Making Sense of the GAO Reports on Gaps & Errors in the SSA Death Master File

There have been many recent developments impacting the insurance industry.  This discussion covers the following:

  • May 3rd 2013 GAO audit report identifying gaps in the SSA DMF and Verus Financial’s comments on same.
  • May 8th 2013 GAO testimony addressing the SSA death verification process, known DMF errors and protected record exclusions.
  • Our overview of how the different SSA databases relate to one another (to add context to the above).
  • Our own assessment as to the completeness and accuracy of the SSA DMF in relation to other known source of death records.
  • Our opinion on the appropriate use of the DMF for GRA audits and ongoing insurer compliance, including recommendations on adjustments that can be made to provide some relief from the burdens placed on insurers while still meeting the well intentioned spirit of recent regulations.

Our comments are based on our assessment of hundreds of pages of topic-specific documents and our fuzzy death matching on tens of millions of policies across a wide range of insurers.

Situation Overview

The Government Accountability Office (GAO) issued an audit report on May 3rd, 2013 titled: “Title XVI Deceased Recipients Who Do Not Have Death Information on the Numident”[1]. The key takeaway is that as many as 182,165 deceased recipients were not listed on the Social Security Administration’s 98 million record Death Master File (SSA DMF) and the report recommended that the SSA should take steps to increase the coverage.  The primary reason many of these deaths were not added to the DMF is because the recipients’ Personally Identifiable Information (PII) on the MBR, SSR, or death report did not match the recipients’ PII on the Numident.  The report attributed much of this to “minor differences in names and dates of birth on the SSR and Numident. Specifically, recipients had spelled their names differently or used married names on one of the records or the year of birth differed by more than 1 year while all other identifying information matched.” That sounds a lot like the SSA can benefit from the very same “Fuzzy” matching logic that is being used for the insurance industry.

On May 8th, 2013 the GAO released Testimony before the Committee of Homeland Security and Governmental Affairs discussing Preliminary Observations on the Death Master File[2].  The SSA disclosed its verification procedures (or lack of), as well as examples of known errors and additional insight into differences between the full 98 million restricted DMF and the abridged 89 million extract available for license from the National Technical Information Service (NTIS), which excludes State protected records.

Today we are discussing what these reports can mean to the insurance industry.  We reference additional comments made by Verus Financial President Jeffrey Drubner in a May 10th, 2013 follow-up article published in LifeHealthPRO titled “GAO critical of Death Master File” in which Drubner defended the accuracy and completeness of the DMF.[3]  We will also draw on our own extensive experience to provide what we believe is a balanced assessment of this issue.

Understanding the terms: Numident, MBR, SSR, SSA DMF, OASDI, NTIS, DACUS, OMG!!

As perspective, Numident is an acronym for “Numerical Identification System.”  It contains hundreds of millions of records representing all Social Security Numbers since they first were issued in 1936 along with the applicant’s name, place and date of birth, and other information.

A Master Beneficiary Record (MBR) contains information about someone who has applied to the SSA for Disability or Retirement benefits.  The MBR is used by the SSA to administer the Old Age, Survivors, and Disability Insurance (OASDI) program.  The Supplemental Security Records (SSR) is the main file used by the SSA to administer the Supplemental Security Income (SSI) program.[4]

To prevent erroneous payments to deceased recipients, the SSA’s Death Alert, Control and Update System (DACUS) is used to match reports of death received from Federal, State, and local agencies against the MBR and SSR.  The SSA records death information from DACUS onto the Numident and then takes an extract from the Numident to add to the DMF.[5]  A subset of that file, excluding state protected records, is made available to the public via NTIS.  You can review our prior blog for a discussion on recent legislation to restrict access to the DMF.[6]

In summary, the government maintains a large file of every person that has ever been issued an SSN.  They use other databases to track who has applied for or received benefits.  When someone dies, their information is added to yet another database.  If the SSA can accurately match the information from these transactional databases back to the original file of all SSNs, then the decedent information is updated and potentially available for inclusion onto the SSA Death Master File that insurers are familiar with.  Ironically, for the same exact reasons insurers cannot always match their own books and records to the DMF, neither can the SSA itself.

Steps being taken to improve the coverage on the SSA DMF – are they enough?

The SSA is taking some steps to improve data quality and matching including the ongoing efforts to migrate to the Electronic Death Registration System (EDRS), which as of March 2013 had 35 participating states.  In 2012 the SSA also launched two new claims processing initiatives (the Claims Enumeration Mini-Path and Autoclear project and the Internet Claim/CLIENT Release 5) to “ensure greater consistency between the name and Social Security number (SSN) information on the MBR, the SSR, and the Numident.” The SSA states that these initiatives will “significantly improve the integrity of, and consistency between, the death data on the MBR, SSR, and the Numident for current beneficiaries.”[7]

In other words, the SSA is aware of the problem and has been taking steps to address it.  While this can help close the gap on the omission of death records similar to those 182,165 (which would represent less than 2 tenths of 1% of the entire DMF) it does not fully address the larger issues which are:

1)      Insurers do not have access to the protected records.

