This Bill is for the Byrds
The Senate version of the GOP budget reconciliation bill doubles down on draconian provisions of the House bill related to the Supplemental Nutrition Assistance Program (SNAP), in particular the state SNAP benefit cost share and excluding immigrants here lawfully for humanitarian reasons. But a very DC insider process known as the “Byrd Bath” is forcing Senate Republicans to rethink these proposals.
To recap, the cost share policy would force states to pay for a portion of SNAP benefits to recipients in their states, ranging from 5% to 25% of the total cost, depending on states’ SNAP benefit error rates. States will not be able to pay even 5% without breaking their budgets, meaning Congress is leaving it to states to do the dirty work of cutting people off benefits that they need to feed themselves and their families.
The GOP reconciliation bills would exclude immigrants who are in the United States legally as refugees, asylum seekers, or through various parole programs. This “would make up to 250,000 people in the country legally ineligible for SNAP,” according to the Congressional Budget Office (CBO). The bills also create a new $1,000 fee for asylum applications.
But Byrd Bath to the rescue! First a little background. Senate Republicans are using the budget reconciliation process because reconciliation bills can’t be filibustered. This means it can pass with a simple majority (51 votes) instead of a filibuster-proof 60 votes. Since 1990, a special rule has been in place called the Byrd rule, named after West Virginia Senator Robert Byrd, its primary sponsor. This rule requires reconciliation bills to include only provisions that directly impact the federal budget. The “Byrd Bath” is the process where the Senate Budget Committee and the Senate Parliamentarian study the bill to see if it includes any extraneous provisions — items that would not directly impact the federal budget. Anything identified as violating the “Byrd rule” may not be included in the reconciliation package. (Thank you Indivisible for the clear explanation of this process!)
This weekend, the Senate Parliamentarian ruled that both of these provisions — the state SNAP benefit cost share and the exclusion of refugees and asylum seekers from SNAP eligibility — violate the Byrd rule, and must be stricken from the legislation. This is good news! But as FRAC notes in the news release linked just above:
“The fight is far from over. The Senate will continue to try to hit the current target to cut hundreds of billions of dollars from SNAP and will be looking for a work around for these two provisions and/or seek other cuts. The bill remains problematic and urgent advocacy is needed – Leader Thune (R-SD) still intends to move the bill in the coming week.”
Allowing these provisions stricken in the Byrd bath to go through would be a cardinal sin!
Troubling New SNAP Data Collection
Meanwhile, a group called Data Index uncovered a notice of an alarming new National SNAP Information Database, which will include personally identifiable information (PII) like Social Security Number (SSN), name, address of SNAP participants (and applicants!).
This is concerning for a number of reasons:
This would be the first national database with details on all individuals participating in SNAP, raising the specter of USDA sharing individuals' private data with, say, U.S. Immigration and Customs Enforcement (ICE).
The data would be stored in the Amazon cloud (AWS). Breaches happen, and the prospect of putting 42 million Americans’ personal and very sensitive data in a private sector cloud is extremely troubling.
One of their goals is to use the database to identify noncitizens on SNAP who aren't supposed to be, and to track down overpayments. This will have a massive chilling effect on mixed status families (e.g., citizen children living with noncitizen adults, or a citizen family with a noncitizen grandmother living with them) applying for SNAP. Even if every member of the SNAP unit is a citizen or in an citizenship category eligible for SNAP, they might fear putting a noncitizen family member at risk, even if their loved one is not benefiting from SNAP. This feels like a workaround to the public charge policy that courts struck down.
The section describing the permitted uses of the data is incredibly vague: it allows USDA to share the data with other agencies if the data request is
"deemed by USDA to be for a purpose that is compatible with the purpose for which USDA collected the records." This could mean anything.
Finally, perhaps the biggest red flag: this data collection was approved by the Office of Management and Budget yesterday without any public comment period, because it was deemed to represent "no material or nonsubstantive change to a currently approved collection.” Excuse me? How is a national database, stored in the Amazon cloud, of every SNAP participant and applicant, including their name, address, and social security number, a nonsubstantive change? This has never existed in a national database before.
For context, SNAP is largely administered at the state level. States collect the data they need from applicants and participants to be sure they meet the eligibility criteria and receive the correct level of benefits. Up to now, states have not shared ANY of this with USDA. They only report to USDA the number of people and households participating in SNAP, and the total amount of benefits distributed to those people, by month.
USDA’s Food and Nutrition Service (FNS) hires quality control (QC) reviewers to go to every state SNAP agency and select a statistical sample of SNAP case files. QC reviewers collect detailed information on these cases, and these data are combined into a national SNAP Quality Control database (SNAP QC). These data are publicly available, though not widely advertised, and they require statistical programming software to make use of them.
The SNAP QC data are used by FNS to produce a number of annual reports on SNAP characteristics and trends in participation, as well as for frequent data analyses day to day and in a microsimulation model built and maintained by Mathematica Policy Research. Mathematica simulates proposed SNAP policy changes, as requested by FNS. I used to work for Mathematica running these simulations.
Two crucial differences between the SNAP QC data and the National SNAP Information Database described above are that:
The QC data are based on a sample — a small percentage of all SNAP households.
The QC data do not include any PII. Nothing can be linked to any specific individual.
The second one is especially important. Putting personally identifiable information on 42 million SNAP participants, plus however many applicants not yet participating (or even eligible) in a private company’s cloud service is a massive security risk. Moreover, while SNAP QC data are used to understand SNAP participant characteristics and economic circumstances to drive better policy, this new collection would be used, apparently, to target individuals deemed to be ineligible for SNAP. Would they be cut off of SNAP, or more? Would they be deported? It feels like this move is another step in the Trump administration building a massive surveillance system.
It is unconscionable that OMB, whose director Russel Vought is the architect of Project 2025, approved this without allowing any public comment. This should be a major news story.
Action items:
Use FRAC’s action alert tool to send a message to your senators and representative that they must defeat the big, brutal GOP reconciliation bill: https://secure.everyaction.com/lmBv70YdQkCuGFcip4BvgA2
Call your senators and representative too to be sure they hear you loud and clear.
While you’re at it, tell them how you feel about this stealth National SNAP Information Database.
Thanks for highlighting attacks on providing food for the poor. Hey, church-goers: WWJD? And this new aspect of massive surveillance seems particularly sinister.