• Rebecca Goldstein | School of Law at UC Berkeley

• Edward Miguel | Dept. of Economics at UC Berkeley

• Dean Karlan | Innovations for Poverty Action

Public Health Capstone
Graduate Student Instructor, Spring 2026 | UC Berkeley

Environmental Economics
Graduate Student Instructor, Fall 2020, Spring 2021 & Fall 2022, Spring 2023 | UC Berkeley

Statistics and Econometrics
Graduate Student Instructor, Spring 2025 | UC Berkeley

Teaching Assistant, Fall 2017 & Fall 2019 | Bard College

Microeconomics
Lecturer, Spring 2023 | Parami University

Teaching Assistant, Fall 2012, Fall 2013, & Fall 2014 | Smith College