2)      Per GAO testimony, the SSA does not verify all deaths and admits some are erroneous.

SSA DMF coverage and the impact of the November 2011 purge of protected records.

As noted in the May 8th GAO testimony, the full version of the DMF currently contains approximately 98 million death records but the version available to the public, and by extension to insurers, contains approximately 89 million records.  However, approximately 4.2 million of those missing records were removed as part of the November 2011 purge of protected records.[8]  For those of us that retained those records, the gap is smaller.  Never the less, that same purge also noted that approximately 1 million fewer decedents will be added to the NTIS’ DMF annually which means the gap will continue to widen.

Existing SSA practices in verifying reported deaths, or not

In the interest of brevity, it is worth reviewing the GAO testimony as referenced in its entirety.   The key takeaway is that the SSA does not verify death records submitted from what it deems to be reliable sources, but it does verify those from less reliable sources.  According to SSA officials, the agency only verifies death reports for individuals currently receiving benefits because “it is essential to its mission to stop payments to deceased beneficiaries”.  A chart is shown below:


Table 1: Types of Death   Reports and Whether SSA Verifies Them


SSA verification of   death

Source of Death Report



State (EDRS)



Funeral directors



Family members



Post offices



Financial institutions



Centers for Medicare & Medicaid Services



Department of Veterans Affairs



State (non-EDRS)



Source: GAO analysis.

The SSA’s analysis of existing DMF records also identified potentially erroneous information including:

  •  130 records where the date of death was recorded to occur before the date of birth
  • 1,295 records where the recorded age at death was between 111 and 129
  • 1,791 records where the recorded death preceded 1936, the year SSNs were first issued

The SSA believes some of these questionable records were added prior to the mid-1970s when manual keying errors were more likely.  Today, the SSA uses a variety of edit checks to minimize these conditions.   Our own studies show somewhat different statistics than these, but are not material for today’s purpose and are therefore being omitted from this discussion.

The impact of DMF errors on audit-level “GRA” reviews

In the May 10th article on LifeHealthPRO, Verus Financial President Jeffrey Drubner stated that “The DMF is an extremely accurate and valuable tool that may be utilized by insurance companies and financial institutions to identify deceased individuals, and the GAO’s testimony is simply part of on-going efforts by the government to make the DMF even more reliable.”

We agree with these comments.

In this same article, Drubner noted several widely referenced statistics claiming DMF accuracy rates exceeding 99.5%.  While the SSA admits that no all-inclusive study has been conducted, it is our opinion (partly based on insurer feedback) that the stated accuracy rates are generally reasonable.  We also believe that when relied upon as part of an audit-level Global Resolution Agreement (GRA), the errors that do exist are further mitigated by the following conditions:

  • Most errors in which living people are added to the DMF today are removed fairly quickly.  We will be conducting a study to demonstrate this but the best explanation is simply that when a living person is added to the DMF in error, their credit card issuer will suspend their accounts almost immediately.  This in turn will lead to the individual applying to have the erroneous death notice removed from the DMF.  For a policy to be reportable under audit, the insured would generally need to have been deceased for about 3 years, which is ample time for most false death reports to be corrected.
  • Unlike organizations that rely on the DMF to immediately terminate credit or benefits, insurance companies not only have the aforementioned “3-year” delay (within an audit), but also have an opportunity and obligation to investigate the death report before initiating the claims or reporting process.  Still, it is important to acknowledge that the research window is small and insurers would bear additional research expenses in dealing with the occasional false positive.
  • Not all errors are created equal.  It is true that the addition of a living person to the death master could cause a completely false positive, but many errors relate to more minor conditions such as isolated typos in SSNs, names, or dates of birth or death.  When using GRA level standards that generally require 4-point verification to be considered a valid match, such errors are less likely to result in false positives then they are to preclude the identification of a legitimate decedent.

Thus, from an audit level Global Resolution Agreement (GRA) perspective, Verus’ Mr, Drubner is in our opinion correct.  The exception being some cases in the AIG and MetLife GRAs in particular where lesser quality matches are accepted due to missing data points in the insurers’ historical books and records.

The impact of DMF errors on RSA and NCOIL state specific regulatory reviews

Once again, the reality is that the SSA DMF is overwhelmingly reliable and for all practical purposes the only source with enough critical mass and transparency to be considered the de facto standard for insurers.  However, now that the SSA has taken greater steps to provide visibility into some of the problems that exist, it creates an opportunity to start a dialogue about some subtle changes that the states can consider which will maintain the spirit of the new regulations while providing some important relief to insurers that are struggling to do their best with an imperfect DMF.  Specifically:

