Algorithms Can Catch Welfare Gamers

Fraud is a perennial problem for governments that provide any form of welfare assistance to citizens who are either indigent or temporarily disadvantaged due to the lack of a job or illness. Notwithstanding the concern for people who are unemployed, it is demonstrable that many welfare systems in high-income countries are so flawed that undeserving recipients game the system shamelessly. And to me this is a far more legitimate objection than the ideological arguments that I have encountered.

Deboarah Orr, writing in the Guardian here tries to understand the complex conflicts that come with being a benefits recipient. With the admission that she was also gamed the unemployment systems while holding a low-paying job, she attacks the recent proposal asking British citizens to report on benefits cheats. The sensible reason that she proffers is that there are a variety of reasons that lead people to try to gain from the system and rarely are undeserving recipients bad people trying to enrich themselves. It is difficult to believe that a one individual within a neigbourhood of people who know one another would easily snitch on others for the pecuniary benefits.

To my mind though, it is still possible to design a mechanism for finding those who game the system without asking them to betray their neighbours and friends. A better way would be to build a database of all recipients and make the data talk by building an algorithm that would sniff out the undeserving recipients. Indeed, if as this blog post stated, Steven Levitt and Ian horsely could construct a reasonable algorithm to detect individuals with a predilection towards terrorism, then I am certain that it is possible to design one for those who game welfare. So the market for devices to catch dishonest beneficiaries of welfare is just poorly designed. I would not mind giving it a try because there may be substantial gains to the public from such a system.