Consider their headline number - Missouri. Testing cost $336k and found 48 drug users. Assuming welfare costs $10k/person, it saved $480k making it a net positive. All the other states they discuss had significantly higher returns (5-10x).
This all assumes not a single drug user was deterred from applying and also that not a single welfare recipient was deterred from using drugs.
Further, none of these measure the question of whether poor people do more or less drugs. Most of these programs only drug test some people and have expensive human-administered programs to decide who to test: "complete a written questionnaire about drug use"“suspicion-based drug testing for each applicant”,“reasonable suspicion exists” that they might be illegally using “a controlled substance or controlled substance analog.”
The fact that these programs found very few drug users is merely due to the fact that welfare recipients aren't dumb enough to put "yes I do drugs" on a written test.
This article is pulling the social sciences shuffle - find a socially undesirable conclusion, write an abstract hinting that it's false, and bury the truth in the data tables.
The reason these rely on surveys instead of urine tests is because courts have deemed mandatory testing to be an unreasonable search, while random testing based on reasonable suspicion hasn't been shut down by courts (yet). Florida tried to have mandatory testing for all welfare recipients but the courts threw it out.
Nevertheless, these results are really astonishing and I think you're looking at them all wrong. Some bureaucrat in Utah thought that 460 welfare recipients were suspicious and ordered them tested, but only 18 of them tested positive. The bureaucrat was wrong 96% of the time, and the 4% usage rate is both under the drug use rate for the general population of that state (6% self-reported) and over-represented (because these were the 10% specifically selected people that the agency thought were using drugs).
Anyway, your profit and loss analysis seems fishy to me. It is highly unlikely that by denying 14 applicants their welfare checks the state of Utah saved anything near 100% of the face value of said check. In fact I would say it is quite likely that the state paid some fraction (with no upper bound) of that on subsequent support for and/or enforcement upon the unsupported people and their families, not to mention the general economic loss of not putting that money in motion.
If you want to do a more detailed cost/benefit calculation, go ahead. I'd love to see it. But you do need to include the benefits which that article pretty explicitly does not.
not to mention the general economic loss of not putting that money in motion.
Can you explain what this means? Sometimes I hear language like this describing Keynesian stimulus, but you do know we aren't in a recession, right? We don't have any slack labor force that can be tricked into working by reducing their real wages while holding nominal wages fixed.
Consider their headline number - Missouri. Testing cost $336k and found 48 drug users. Assuming welfare costs $10k/person, it saved $480k making it a net positive. All the other states they discuss had significantly higher returns (5-10x).
This all assumes not a single drug user was deterred from applying and also that not a single welfare recipient was deterred from using drugs.
Further, none of these measure the question of whether poor people do more or less drugs. Most of these programs only drug test some people and have expensive human-administered programs to decide who to test: "complete a written questionnaire about drug use" “suspicion-based drug testing for each applicant”, “reasonable suspicion exists” that they might be illegally using “a controlled substance or controlled substance analog.”
The fact that these programs found very few drug users is merely due to the fact that welfare recipients aren't dumb enough to put "yes I do drugs" on a written test.
This article is pulling the social sciences shuffle - find a socially undesirable conclusion, write an abstract hinting that it's false, and bury the truth in the data tables.