29 August 2013

The Current Cost of Investigating Welfare Fraud isn't Worth It

Just because an idea sounds reasonable and seems to make sense doesn't mean it is good policy. Sometimes a second look is needed.

I also should note that I am triggered to write this post because of an article (The Hand that Feeds You) by Al Jazeera America, and it is refreshing to see a new mainstream media outlet.

The prime example of a policy that sounds reasonable but is actually silly is mandatory drug testing of welfare recipients. It seems so much like common sense - government is subsidizing one's living and should thus check that one is living in accordance with law. Many companies require drug testing for their employees, so why not welfare recipients?

I've not ever worked for an employer that requires drug testing, haven't ever been randomly drug tested in school, so I'm not very familiar with the process. That said, while drug testing by employers is legal and use of drugs may change how certain laws apply, those laws, "[do] not encourage, authorize or prohibit drug tests" (EEOC). This fact weakens the idea, but doesn't completely make it seem unreasonable.

What makes drug testing welfare recipents seem unreasonable is it just doesn't do that much, such that the cost makes one question the resources, time, and effort going into these tests. Drug testing an individual apparently averages about $40 (NC Policy Watch). There are ways to try to avoid this cost, such as Utah using a screening questionnaire to decide who is high risk for drug use, and then only testing those who are high risk. That said, in one year it cost $6,000 for the screening questionnaire, and another $25,000 to test those who are high risk (Desert News). How many people of the 4,730 screened did that $31,000 catch? 12

Florida tried all out testing, only to find that:
Ushered in amid promises that it would save taxpayers money and deter drug users, a Florida law requiring drug tests for people who seek welfare benefits resulted in no direct savings, snared few drug users and had no effect on the number of applications, according to recently released state data (New York Times). 
To be fair, some outlets do report savings, as in savings of $40,800-$98,400 a year for a $178,000,000 program (Tampa Tribune). Oh, and those savings require one to assume that those tested are going to stay on the program for a year, and that we will continue to see 2% rejected a month.

These changes are also coming when we see headlines, like this one in 2011, "Welfare Spending Cut in Half Since Reform" (referring to the reform under President Bill Clinton) which includes charts like this:

Chart from CNN

The program that got me onto this subject, however, was the Al Jazeera America article on food stamp fraud. Here, it would seem to make sense to be stringent about investigating food stamp fraud, but in reality it doesn't make any fiscal sense. For example:
California found potential problems in fewer than 2% of its 150,000 SNAP investigations, costing the state about $35 million to find less than $20 million in wrongful payments... All told, the USDA contributed $126 million, and states spent millions more, for investigators to recover about $115 million. (Al Jazeera America)
So it clearly is costly, but these people who are committing the fraud must be getting absolutely rich off of it, right? Wrong. One study suggests the average income of those convicted of welfare fraud, including the welfare, was just $13,356 (Al Jazeera America). The scholarly article actually gives us a pretty good profile of the fraud:
Most fraud convictions were for unreported income with a median amount of $2,423, which on average, amounted to only $164 per month per household member. This represented part-time earnings of less than five-months’ duration that were scattered over a year or more, suggesting that parents made repeated attempts to plug gaps in inadequate subsistence budgets. For the majority, the fraud conviction was their first offense, and the mean amount of the overpayment was less than $5,000. With 64% receiving no child support, the median income from earnings and other sources—aid plus overpayment—was $1,113 per month. To place this figure in perspective, if defendants had reported all of their income, 73% still would have incomes far enough below the poverty line to continue to qualify for welfare. [emphasis mine] (Swan et al)
It should also be noted that these recipients are jumping through a number of hoops just to qualify, having to go so far as to waive their 4th Amendment rights (Al Jazeera America).

That said, investigating some types of fraud does sometimes make sense. To make the same contrast Al Jazeera does:
Every year, states conduct nearly 450,000 welfare fraud investigations. About 14,000 people are eventually criminally prosecuted for welfare fraud, most of them charged with felonies. (As a comparison, the IRS launches about 5,000 fraud cases each year, and about half of them end in convictions). (Al Jazeera America)
Pre-sequestration, there was a gap between what was owed to the government and what was actually paid (aka tax fraud) of about $400 billion (Politico). The sequester though appears to have cut $267 million from enforcing the tax code, meaning a loss of revenue of around $1.5 billion. The financial return there is extreme and obvious, unlike drug testing welfare recipients (which has a marginal return if you are comfortable making certain assumptions) and investigating welfare fraud (which costs more than it returns). Investigating tax fraud has significant and worthwhile returns.

Oh, and just for a final kicker I leave you with this quote:
Neither SNAP nor TANF adjust eligibility levels based on local costs of living — David’s salary would have been too much whether he lived in McAlester [, Oklahoma] or Queens, New York — and SNAP doesn’t cover a number of household necessities, like soap, toilet paper, pharmaceuticals and diapers. (Al Jazeera America).


  1. I would question the basis of the argument that drug testing for welfare recipients is not worth it. There are a few ideas not addressed in this article. One is how many welfare recipients dropped out of the system as soon as drug testing was mandatory?
    Take a look at the year over year drop in applications/approvals for TANF in Florida after the enactment of bill took effect, which was August 2011. If you were taking drugs and saw that you needed to be tested to receive TANF, the odds are you would not take the test at all, which is what happened in the following months. Over 35% fewer TANF recipients were approved in Sept 11, than in Sept 10. That accounts for a significant savings.

    1. There was a national decrease seen in TANF during that time frame, even when removing from consideration Florida's numbers. Furthermore this was a time when a lot of people were worried about TANF seeing as Congress failed to fully reauthorize it in 2010 and thus it existed in 2011 as a block grant program.

    2. In regards to Florida TANF application approvals, does it seem fishy that they dropped 35% yoy in Sept 11, dropped 46% yoy in Oct 11, dropped 44% yoy in Nov 11, and then the drug bill was dropped in late Oct and the TANF changes went into effect for Dec 11 and then application approvals increased 18% yoy in Dec 11? The implementation of the bill coincides directly with the yoy decreases and increases in TANF approvals. The block grant couldn't explain that. It's a pretty clear case of drug users leaving the program and then signing back up.

    3. This comment has been removed by the author.

    4. Let's just plot the number of applications for TANF from October 2009 to December 2012 (data available here: http://www.acf.hhs.gov/programs/ofa/programs/tanf/data-reports). The Florida law was in effect from July 1, 2011 until October 24, 2011 (as discussed in the court case: http://thinkprogress.org/wp-content/uploads/2014/01/document.pdf), and I have indicated this period with red lines.
      Here is just Florida: http://imgur.com/8bwG5tN
      Here is the US and Florida: http://imgur.com/HzIfPQb

      Edit: Forgot to include the court case, hence the deletion and reposting.

    5. I agree that they have flawed thinking. Maybe you don't get more money right away, but you do make things right by seeing that the correct people get the benefits, and , in time, it pays off!

  2. I see you show the applications remained close to the same, even though they did drop some in the months after July 2011. What counts is the number of approvals, which dropped almost 50% in some months yoy after drug testing was required. A significant number(35%) in Sept 11, decided to fill out the application, get approved, but then not take the drug test, because they would have failed. Notice that only cash assistance programs dropped significantly due to drug testing, while food stamps and medicaid approvals stayed the same yoy, because they were not drug tested. http://justoneminute.typepad.com/main/2012/04/drug-testing-and-public-assistance-in-florida.html -Derek