Census-FNS-ERS Joint Project

This page provides the following information about the Census-FNS-ERS Joint Project: Next Generation Data Platform:

Overview

The Census-FNS-ERS Joint Project is a strategic partnership between ERS, the U.S. Census Bureau (Census), and the Food and Nutrition Service (FNS), the agency responsible for administering USDA’s domestic food and nutrition assistance programs. The Joint Project is a long-term cross-agency effort to acquire, for statistical and research purposes, administrative data from States on USDA food assistance programs and to leverage those data through linkages to surveys conducted by Census and to other administrative data files.

The project builds resources to conduct policy-relevant analysis using administrative data and linking multiple data sources, gaining insights and results that are unobtainable using any data source alone. The research helps inform USDA, Congress, State agencies that administer food assistance programs, and the American public on who participates in USDA food assistance programs, how program participation affects the lives of those individuals, and who does not participate and why. At the same time, the Joint Project informs decisions about how to improve Census surveys (which collect data on food assistance participation), the decennial Census of Population and Housing, and data-linkage processes.

Census conducts a variety of research projects that use administrative records linked to census and survey data. These projects are supported by the Data Linkage Infrastructure in which Census is expediting the acquisition of federal and federally-sponsored administrative data sources, improving data documentation and linkage techniques, and leveraging existing systems for governance, privacy protection, and secure access to these data. Census has been leveraging external data sources since the 1940s; the Joint Project is not unique—it is another important example in a long history of Census work in this field.

The Joint Project is pioneering in its use of USDA administrative data, and this partnership has led to the development of the Next Generation Data Platform where USDA administrative data reside and link to other data sets. 

The Next Generation Data Platform benefits ERS, Census, FNS and the States, and supports guidance provided by the U.S. Office of Management and Budget in M-14-06 (Guidance for Providing and using Administrative Data for Statistical Purposes). The project partners within the Census Bureau are the Center for Economic Studies (CES) in the Research and Methodology Directorate and the Economic Reimbursable Surveys Division (ERD) in the Economic Directorate.

Goals

The goals of the Census-FNS-ERS Joint Project are to:

  • Acquire at Census the State-level administrative data of the Supplemental Nutrition Assistance Program (SNAP, formerly known as the Food Stamp Program) and the Special Supplemental Nutrition Assistance Program for Women, Infants, and Children (WIC);
  • Leverage those data through linkage to other program and national survey data files;
  • Collaborate on understanding data quality characteristics of administrative records; and
  • Conduct joint research on issues that inform FNS and USDA on administering SNAP and WIC and inform Census on survey-improvement activities.

The Joint Project is a national undertaking. State SNAP and WIC agencies that participate send their confidential micro-data (involving millions of records that must be maintained in a secure data environment) to Census in exchange for State-specific analyses and reports. ERS researchers with Census Special Sworn Status and Census staff use the States’ administrative data and linked data for joint research.

The Joint Project has added an important component of studying the complex relationships between SNAP eligibility requirements and participation decisions of veterans (and their families) as well as issues of veteran’s access to health care. Census conducts the linkage between administrative records of the U.S. Department of Veterans Affairs (VA) or the Department of Defense (DoD), depending on availability, and State-level SNAP administrative records and Census surveys and censuses.

Partnerships

The two USDA programs in the Census-FNS-ERS Joint Project are SNAP and WIC. SNAP is USDA’s largest food assistance program, serving about 40 million people per month, about 12 percent of the Nation’s population, at a cost of approximately $65 billion in fiscal 2018. WIC provides nutrition assistance to some of the country’s most vulnerable populations, serving about 7 million people per month at a cost of approximately $5 billion.

Understanding USDA food assistance programs and policies gains strength through access to confidential micro-level administrative data on participation, duration, and amount of benefit receipts. Administrative records are known for their completeness and accuracy. Their power as a research tool is multiplied when linked with either a survey or another administrative file, supporting research that could not be conducted using any one data source. The participation of FNS and ERS in the project is essential for conducting research that informs policy and operations decisions of UDSA food assistance programs.

The availability of SNAP and WIC administrative records enhances Census operations and extends demographic and socioeconomic research capabilities through record linkage and statistical matching. Administrative records data can advance the mission of the Census Bureau by making operations more efficient, improving data quality, reducing data collection and processing costs, and reducing respondent burden. CES staff combine information from multiple sources to create new data products that could not be produced using single data sets.

The Joint Project addresses several barriers that historically have made it difficult to conduct research using administrative data from USDA food assistance programs. The project needed the cooperation of all three agencies to overcome these barriers: State-level USDA data, restrictions on allowable uses of administrative data, confidential data requirements, and statistical complexities.

