Research | Mitali Roy Mathur

Social Protection

Government Scheme Overload: Take-Up and Targeting in a Multi-Scheme Context

Mitali Roy Mathur

Presented at: Advances in Field Experiments (2026), GPRL-IPA Research Gathering (2026), CEGA Staff Meeting (2025)

Funding: CEGA, J-Pal Social Protection Initiative, UC Davis Economics Department ($44,575)

Research question: How does the quantity and quality of government welfare schemes offered to individuals affect take-up and welfare?

Methods: Randomized Controlled Trial in India

Keywords: Social Protection Policies, Take-Up, Targeting, Choice Overload

Abstract Press

Developing countries are often faced with a dilemma: governments offer a variety of social protection policies to people in need, but the rates at which individuals take-up these programs are low. Canonical models typically analyze take-up of \textitsingle programs in isolation, concluding that incomplete take-up is either a result of efficient targeting or a failure of information, hassle, or stigma. Building on this literature, I study multi-scheme contexts: settings in which individuals are eligible for multiple programs and may come to be aware of multiple programs at once. My paper studies how individuals choose how many and which programs to apply to when exposed to a basket of schemes. I first define a model of take-up across a set of schemes an individual is eligible for. I introduce two frictions: awareness and attention. Then, I implement a randomized controlled trial that experimentally varies the quantity and quality of information about benefits shared to informal workers who are eligible for multiple programs. I vary quantity by varying whether or not citizens learn about one scheme at a time with opt-in continuation versus all eligible schemes at once. I vary quality through sharing information about random or recommended schemes. My research emphasizes the importance of studying social protection schemes more comprehensively and highlights that one major bottleneck to take-up is not simply a lack of awareness, but a complexity in navigating the quantity of potential schemes.

Migration

Ahead with H-4 EAD: The Effects of Granting Work Authorization to High-Skilled Female Immigrants

Mitali Roy Mathur

Presented at: Economics of Migration Junior Workshop (2026), Junior Economics of Migration Zoom Seminar (2026)

Research question: What are the labor market, household, and selection effects of offering work authorization to the spouses of H-1B workers?

Methods: Natural Language Processing, K-Means Clustering, Difference-in-Differences

Keywords: High-skilled immigration, Joint migration, H-1B visa, H-4 visa

Abstract

Countries often host immigrants under a variety of visa categories. Immigration policy strictly regulates the quantity, eligibility requirements, and rights of immigrants in each visa group. Given norms around family unification, many immigrant visa categories include visas for dependents. Policymakers and researchers typically focus on studying the effects of immigration driven by lead migrants. Com- paratively less attention has focused on the dependents of these migrants. Understanding how policy changes impact immigrants on dependent visas plays an important role in studying the joint decisions migrants make. In this paper, I evaluate the effects of a 2015 change in immigration policy in the U.S. that allowed H-4 visa holders (spouses of high-skilled H-1B holders) access to employment authorization (EAD). I use natural language processing (NLP) and unsupervised machine learning (ML) to identify likely H-1B and H-4 visa holders in the American Community Survey. After validating my classification approach through a series of checks, I use a difference-in-differences strategy to isolate the policy’s labor market and family effects. I find that the work authorization restriction created a latent labor supply of high-skilled female immigrants that were ready to work, but were legally barred from doing so. One year after the policy removed these restrictions, the employment rate of H-4 visa holders increased by 42.2 percentage points relative to similar naturalized citizens, effectively closing the employment gap between citizens and H-4 visa holders. I also find that the H-4 EAD policy increased the fertility rate of H-4 visa holders by 6.3 percentage points relative to similar non-citizens. My results are driven by incumbent H-4 visa holders who were already in the U.S. prior to the policy change, indicating that the mechanism was not selection. Additional descriptive evidence suggests that the H-4 EAD policy did not change the composition of H-1B or H-4 workers. Overall, my research uses the example of the change in policy impacting H-4 visa holders to show how a simple machine-learning-based approach to visa classification can be used to analyze the effects of specific immigration policies. I find that small changes to the legal rights of dependents have short-term effects in realigning labor supply and family formation towards equilibrium, without altering the composition of new immigrants.
Immigration and Macroeconomic Outcomes in OECD Countries

Gaetano Basso, Mitali Roy Mathur, and Giovanni Peri

Research question: What drives immigration across countries, and what are the macroeconomic effects of immigration in OECD countries?

