>To add insult to injury, means testing[2] often costs more than the cost of fraud in social benefit programs!
The point of means testing isn't to fight fraud, it's to better divert a limited pool of resources to the neediest people. If you have a pool of money to distribute, only giving it to the poorest 25% (ie. means testing) means the recipients can get 4x the money compared to giving it to everyone. Sure, there are ways this can be done badly through onerous requirements or whatever, but the cost of means testing vs the fraud rate is a completely irrelevant metric.
The problem is that means testing programs often have a cliff where you go from qualifying to not qualifying instead of being a gradual ramp-off. In other words, earning $1 more can remove $100 in support from the government instead of something more reasonable like only removing 10c at the beginning and up to $0.9 if you’re already on your way out.
The common criticism of means testing is that more money is spent on the bureaucracy implementing the testing than if you just took the money you otherwise spent on bureaucracy and gave it directly without means testing.
There's only so much money to go around, if I chose to spend more money on preventing fraud than the fraud would cost, I'm prioritizing righteousness over helping people.
If you have funding for 500 people, and fraud would cost you funding for 10 people. If you spend the cost of funding 40 people on fraud prevention, you're harming the 30 people you can no longer fund to keep those 10 from undeserved funding.
The point of means testing isn't to fight fraud, it's to better divert a limited pool of resources to the neediest people. If you have a pool of money to distribute, only giving it to the poorest 25% (ie. means testing) means the recipients can get 4x the money compared to giving it to everyone. Sure, there are ways this can be done badly through onerous requirements or whatever, but the cost of means testing vs the fraud rate is a completely irrelevant metric.