State-level data: One barrier to research is that no national data exist for all program participants in either SNAP or WIC. Data for SNAP and WIC are collected by States, and reside at the State level. The Joint Project is undertaking a large-scale effort to contact all State SNAP and WIC agencies (which are different agencies at the State level) and to negotiate data-sharing agreements. The Census Bureau has the staff and expertise to make the contacts and develop the agreements. FNS involvement enhances the willingness of States to enter into data-sharing agreements for the project. In addition to the contributions of the Federal partners, the States’ SNAP and WIC agencies have State-specific expertise on their own Management Information Systems. Staff in the State agencies are the critical first link in the process of extracting data and providing explanations to the Joint Project on the structure, format and meaning of the fields.

Data restrictions: Statutory and regulatory restrictions on the use of SNAP and WIC administrative data present a second barrier. Specifically, the data can be used for research only if that research informs program administration. FNS approval of the Joint Project’s research enables it to meet the statutory and regulatory requirements. FNS has programmatic expertise on the issues important to SNAP and WIC administration. Another statutory restriction is that accessing confidential Census data requires research to provide a benefit to Census. A final restriction is that ERS researchers must obtain Special Sworn Status as Census agents.

Data management: Once the Census Bureau receives a State’s data extract containing millions of confidential records, the data need to be cleaned and edited, reformatted, and linked in a secure data environment where access is restricted to approved researchers working on approved projects. Within the Bureau, the Center for Economic Studies and the Economic Reimbursable Surveys Division have the infrastructure of the physical and intellectual capital for managing data to maximize use while maintaining data confidentiality and integrity.

Statistical complexities: ERS researchers contribute expertise in economics, other social sciences, and statistical disciplines. ERS links across administrative and survey data sets to conduct research, overcoming statistical complexities in using and merging large-scale data sets, such as modeling a USDA food assistance program and people’s behavioral responses to the program. ERS also provides financial support to facilitate data acquisition and management activities. ERS works in tandem with FNS and Census to identify the policy and economic issues undertaken by the Joint Project.

Data

Certain data collected in the Census-FNS-ERS Joint Project are available on a restricted-use basis to authorized Federal and non-Federal researchers working on approved projects through Federal Statistical Research Data Centers (FSRDC). Researchers interested in submitting an application to access the confidential SNAP and WIC data (for States and years for which the data are available) are asked to contact the FSRDC where they would conduct the research.

Census ERD staff reach out to State SNAP and WIC agencies to solicit their interest and participation. FNS encourages States to share USDA administrative data with the Census Bureau; some States also provide administrative data from the Temporary Assistance to Needy Families (TANF) program. For further information on administrative data at Census, including SNAP and WIC data in the Census-FNS-ERS Joint Project, see the Census Bureau Administrative Data Inventory.

Looking ahead

The scale of the Joint Project is large, and its scope is broad. While the project has achieved some long-term goals, it is currently working toward other goals.

At its fruition, the Census-FNS-ERS Joint Project:

  1. Obtains, updates, and maintains large-scale administrative SNAP and WIC micro-data files obtained from States (including confidential data on millions of records that must be maintained in a secure data environment), accompanied by critical information on the particulars of a State’s distinct legal, computer and technical-support infrastructure;
  2. Houses administrative data relevant for ERS research (and Census survey-improvement activities) that are obtained across the Federal government, including data supplied by the Departments of Labor, Treasury, Health and Human Services, Veterans Affairs, and the Social Security Administration;
  3. Provides ERS researchers with Special Sworn Status as Census agents access to Census’s premiere national surveys of households, including confidential data that are not publicly released;
  4. Leverages the information content of any one file by creating and updating linked databases using state-of-the-art file-linking protocols and infrastructure provided by Census staff (for example, Census’s Person Identification Validation System (PVS) methodology);
  5. Enhances the effectiveness of using the Platform’s data for ERS research through: 

(a) Improved frameworks and documentation for understanding the quality of raw administrative data as received from data suppliers, and
(b) Refinement of the data-intake process—the procedures by which raw administrative data are cleaned, organized, and prepared for analysis files used by researchers;

6. Provides access to data collected in the Census-FNS-ERS Joint Project; data access is provided to authorized Federal and non-Federal researchers working on approved projects through Federal Statistical Research Data Centers.

Readings

ERS reports

Participation in USDA’s Supplemental Nutrition Assistance Program (SNAP): Effect of Local Labor Market Conditions in Oregon

In this report, ERS explores the relationship between local economic conditions and the duration of able-bodied adults’ enrollment in SNAP. Researchers examine whether improvements in local labor market conditions in Oregon shorten enrollment times for these SNAP recipients.

Annual and Monthly SNAP Participation Rates

A key aspect of SNAP is the extent to which it reaches its target population—the rate of participation among people who are eligible for SNAP benefits. This report estimates an annual SNAP participation rate, which counts the number of people who participated at some time during the year as a share of people who were eligible at some time during the year.