Methods: Local projections, Shift-share IV

Keywords: OECD Economies, Immigration, Population Change, Immigration Policies, Productivity, Investment, Employment-Population Ratio

Abstract Paper

OECD countries, and particularly Europe’s leading economies, have experienced a significant decline in native population growth alongside large increases in net immigration. We compile a new and comprehensive dataset of net immigration rates from all origins to OECD countries and document the characteristics of these flows, establishing that most of the net immigration increase was from non-OECD countries and was significantly high-skilled, over the period 1990–2024. We show that the substantial cross-country heterogeneity in immigration dynamics is difficult to explain by push factors or network effects, while immigration policy – as captured by summary indices – account for only a small share of the observed variation. Unexpected immigration shocks and surges were relevant phenomena of these years. We then examine the relationship between immigration and growth in population and GDP per capita, focusing on labor productivity, capital investment, and total factor productivity (TFP), using local projections and several sources of identifying variation. We find that immigration from non-OECD countries has been a significant driver of GDP per worker growth, primarily through higher investment. High-skilled immigration, in particular, is associated with stronger human capital accumulation, faster TFP growth, and greater capital deepening. Positive immigration shocks and surges and more open immigration policies, are likewise associated with improved economic performance. By contrast, native population growth has either no effect or a weakly negative effect on GDP per capita and productivity growth. The estimated coefficients on immigration, while positive, are often imprecisely estimated. Still, their magnitude and direction are important and consistent with a large and growing literature documenting the positive effects of immigration – especially high-skilled immigration – on productivity, entrepreneurship, and economic growth.
Landscaping internal migrant vulnerabilities and labor department schemes

Mitali Roy Mathur

Research question: Which government schemes are most important to internal migrants in India?

Methods: Qualitative interviews

Keywords: Internal Migration, Social Protection Programs, Vulnerability

Abstract

I led a research project in collaboration with the research team at Jan Sahas’s Migrants Resilience Collaborative. In this project, I built a database of different labor department welfare schemes in India. Next, I constructed a unique vulnerability index using a weighting technique to rank schemes based on vulnerabilities migrants experience tied to their demographic profiles. Finally, I designed an application in which users can enter different demographic profiles of migrants and see a ranking of types of schemes that may be most important to that migrant persona. This research contributes to the literature on optimal program design with respect to welfare targeting and lays the foundation for future research into interventions that can increase internal migrants’ take-up of specific schemes.

Gig Work

Uncertainty and Frictions in Gig Work: Evidence via Conditional Income Guarantees

Mitali Roy Mathur, Aleksandr Michuda, and Shotaro Nakamura

Funding: IDinsight Digital Economy Research Impact Initiative ($200,000)

Research question: How do conditional guarantees in the gig economy affect labor supply, earnings, and welfare?

Methods: BDM auction, MTE, field experiment

Keywords: Willingness to pay, contract design, gig work

Abstract

Gig economy workers face a fundamental trade-off between the flexibility of piece-rate payment and the uncertainty of earnings, yet causal evidence on how workers value this trade-off remains scarce. We propose a field experiment with ride-hailing drivers in which each driver states their willingness-to-pay (WTP) for four fixed guarantee contracts varying in trip requirements, scheduling flexibility, and guarantee personalization. A random price P ∗ is drawn from a continuous distribution; drivers whose WTP meets or exceeds P ∗ are enrolled at that price, generating a continuously distributed instrument that identifies marginal treatment effects (MTEs) for the full sample. Post-enrollment, a Karlan-Zinman surprise randomization assigns enrolled drivers to improved or unchanged contract terms, providing clean identification of moral hazard separate from selection. The design enables us to: (i) measure money-metric valuations of labor supply costs, scheduling flexibility, and guarantee personalization through within-driver bid differences; (ii) separate adverse selection from moral hazard using WTP variation among enrolled drivers and the post-enrollment surprise randomization; (iii) estimate a full MTE curve over the entire population of drivers; and (iv) conduct structural welfare analysis including optimal contract design, mandatory versus voluntary enrollment comparisons, and personalized guarantee simulations
Understanding the challenges of female surveyors

Mitali Roy Mathur, and Lipika Biswas

Research question: What are the unique challenges that female surveyors face?

