renting the nation's death file

Problem statement

Knowing who has died is a basic input to almost every payment system a government runs — pensions, Medicare, veterans' benefits, tax refunds, student loans — yet the United States has no national death registry. Deaths are registered by 57 separate vital-records jurisdictions (the 50 states, New York City, the District of Columbia, and five territories), each of which owns its own records; the closest thing to a national file is the Social Security Administration's Death Master File (DMF), which SSA assembles as a byproduct of administering its own benefits by buying state records on a per-record fee schedule and combining them with reports from families, funeral homes, and other agencies. Federal law (Section 205(r) of the Social Security Act) then restricts which agencies may see the state-sourced portion of the file, so the "full" DMF reaches only benefit-paying programs, and Treasury's government-wide Do Not Pay (DNP) system got access only through a three-year pilot that runs from 27 December 2023 to 26 December 2026. The unsolved problem is institutional design: a public good produced by 57 suppliers, priced by a fee schedule that GAO found was set without the statutorily required cost data, funded by reimbursements from agencies that "may elect" not to keep paying, and gated by a statute written for one agency's benefit administration rather than for a national data utility.

Why this matters

The value of getting this right is measurable: Treasury reported that the first pilot year (calendar 2024) identified, prevented, or recovered about $113.5 million in improper payments against $4.6 million in implementation cost, and projected more than $337 million in net benefits over the three years. The cost of getting it wrong is also concrete — SSA's payments to states more than doubled to $23.8 million in 2024, are projected at $25.9 million for 2025 and about $132 million over five years, and are passed through to agencies whose continued participation SSA cannot guarantee. On the other side of the ledger, erroneous death entries can close a living person's bank accounts or deny them credit and employment, and three territories still report deaths on paper, so timeliness and accuracy vary by jurisdiction. Every federal, state, and private payer that cannot access or afford timely death data keeps paying the dead.

What’s been tried and why it hasn’t worked

SSA has purchased state death data for decades through contracts negotiated with the states' association (NAPHSIS); the current schedule pays per record according to how fast the state transmits it — for example the rate for records submitted within 0–6 days rose from $3.73 to $9.08 and the rate for records submitted after 120 days rose from $0.01 to $1.87. GAO found the schedule was negotiated on timeliness alone: the statute requires compensation to comprise a fee for use of the data plus "the full documented cost to the state of transmitting the data to SSA," but "SSA did not obtain the required state cost information and therefore did not consider it during negotiations," and many vital-records offices "do not directly track the full costs for transmitting death data" because the same staff register births, marriages, and divorces without separating their time. SSA's method for splitting the total among receiving agencies "was not based on considerations related to agencies' proportional share of costs," agencies cannot see their final share until all agreements are signed, and SSA "has no guarantee that agencies currently receiving the full DMF will elect to continue receiving it." The Social Security Advisory Board argued in 2019 that the whole function should move from SSA to Treasury's Do Not Pay portal and be funded from general revenue rather than the Social Security trust funds, because death data benefit all agencies, not just benefit administration — but SSA still interprets Section 205(r) as barring it from sharing state data with the private sector and most non-benefit-paying agencies, and the DNP arrangement remains a pilot with an expiry date. Each fix so far has patched access for one more user rather than redesigning who owns, prices, and distributes the national file.

What would unlock progress

Progress requires treating death data as a shared national data utility with an explicit governance model: a cost-accounting standard that vital-records offices can actually apply, a cost-allocation rule among federal (and eventually state and private) users that is transparent before agreements are signed, and a statutory basis for access that is tied to purpose rather than to the identity of one benefit-paying agency. Adjacent precedents exist — shared-service pricing for inter-agency data hubs, club-good cost-sharing formulas in consortia, and activity-based costing templates used in public health registries — but none has been mapped onto the 57-supplier, multi-buyer structure of U.S. death registration. A pilot that costs one jurisdiction's transmission activities properly, or a simulation of how alternative fee schedules change state timeliness incentives, would give the parties something better than a flat per-record rate to negotiate over.

Entry points for student teams

A public-policy or operations team could build an activity-based cost model for a state vital-records office (using published budgets and interviews) to show what "full documented cost of transmitting" actually is, and compare it with the negotiated fee schedule. An economics team could model the fee schedule as a mechanism — how does paying $9.08 for a fast record and $1.87 for a slow one change a state's incentives, and what schedule would minimize total federal cost for a target timeliness? A design/data-governance team could draft a charter for a national death-data utility (ownership, access tiers by purpose, cost allocation, error-correction rights for people wrongly listed as dead) and stress-test it against the current statute. Relevant skills: public finance, mechanism design, data governance, health-informatics policy.