Evaluating Behavioral Policy Applications Using Administrative Data

Paper Session

Friday, Jan. 6, 2017 1:00 PM – 3:00 PM

Hyatt Regency Chicago, Regency A
Hosted By: American Economic Association
  • Chair: Tatiana Homonoff, New York University

Direct Outreach to Small-Scale Borrowers: A Field Experiment Using USDA Administrative Data

Jake Bowers
,
University of Illinois
Nathaniel Higgins
,
Social and Behavioral Sciences Team
Dean Karlan
,
Yale University

Abstract

USDA’s Farm Service Agency (FSA) introduced a Microloan program in January 2013, a program which is designed specifically to serve the needs of small farms, beginning farmers, niche farmers, and farmers from historically socially disadvantaged groups. These loans are smaller than traditional FSA loans, and are designed to be more convenient and accessible to nontraditional producers, with a shortened and streamlined application and relaxed eligibility criteria. Because these loans are specifically targeted to farmers who often have less contact with FSA than traditional or large operations, the outreach strategy for informing potential loan recipients is an important component of program success. In this paper we report on data from two phases of direct outreach to farmers using USDA administrative data to market the program. In 2015 we sent personalized letters to randomly selected farmers in the southern U.S., an area that is dense with historically disadvantaged farmers, using the Agricultural Census as the administrative data source. In 2016 we repeated the exercise on a national sample, using FSA's list of existing customers as the population from which individuals were randomly drawn rather than the Agricultural Census. We report on the success of this direct outreach at generating interest, applications, and successful loans through the microloan program among targeted farmers.

Messaging to Promote Workplace Retirement Savings

Tatiana Homonoff
,
New York University
Will Tucker
,
Social and Behavioral Sciences Team
Jacob Goldin
,
Stanford University

Abstract

Simple interventions like changing the default or sending a short message can induce individuals to save more for retirement. However, messages that emphasize high savings rates may increase the amount that savings plan participants save while reducing the total number of plan participants. We study this possibility in the context of a field experiment designed to increase retirement savings by U.S. military service-members. We find that service-members who received a message emphasizing a low contribution rate were more likely to participate in a savings plan than were service-members whose message emphasized a high contribution rate, or no rate at all.

Medicare Letters To Curb Overprescribing

Adam Sacarny
,
Columbia University
David Yokum
,
Lab @ DC
Amy Finkelstein
,
Massachusetts Institute of Technology
Shantanu Agrawal
,
Centers for Medicare and Medicaid Services

Abstract

Inappropriate prescribing is a rising threat to the health of Medicare beneficiaries and a drain on Medicare’s finances. In this study we used a randomized controlled trial approach to evaluate a series of low-cost, light-touch intervention aimed at reducing the inappropriate provision of prescription drugs in the Medicare Part D program. The first such trial addressed Schedule II controlled substances. Potential overprescribers were sent a letter explaining that their practice patterns were highly unlike those of their peers. Using rich administrative data, we were unable to detect an effect of these letters on prescribing. Subsequent trials build on this null result with alternative interventions.

Low-Income Nonfilers and Refundable Credits

John Guyton
,
Internal Revenue Service
Pat Langetieg
,
Internal Revenue Service
Day Manoli
,
University of Texas-Austin
Mark Payne
,
Internal Revenue Service
Brenda Schafer
,
Internal Revenue Service
Michael Sebastiani
,
Internal Revenue Service

Abstract

This project examines how reminders affect tax filing among lower-income nonfilers (individuals who did not appear on a filed tax return but had income reported by third parties to the Internal Revenue Service). We present novel data on this population and results from two randomized controlled trials. The results demonstrate that one-time reminders increase tax filing, both to claim tax refunds based in part on withholdings and Earned Income Tax Credit benefits, as well as to voluntarily pay balances owed to the IRS. However, these effects do not persist. Consistent with recency effects, individuals who owe a balance due appear more likely to recidivate into nonfiling than those who receive refunds. Follow-up reminders continue to increase tax filing, particularly among individuals who previously had to pay balances to the IRS instead of receive refunds.
Discussant(s)
Hunt Allcott
,
New York University
Damon Jones
,
University of Chicago
Angela Kilby
,
Northeastern University
Alex Rees-Jones
,
University of Pennsylvania
JEL Classifications
  • H0 - General