Attachment 2- Statement of Work.docx

Statement of Work Household Consumer Food Purchase Panel

I. OBJECTIVE:

The USDA requires high-quality household consumer food purchase panel data to support current, ongoing, and planned research and data analytics activities. USDA will use the data for projects that support economic and policy research within the realm of USDA’s mission. Use purposes include (but are not limited to):

1. Describing the types and amounts of foods SNAP (Supplemental Nutrition Assistance Program) participants purchase using SNAP benefits and other methods of payment and comparing these households’ purchases against those of other households not participating in SNAP.

2. Estimating the effects of various SNAP food restrictions introduced via food restriction waivers (FRWs) on households’ food purchases, food purchase healthfulness, shopping behaviors, and other outcomes.

3. Using machine learning prediction methods to impute SNAP benefit distribution days.

4. Examining item substitutions over the SNAP distribution cycle.

5. Estimating food demand systems to obtain income and price elasticities across different household types (e.g., by SNAP participation status and by income quartile).

The data will support research projects on a variety of topics of interest to USDA agencies. The data may be used in combination (or linked to) other data to conduct this research, including commercial and proprietary data sources as well as administrative data sources owned by Federal, State, or local governments. USDA research projects are often undertaken in collaboration with experts within the Federal government and external to it, including but not limited to economists, policy researchers, and survey and data methodologists. Research using the data may be conducted by teams consisting of only USDA staff or a combination of USDA staff and external researchers at academic or non-profit institutions. The resulting aggregated findings (not the raw data) will be disseminated through multiple media, including but not limited to professional academic journals, USDA publications, the USDA website, internal briefings, and presentations at meetings of professional economics or other research associations. Before public release, all publications and reports will be reviewed by USDA to meet data disclosure requirements and to strictly maintain privacy and confidentiality requirements of data purveyors. USDA will also provide a two-week window (ten business days) for data purveyor review of outputs and projects to assess compliance with the data agreement and confidentiality requirements, after which USDA assumes the approval for outputs and projects has been granted and will proceed with the projects and outputs. The USDA will retain the right to use the data and information indefinitely to undertake its projects as specified above.

II. APPROACH:

USDA intends to award a contract to an organization (hereafter the “contractor”) to collect household food purchase panel data, deliver it to USDA, and provide adequate documentation and support to enable the use of the data for conducting research and improving data products. The work comprises the following tasks:

Task 1. The contractor will deliver nationally representative household food purchase panel data with supporting documentation and technical assistance meeting the following criteria:

1.A. Data:

1. Data includes high-quality, reliable information on a nationally representative panel of respondents and their grocery sector purchases.

2. Purchase data includes information on:

a. Purchase baskets including component foods/goods,

b. Timing of purchase,

c. Retailer or retailer format at which purchase is made including but not limited to on-line food purchases, supermarkets, convenience stores, drug stores, wholesale clubs, and mass-merchandisers,

d. Prices of basket and component goods,

e. Item-level identifiers, such as GTINs or UPCs, and

f. Use of one or more payment methods to include SNAP and WIC (Special Supplemental Nutrition Program for Women, Infants, and Children) EBT (Electronic Benefits Transfer) as well as other methods.

3. Data includes a refined product taxonomy useful for:

a. Understanding the characteristics of purchased goods, especially food-at-home goods,

b. Categorizing goods, and

c. Determining whether goods are targeted by state FRWs.

4. Data includes high-quality information on household characteristics, including:

a. Receipt/use of SNAP and WIC benefits,

b. Economic characteristics,

c. Demographic characteristics, and

d. Location of residence.

5. Data includes information on households and purchases from 2019 to 2031. The first deliverable data collection will constitute a history spanning 2019 to September 30, 2026. Subsequent data collections will cover the period from October 1, 2026 through June 30, 2031, to be delivered on a quarterly or biannual basis (see Section IV for details).

6. Data is cleaned, prepared for delivery, and deliverable one month after collection (e.g., data spanning January-March 2027 is deliverable by May 1, 2027).

