Malcolm Gladwell, in the latest issue of the New Yorker, argues that the traditional approach to homelessness is inefficient. It'd be cheaper, he suggests, to identify the hardest cases—the chronic homeless—and hand them each a key to an apartment, then monitor them with dedicated social workers and health-care specialists. That's because continually treating them for substance abuse and other medical and societal ills, jailing them, sheltering them, and ejecting them back to the street costs even more. It's an interesting argument, raising questions of how societies should balance economic efficiency with the need for proper incentives and an intuitive moral code. But Gladwell's argument rests on a flawed interpretation of government costs—though in an email to Gelf he explains why his argument still holds.

Gladwell's crucial anecdote is the story of Murray Barr, a homeless man in Reno who cycled through the city's emergency rooms. Some rough calculations by concerned local cops set his unpaid medical bills at one million dollars—hence the article's title, "Million-Dollar Murray." Gladwell then compares such costs to the cost of his apartment-and-social-work scenario, and writes that the latter is much less pricey, sometimes just one-third as much.

But hospitals often charge the uninsured marked-up prices, assuming that many of the bills will never be paid. The few uninsured people who do strive to pay, often in installments over many years, are subsidizing the many who can't or don't. So an unpaid medical bill doesn't necessarily translate into actual cost. Lucette Lagnado has written extensively about this business model in the Wall Street Journal. From a 2003 article:

Most major U.S. hospitals are required to set official "charges" for their services, but then agree to discount or even ignore those charges when getting paid by big institutions such as insurance companies or the government. As a result, almost no one but uninsured individuals ever faces the official charges. In some ways, hospital charges are like automobile "list prices" or hotel "rack rates"—posted prices that everybody knows nobody pays. But in the case of hospitals, the pricing disparity isn't publicly known and falls most heavily on the vulnerable.

Before a magazine as heavily fact-checked as the New Yorker publishes such a provocative argument, it should seek better evidence for its lead anecdote than second-hand, back-of-the-envelope calculations based on "list prices."

Gladwell, when emailed by Gelf, agreed that Million Dollar Murray might be closer to $500,000 Murray if circumstances were different, but said his central point holds:

The prices charged by hospitals vary widely, and estimates of costs to the uninsured are always higher than quotes to people who belong to large insurance groups, with stronger purchasing power. So maybe if Murray were insured by Blue Cross, he would be $500,000 Murry. Pretty much everyone involved in this work, though, is convinced that regardless of how you price medical services, rent and a share of a caseworker is still substantially cheaper than having someone check into the [emergency room] twice a week, and spend three weeks in the hospital every year with head trauma or complex pneumonia. Interestingly, you could also make the argument that the Murray number undestimates his true cost to the system, because it doesn't take into account his share of ambulance services (which are prohibitively expensive) and his share of police services, and indirect costs, like what impact a visible homeless presence has on local business. There are plenty of business owners in, say, downtown Denver or San Francisco who would think that is the biggest cost of all.