The Targetting vs. Universality Debate

A key issue any future opposition government is going to have to deal with is a very old one in First World social policy, but one barely talked...

A key issue any future opposition government is going to have to deal with is a very old one in First World social policy, but one barely talked about in our public sphere: should the government extend new social benefits to everyone, or should it try to target them narrowly on those who “need them most”?

Pose the question in those terms, and the answer seems obvious: why would you spend scarce public resources on people who don’t really “need them most”? Intuition comes down heavily on the “Target the Spending” side of the equation, and people tend to then favor it instinctively, before working through the dozens of unintended consequences and drawbacks of a targetting regime.

There are two big, insoluble problems with targetting. The first has to do with work incentives: the more targetted social spending is, the bigger a disincentive it creates for people to work.

Just as an illustration, imagine a program that targets only those who really need it most: people with no money at all. Imagine the government says “ok, we’ll give you Bs.100 a month, but only if you have no other source of income.”

At that point, what happens to the incentives to work of people who, say, are only able to find a job that pays Bs.10 a month? Well, if you have to give up a Bs.100 benefit to get Bs.10 income…you’re just not going to take the job.

That’s not a very good program.So it’s back to the drawing board for the Social Policy maker.

Now he may say, “OK, we’ll give you Bs.100 if you have no income, but we’ll take away Bs.9 of that for every ten dollars in work income you get. So, if you’re offered a job paying Bs.10 a month, you will only get a Bs.91 worth of benefit that month. Your final take-home pay packet will be Bs.101 each month.

So now you do have “a reason to work” – that extra Bs.1 a month, but you will, in effect, be paying a marginal tax rate of 90% on your income up through the end of your eligibility for the benefit program.

People facing 90% marginal taxes do not work hard.

Follow that logic down the line and you come to realize the First Iron Law of Social Policy Design: the more targeted the social benefits, the bigger the work disincentives.

And then we come to the second fatal flaw of targeting: the more exquisitely you want to aim your social spending at only the people who need it most, the more you need to go poking your big fat government nose into poor people’s household finances to assess just how much they’re making.

Means testing” is the kind of thing even highly sophisticated, highly competent First World governments do badly. Bureaucracies typically find it onerously expensive to carry out, and subjects often find it inherently demeaning and unfair. Inevitably spawning its own sprawling bureaucracy rife with opportunities and incentives to cheat, a means tested welfare state is just not something amateurs ought to be messing with: it’s the 42 km. marathon of public administration; the Venezuelan state wheezes up a flight of stairs.

In fact, the policy that carries least work disincentives with it is a fully universal one: “we’ll give you this money however much or however little you work.” Only full universality avoids the trap of creating outsized disincentives for people on low-incomes who receive social assistance. Carry that thought a little bit further and you can imagine a policy where low earners get more benefit for working more – one where marginal tax rates are lower if you work a crappy job than if you don’t work at all.

They actually do that in the U.S.: it’s called the Earned Income Tax Credit. And it works rather well.

These debates are decades old in the welfare states of the rich countries, but largely unheard of in Venezuela. Part of me is convinced that, if the opposition does get into power in 2012, we’re going to be spending a lot of time and effort needlessly, painfully reinventing this wheel.