1.B. Supporting Documentation:

1. Supporting documentation for the data will be delivered with initial data delivery and updated documentation with subsequent deliveries.

2. Documentation will include information on sampling design, weighting methodology, and known sources of measurement error.

3. Documentation will include data dictionaries to describe all data and variables received with each delivery.

1.C. Technical Assistance:

1. The contractor will provide technical assistance on interpreting, cleaning, and using the data as requested by USDA.

2. Assistance will be provided via email or teleconference.

Task 2. In addition to the data on the nationally representative consumer panel outlined in Task 1, the contractor will provide the same data on respondents/households and their purchases as outlined in Task 1 for all available respondents (including those not in the panel outlined in Task 1) during the same 2019-2031 period. This “full” data will support state-level analyses. If all data outlined in Task 1 is included in Task 2, Tasks 1 and 2 can be met by providing the “full” data outlined in Task 2 with sufficient information to recreate the panel outlined in Task 1.

Optional Task 3. The contractor will provide data for the nationally representative consumer panel outlined in Task 1 (and for the full set of respondents outlined in Task 2) on purchases made in other sectors, including:

1. Baby sector.

2. Limited service restaurant sector.

The contractor will also provide accompanying documentation and technical assistance for this purchase data.

III. PERIOD OF PERFORMANCE: Estimated 9/15/2026 to 9/14/2031 One Base Year period and four Option Years.

IV. DELIVERABLES:

1. Initial data spanning 2019 through September 30, 2026, and initial documentation

a. Data will be delivered no later than November 1, 2026.

b. Data will be delivered in an acceptable format to USDA Technical POC (TPOC) (e.g., parquet files).

c. Documentation will include necessary information to clean and use the data for research purposes (e.g., on data sources, collection procedures, estimation/imputation methods, data dictionary, variable dictionary, data quality, and limitations) and will be delivered in an acceptable format to both the contractor and USDA TPOC (e.g., .docx, .pdf, .xlsx, .json).

2. Periodic data updates and documentation updates

a. Data from October 2026 through June 2027 will be delivered every 3 months ; see table below for specific due dates for data delivery and the data period covered by each delivery. Minor delays of data delivery will be allowed subject to approval from USDA.

b. Data from July 2027 and forward will be delivered on a biannual basis ; see table below for specific biannual due dates for data delivery and the data period covered by each delivery

c. Data deliveries will include data for the previous three-month or six-month period specified and also update any data previously delivered (i.e., if additional cleaning or collection of data in previous periods is needed post-delivery).

d. Data shall be fully compatible, consistent, and able to integrate with earlier data and information (i.e. by replacing/updating previous data deliveries or appending to previous data deliveries), particularly with respect to file format, variable names, identifiers, etc. In cases where this is not possible or changes are necessary, the contractor must provide adequate technical assistance to explain changes and aid USDA in making necessary updates.

e. Updated documentation will be provided to reflect changes in data sources, collection procedures, estimation/imputation methods, data dictionary, variable dictionary, data quality, limitations, and other relevant aspects of the data.

3. Technical assistance with data on an as-needed basis throughout all data deliveries:

a. USDA requires technical assistance from data provider (not to exceed 10 human-time hours per month) relating to interpreting, cleaning, and using the data.

b. This assistance will be requested by USDA staff as needed, either via email or teleconference. Contractor shall respond to all technical assistance requests within 48 hours and if more time is needed to provide an adequate response, the contractor will inform USDA.

c. The contractor will be available for two meetings per month with USDA staff to discuss ongoing issues and data updates and to answer questions.

4. Approval of data-related outputs on an as-needed basis:

a. USDA requires a no-more-than-two-week turn-around (ten business days) from the contractor to approve outputs from existing projects based on the data, after which USDA assumes the approval for projects or outputs is granted.

5. Approval of new projects on an as-needed basis:

a. USDA requires a no-more-than-two-week turn-around (ten business days) from the contractor to approve new projects based on the data, after which USDA assumes the approval for projects or outputs is granted.

5 | Page