This paper studies how public sector managers impact public service provision, both through the amount of services provided and how those services are provided. I use novel administrative data containing 30 million application review decisions that allocate $70 billion per year in public benefits for Medicaid, SNAP, and TANF in the state of Texas. Managers in this setting play an important role overseeing teams of case workers but have more limited tools and incentives for impacting outcomes. First, I document wide variation in throughput (cases reviewed per case worker), permissiveness (share of cases approved), and miss rates (share of cases that are denied and successfully reapply) across manager teams. Second, I use variation in case worker-manager assignments to show that managers explain 8-10% of the overall variation in case worker throughput, permissiveness, and miss rates. Improving managers from the bottom quartile of throughput to the 75th percentile would increase organization-wide throughput by 5.6%, which is 3.7 times greater than doing the same for the equivalent number of case workers. Third, I find that manager throughput ("quantity") and accuracy ("quality") are uncorrelated, reflecting significant heterogeneity in both manager productivity and preferences. This means that staffing policies aimed at improving manager throughput won't have unintended consequences on accuracy or decision-making, or vice versa. My findings suggest that managers are useful intermediaries through which policy changes and bureaucratic improvements can be targeted.

Governments generally rely on private vendors to distribute in-kind benefits. The types of vendors that participate can affect beneficiaries, local markets, and program costs. We study the effects of a dramatic increase in the number of food stores accepting SNAP benefits during the Great Recession. To do so, we combine several datasets: administrative records on all SNAP stores, information on all food stores in the U.S., a large household purchasing panel, and a panel of retailer transaction records. We find that the new SNAP stores are largely non-traditional grocery retailers like club stores, dollar stores, drug stores, and mass merchandisers. The adoption of SNAP by these stores decreases the average distance to the nearest SNAP store by about 8% between 2008 and 2012. We find that SNAP adopting stores experience increases in overall sales of about 5%, increase product variety, but do not change pricing. We find that stores who experience greater competition from these new SNAP adopting stores lose about 1% in sales, but do not respond by changing their pricing or variety. Using this information, we estimate a model of retail demand to measure welfare gains from decreases in shopping costs from SNAP adoption for both SNAP eligible and ineligible households. We find welfare gains for SNAP eligible households, while SNAP ineligible households experience minimal welfare gains.

This paper examines how household shopping trips, food expenditure, and nutrition are impacted by the entry of dollar stores. We use an event study approach with food retailer location data from Nielsen TDLinx and the household food expenditure from the IRI Consumer Panel dataset. We find that when a dollar store enters a household’s zip, households shift food expenditure to dollar stores from other food retail channels, with larger effects for low-income, low-access, and non-metro county households. Households shift food spending away from perishable product groups with limited offerings at the dollar store like dairy and meat, but by a relatively small amount. We find small impacts of dollar store entry on nutrition, particularly for households living in low-access areas. These effects are about 0.03 standard deviations, or about 5% of the nutrition-income gap, and are driven by decreases in dark greens, legumes, protein, and healthy fat and increases in refined grain. This suggests that household store choice could have an important role in food access and nutrition.

In many settings managers are promoted internally and selected from the best workers. However, the skills needed to be an effective manager often differ from those needed to be a good worker. It is unclear if good workers can replicate the processes and decision-making philosophy that made them a good worker across a team of workers. In this paper, I explore what makes an effective manager. I use rich microdata for public sector workers and managers administering public benefit programs in Texas to measure a worker's performance and decision-making before and after being promoted to manager. I use variation in worker-manager assignments to measure the causal impact of workers and managers. First, I study if workers that get promoted tend to have different quantity, quality, decision-making, experience, or demographic characteristics. Second, I determine what factors predict manager performance and decision-making, including historical worker performance and the performance of the manager that trained them.

Interviews for public benefit programs have multiple functions that are not well understood. First, interviews are used to gain important information about clients that could improve a case worker's ability to identify ineligible households. Second, interviews are used as an application screen that could better target benefits. Third, interviews are an important opportunity for applicants to receive one-on-one assistance with their application. The relative importance of these factors and how they correlate with beneficiary need and behavioral biases are important for understanding optimal interview policy. I utilize rich micro-data for SNAP in Louisiana that includes detailed applicant information including the completeness of the original application and what information was updated by the case worker. I use interview waivers that suddenly removed and reinstated interviews as variation to identify how interviews change program participation. I then explore the role of each mechanism in the overall participation effect, and what types of households were most impacted by each mechanism.

This paper investigates whether the federal school nutrition reforms enacted after the passage of the Healthy Hunger-Free Kids Act of 2010 had spillover effects onto the healthfulness of student household grocery expenditure. Focusing specifically on the Smart Snacks in Schools regulation and household grocery expenditure on unhealthy snack food, a difference in difference framework is used to compare the relative changes in unhealthy snack expenditure for households in states with and without strong state snack regulations prior to the federal regulation. The results indicate that the federal regulation decreased unhealthy snack expenditure by about 5% after three years for households with a student. This suggests that spillovers from school policy enhance the impact of the policy.