Methods: Descriptive survey

Keywords: Female surveyors, fieldwork, gig work

Abstract Read More

I co-led an internal project with a field manager at IDinsight to understand the challenges that female surveyors face. We were interested in the gender dynamics of our field teams in order to shed light on ways in which IDinsight can improve its processes and to understand the gender dynamics of a unique form of gig work. Our learnings highlighted challenges surveyors experience while menstruating, with gender-based bias, and with safety.

Gender

The Diversity of Conference Presenters in Virtual and In-Person Conferences (2017–2023)

Jenna Stearns, Keer Yang, and Mitali Roy Mathur

Funding: Academic Female Finance Committee ($19,750)

Research question: How did the transition from in-person to virtual conferences impact women’s participation?

Methods: Web-scraping, Difference=in-differences

Keywords: Female labor force participation, Academic conferences

Abstract Data

We developed a publicly available dataset to document the gender and institutional affiiation of presenters, discussants, and session chairs at four major finance conferences - the AFA, SFS Cavalcade, WFA, and FMA. The data span in-person, virtual, and hybrid formats, enabling research on whether virtual conferences affect gender representation and access to prestigious speaking roles in academic finance
Evaluating the implementation of a women’s empowerment program

IDinsight

Research question: Did IWWAGE’s women’s empowerment program achieve its intended outcomes?

Methods: Descriptive survey, process evaluation

Keywords: Process evaluation, women’s empowerment

Abstract

I led the first of two rounds of a process evaluation to understand the implementation of the Strengthening Women’s Institutions for Agency and Empowerment program run by the Institute for What Works to Advance Gender Equality (IWWAGE). The program is implemented by different partners in four states and focuses on establishing gender resource centers and introducing training programs for women in self help groups. The first round entailed detailing out a theory of change to understand the program and collecting data via a phone survey on key process indicators. The results from the second round highlight the opportunities for IWWAGE to improve its program implementation while scaling.

Data Quality & Methods

Understanding best practices to improve response rates in phone surveys in developing contexts

Abhilash Biswas, Karan Nagpal, and Mitali Roy Mathur

Research question: How can we improve response rates in phone surveys in developing contexts?

Methods: Phone survey

Keywords: Phone Surveys, Response rates, Data quality

Abstract Read More Paper

During the COVID-19 pandemic, in-person primary data collection was put on hold. As part of the Data Delta innovations team at IDinsight, we began working on building systems and running small experiments to understand how best to implement high-quality phone surveys. Our research culminated in many findings on key processes for executing phone surveys: callback protocols, best times of day to call and phone survey lengths, rapport-building, and reaching women.
Improving data quality and data use for India’s Aspirational Districts Program

IDinsight Data Delta Team

Research question: How can we improve the quality of data reported by India’s Aspirational Districts?

Methods: Household survey

Keywords: Primary data collection, Household surveys

Abstract Read More

As a member of the Data Delta team at IDinsight, I worked on two different phases of a project with Niti Aayog as part of their Aspirational Districts Program (ADP). In the first phase, I worked on managing data quality for a primary survey of 27,000 households across 27 of the poorest districts in India. In the second phase, I worked on a team to understand the pipeline in which administrative data is collected in each district to identify potential areas of improvement in sampling, training, and data collection processes and implement automated data quality checks.

Behavioral Nudges

Evaluating the role of behavioral messages at improving maternal health outcomes

IDinsight

Research question: How can behavioral messages improve maternal health outcomes in South Africa?

Methods: Randomized controlled trial

Keywords: Maternal health messaging, mobile health, behavioral nudges

Abstract Read More

With IDinsight, I worked on the design of an ongoing randomized controlled trial that aims to assess the effectiveness of different types of behavioral messages on knowledge, attitudes and behaviors of pregnant and lactating women in South Africa. The study evaluates different messaging techniques that MomConnect, a maternal mobile health platform, can use to improve maternal health outcomes.
Evaluating the effectiveness of behavioral nudges at the honest reporting of COVID-19 symptoms

IDinsight

Research question: Can behavior nudges improve the honest reporting of COVID-19 symptoms?

Methods: Randomized controlled trial

Keywords: COVID-19, behavioral nudges, self-reporting

Abstract Read More

I worked on the analysis of a randomized controlled trial that IDinsight and Praekelt designed to understand the role of behavioral nudges in the honest reporting of COVID-19 symptoms by university-goers in South Africa. We found that an appeal for university-goers to be honest was most effective at increasing honesty in mobile HealthCheck daily self-assessment surveys. The results of this RCT highlight the small impacts that low-cost, digital interventions can have.