  • While the aforementioned cases of living persons being added to the SSA DMF in error are generally reversed quickly, that is still not instantly.  Accordingly, regardless of how often states require DMF reviews we would suggest that an additional “rest” period be added such that the most recent death records can optionally and temporarily be excluded from said review.  For example, if conducting a mandated review on May 15th, 2013 consider allowing the use of death records through no less than March 15th, 2013.  That additional window for recent deaths would not only allow more of the erroneous records to work their way back off of the DMF, but it also recognizes the insurance industry’s position that a reasonable grace period will show greater sensitivity to the grieving process.  Even with such modest delay, the decedents will still be identified in a subsequent mandated review period though many will have already come forward on their own terms to complete the claims process through normal channels.
  • Secondly, while more common typographical mistakes on the SSA DMF do not pose a great concern within a GRA-level audit due to the weighting of the 4-point match criteria, errors specific to SSNs can place a larger burden on insurers who are attempting to comply with Regulatory Settlement Agreements (RSA) or any of the state-specific “NCOIL” regulations.  This is because these requirements shift the burden to insurers on nothing more than an SSN-only match.  This can also be partially solved with a “rest” period however another option is to allow insurers to incorporate some additional criteria across Date Of Birth (DOB) and name to further qualify an SSN-only match for research.  Clearly, all parties would have to work closely together to craft a modified approach that truly meets the needs of the states, insurers and above all, the owners without creating any unintentional conditions that may compromise the core objectives.

Appropriate use of supplemental sources of deceased data

In our opinion, the reality is that despite its gaps and occasional errors the SSA DMF remains the premier and only viable source of death data that should be relied upon for the purposes intended for the insurance industry.  Having conducted many reviews, we at Cross Country Computer (CCC) recommend the following best practice approach:

Begin by licensing the SSA Death Master File and contract for weekly updates of adds, changes and deletes.  As these records are held to a higher standard and defined specifically in the assorted GRA/RSA and state regulations they should be treated separately to allow segregation of matches to ancillary sources.  Retain the 4.2MM records and associated data elements that the SSA removed as part of the protected records purge in November 2011.  In addition to the SSA DMF, Cross Country also has developed a proprietary database of military decedents that is not included on the DMF, as well as a variety of supplemental deceased sources some of which include both name and address.  These supplemental files may be compiled from a combination of state and local, credit bureau, postal and marketing sources.  They often have less transparency in data collection methods, questionable or non-existent verification processes, and lower reliability than the SSA DMF.  As such, they should primarily be used as confidence boosters or reducers.  In other words, if there is a marginal match to the SSA DMF that also matches to a supplemental file, then in most cases there is a greater likelihood that the insured is deceased – There are some exceptions, but we can’t share all of our secrets!  In the absence of an SSA DMF match, some clients use pure supplemental-only deceased matches as a means of narrowing a universe for additional research.  These supplemental sources would become more important in the unlikely event that legislation is ever passed limiting an insurer’s ability to have timely access to the SSA DMF.  Aside from this though, our strong recommendation is to avoid intermixing supplemental sources with the DMF results as doing so risks compromising the integrity and reliability of the overall process.

Concluding Thoughts

Despite the whirlwind of recent controversy surrounding the completeness, accuracy and accessibility of the Social Security Administration’s Death Master File (SSA DMF), we strongly believe that it remains the best and most reliable source of death records available to insurers today and in the foreseeable future.  The SSA’s ongoing effort to further improve its accuracy reinforces our opinion.

We believe that the SSA DMF is the appropriate source for GRA level audit reviews, and would be concerned with allowing the formal use of other sources as doing so will only create a greater number of false positive matches, cause insurers additional burden in their review, and due to the arguably short window available for research, could lead to property being reported in error.

We are an advocate for ensuring the DMFs continued availability to entities who can demonstrate a reasonable use need, including insurers. Because we have not found other available sources of death data to approach the scale and reliability of the SSA DMF, we believe that the optional use of such alternate sources is helpful but should be segregated to ensure the integrity of the SSA DMF results.

We feel that there is an opportunity to make modest adjustments to the Regulatory Settlement Agreements (RSA) and state-specific “NCOIL” regulations in order to maintain the spirit of their intent, while also helping insurer’s reduce the number of erroneous matches that have to be reviewed before the SSA has had an opportunity to correct at least some of them.

We believe that the increased discussion and transparency regarding all of these matters will help lead to greater clarity and consensus among all parties going forward, and will ensure the best outcome for those with a rightful claim to the assets in question.

Please refer the article, Sometimes It Pays to Be Dead, is a layman’s overview of the composition of the SSA DMF with pictorial.

For more information or a printable version of this blog, contact Thomas Berger at (631) 220-6947 or via email to

Tom Berger

About Tom Berger

Thomas Berger is CEO of Cross Country Computer. Tom joined Cross Country in 1991 and acquired the company in 1996. Tom has served as Treasurer of the Unclaimed Property Professionals Organization (UPPO) and holds the patent for APEARS™. Tom was inducted into the Direct Marketing Association of Long Island (DMALI) Hall of Fame in 2012 and is a lifetime member of MENSA, the international High IQ society. Tom holds a BS degree in Management and Marketing from RIT.
This entry was posted in Data, Data Management, Hygiene, Unclaimed Property. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *