Visualization of Social Security Fraud(?): A STUMP Geeking-Out Special
Starting out at a key data source
After my last post, On Social Security Old Age Benefits Fraud(?), Elon Musk and his merry DOGEsters roll on, with additional claims:
Now, this is shorthand, but I can think of alternative explanations:
There is a data default birth year of 1875 from when Social Security was inaugurated 90 years ago (Thanks to Loner from GoActuary for that thought) and somebody was added without birth year at some point
Fat-fingered manual data entry with no data validation or poor data validation — somebody born in 1975 was entered as 1875 (or something similar)
Fraudulent entries came from multiple sources (illegal immigration, identity fraud, tax dodging, credit fraud, etc.)
Now, Elon never says fraud outright - just says they’re coming across crazy data items. That does not require fraud at all.
I have encountered real data in insurance systems generated from processes like the first two bullet items.
The people with screwed-up records may be completely ignorant that the government systems have completely wrong data about them. They may retire at reasonable ages (at which point they may find the error… and may not be able to correct it. More on that in a bit.)
But it does require crappy data systems to allow obviously wrong values to persist.
First, no data validation on entry and putting in defaults when there should be no default values. Or having manual entry when there may be automatic data entry.
If you have manual data entry, you must have checks and controls, unless the data doesn’t matter at all.
Second, data audits should be conducted regularly. These are different from financial audits. Data audits do require knowing information about the data, to be sure, including how it’s being used. Those who audit should be different from those who will be correcting any defects found.
And when people are not allowed to correct information, whether inside or outside the system, one asks why they’re not allowed. We will be hearing more about this, I’m sure.
Perhaps certain types of errors were being corrected, and others, not so much.
And more:
The full table:
Note: These are all the people in the Social Security system who are supposedly alive.
Not all the people in the Social Security system receiving benefits.
People in the responses to this post added up the numbers and noticed:
Musk responded, noting that there are tens of millions of active SSNs more than U.S. legal residents.
Yes, we know there are illegal aliens using SSNs, which is one source of Social Security fraud. There are also people using their dead relatives’ SSNs.
Some of these SSNs in the system may be lying quiescent, though. Just never flagged as dead. Not receiving benefits. Just sitting there.
Some of the ages we see are truly absurd, existing only due to manual data entry without data hygiene practices that would have prevented those >200-year-olds.
Note that none of this has anything to do with accounting audit experience.
Yet.
Improper Payments Data: Available for You! For Free!
After the last post, I found that, oh wow, there have been regular official reports on improper payments in various federal systems.
The U.S. Office of Management and Budget puts out annual reports on improper payments, as a result of the Payment Integrity Information Act of 2019.
This bill reorganizes and revises several existing improper payments statutes, which establish requirements for federal agencies to cut down on improper payments made by the federal government.
The Office of Management and Budget (OMB) may establish one or more pilot programs to test potential accountability mechanisms for compliance with requirements regarding improper payments and the elimination of improper payments.
The bill requires the OMB to update its plan for improving the death data maintained by the Social Security Administration and improving federal agency use of death data.
The bill establishes an interagency working group on payment integrity.
This extends work on a variety of government waste and fraud bills that extend back for years, including such work as the Federal Funding Accountability and Transparency Act of 2006.
But guys, there’s a website and datasets… and man, am I in heaven.
Annual Improper Payments Datasets: https://www.paymentaccuracy.gov/payment-accuracy-the-numbers/
They’ve got a drop-down menu, downloadable datasets, etc. etc. etc.
Let me keep it simple for this post, by only grabbing SSA (Social Security Administration) data.
Here is the information for SSA payments: total payments made, broken out by total payments and improper payments (that they could flag):
Yes, they’re there. They’re less than 1% of the payments.
Over this period, SSA made $7.1 trillion in proper payments (this is SSI as well as old-age and survivors benefits, the main program people think about when you say “Social Security”).
They also detected $56 billion in improper payments.
These are different orders of magnitude. Thus, “only” 0.8% in improper payments.
I can dig into the breakout of the sources of this 0.8% of improper payments, but let me put that aside for right now. The spreadsheet is below.
Spreadsheet
SSA Already Knew About Sources of Error
The things Elon has been parading to the public were already known within the SSA.
If you go to the Annual Improper Payments Dashboard, select “Supplemental Information” and start scrolling… you get all sorts of items of how these improper payments occur.
Here is a short screenshot:
If you start digging into the items, you see items such as this:
So — they have trouble accessing information about the age of applicants? Excuse me?
If they can’t even deal with age issues, it shouldn’t surprise you that there are issues with citizenship data.
There is a whole list of these data issues. Marital status, employment status, income, whether the person is dead…
All of these troubles have been known for years.
It’s not just the SSA that has a problem.
The SSN is being used for identity for all sorts of federal programs — it starts here.
Open the Books has been putting out annual reports on these Improper Payments data, such as this 2023 report. This is where Sen. Kennedy of Louisiana has been getting some of his information.
I was only showing the SSA bit.
I see why Elon and his crew are starting here. So many other federal programs key off the SSNs.
Except, it’s not being used as a key, in a database sense, is it? Because many of these SSNs are being used by multiple people. That’s a huge part of the issue.
The SSA knew there was bad data in there, and it looks like it was nobody’s responsibility to clean it up. If one tried to clean it up… someone might have gotten in trouble.
I didn’t even get into the adjudicated fraud in SSA yet. (That’s in the spreadsheet above.)
More to come.
The discrepancy in the younger ages needs to be looked into as well. When I used US census estimates for 2023 as a proxy for the current number, the 20-29 group has 43,829,532 people. Musk's Social Security value is 47,995,478, a difference of 4,165,996. We did NOT have that many people in those birth years die and not have their deaths recorded. Something else is going on.
My sister's mother-in-law died, and they found at least 4 dates of birth she had used. She did NOT want anyone to know her true age. Even her drivers license had the wrong DOB (YOB). I can only imagine the problem her SSA bureaucrat came across - the drivers license says one date, her marriage license has another, both of which are different than the date she used when originally applying for a SSN. It took quite awhile to figure out what the real date was. And then the life insurance death benefit was adjusted and paid out. SSA did not adjust anything, since she was already dead. Was that fraud? She certainly knew better, but it probably was just vanity.