Improving the Assessment of SNAP Targeting Using Administrative Records

SNAP provides food and nutrition benefits to low-income households based on a formula that adjusts for monthly need. This study assesses how well SNAP is targeted to low-income households by estimating benefit receipt by household income relative to poverty for the period 2008 to 2012.

Supplemental Nutrition Assistance Program (SNAP) Access at the State and County Levels: Evidence From Texas SNAP Administrative Records and the American Community Survey

This report links Supplemental Nutrition Assistance Program (SNAP) administrative records from Texas to the American Community Survey (ACS). The large sample size of the ACS enables the estimation of SNAP access rates for demographic subgroups and counties with the State and for demographic subgroups within the largest counties in Texas, helping SNAP administrators to better target outreach.

Amber Waves

Amber Waves, ERS's magazine, is a window into the agency's broad research program, covering production agriculture, food safety and nutrition, the food industry, rural economies, agricultural trade, and farm-related environmental issues. Published in a web edition, Amber Waves contains in-depth feature articles and research findings, including a variety of articles related to WIC and other food and nutrition assistance programs. The following articles are based on work conducted under the Census-FNS-ERS Joint Project using the Next-Generation Data Platform:
SNAP Participants Are More Likely To Leave the Program When Local Labor Markets Improve (July 2019)
Local Labor Market Conditions Impact Participation in USDA’s Supplemental Nutrition Assistance Program (September 2018)
Illuminating SNAP Performance Using the Power of Administrative Data (November 2016)
Investigating Temporal Effects on Measured SNAP Participation Rates (March 2016)
Linking Administrative and Survey Data Provides a More Complete Picture of Whether SNAP Benefits Reach the Poorest Households (September 2015)
New Analysis Reveals Significant Within-State Variation in SNAP Participation Rates  (December 2013).

U.S. Census Bureau working papers

Stevens, Kathryn, Liana Fox, and Misty Heggeness. "Does Source Matter? Using State-Level SNAP Administrative Records and the Transfer Income Model (TRIM3) to Evaluate Poverty Measurement," SEHSD-WP2018-16, U.S. Census Bureau, April 2018.

Fox, Liana, Misty Heggeness, José Pacas, and Kathryn Stevens. "Precision in Measurement: Using SNAP Administrative Records to Evaluate Poverty Measurement," SEHSD-WP-2017-49, U.S. Census Bureau, October 2017.

Colby, Sandy, Jose Debora, and Misty Heggeness. "How Well Do Individual Reports Supplemental Nutrition Assistance Program (SNAP) Take Up in Household Surveys?" SEHSD-WP-2017-03, SIPP-WP-276, U.S. Census Bureau, Center for Economic Studies, Social, Economic, and Housing Statistics Division, December 2016.

Scherpf, Erik, and Benjamin Cerf. "Local Labor Demand and Program Participation Dynamics: Evidence from New York SNAP Administrative Records," CARRA-WP-2016-10. U.S. Census Bureau, Center for Administrative Records Research and Applications, November 2016.

Harris, Benjamin Cerf. "Within and Across County Variation in SNAP Misreporting: Evidence from Linked ACS and Administrative Records," CARRA-WP-2014-05. U.S. Census Bureau, Center for Administrative Records Research and Applications, July 2014.

Meyer, Bruce and Robert Goerge. "Errors in Survey Reporting and Imputation and Their Effects on Estimates of Food Stamp Program Participation," CES-WP-1114. U.S. Census Bureau, Center for Economic Studies, April 2011.

Conference proceedings

Czajka, John L., Karen Cunnyngham, and Randy Rosso. "Simulated Versus Actual SNAP Unit Composition in Survey Households in Two States," in Proceedings of the 2015 FCSM Research Conference, Washington, DC: Federal Committee on Statistical Methodology, 2016.

Harris, Benjamin Cerf. "Within and Across County Variation in SNAP Misreporting: Evidence from Linked ACS and Administrative Records," in Joint Statistical Meetings Proceedings, Social Statistics Section, 2013.

Journal articles

Scherpf, Erik and Benjamin Cerf. "Local Labor Demand and Program Participation Dynamics: Evidence from New York SNAP Administrative Records," Journal of Policy Analysis and Management, 38(2): 394-425 (2019).

Meyer, Bruce and Derek Wu. "The Poverty Reduction of Social Security and Means-Tested Transfers," Industrial and Labor Relations Review, 71(5):1106-53 (2018).

O’Hara, Amy, Rachel M. Shattuck, and Robert Goerge. "Linking Federal Surveys with Administrative Data to Improve Research on Families," The Annals of the American Academy of Political and Social Science, 669, January 2017.

Last updated: Thursday, September 19, 2019

For more information, contact: Mark Prell