Poverty policy's terrible tradeoffs.

Susan Edmonds canvasses the state of play around poverty, along with a few bits from me on the subject. It's a good piece.

Basically, policy is a pile of terrible trade-offs.

Cash assistance makes recipients better off. But providing it requires choosing among a few poisons.

Focusing assistance on those in most need through tight eligibility requirements makes sure that aid goes to those in most need - but at the cost of demeaning questions and testing and constantly justifying yourself to WINZ.

Targeting cash assistance to those in most need means clawing back cash benefits as someone is able to earn income, and that provides disincentives to work. And it provides incentive to feign eligibility. Worry less about the lying aspect and more about how it can split up families. And targeting also requires clawing back benefits as earned income increases, which provides disincentive to work.

Shifting instead to a guaranteed annual income gets rid of the demeaning questions, if you provide it at a level high enough to avoid having to layer on a welfare system on top. But providing that much assistance blows out the budget very quickly. Treasury's 2010 analysis reckoned that a GAI paying about the average amount received by someone on benefit would require a flat income tax of about 50% to cover the costs - and remember that that will be less than what's received by those currently worst off. So you'd still need to layer a welfare system on top of that.

Kevin Milligan's impossibility still holds. You can't pay a universal benefit high enough to not leave the worst off worse off without either having a very high phase-out rate (and consequent very high EMTRs), or blowing out the budget. And layering a welfare system on top of a GAI brings back all the problems above, albeit hopefully among a smaller cohort.

And you can't pretend that trade-off doesn't exist by appealing to other taxes that aren't currently in place. Why? Because if those taxes made sense, then they make sense regardless of whether you want to run a UBI. You'd then want to put them in on a revenue-neutral basis, replacing other taxes, first. If the new tax is really more efficient, then the deadweight costs of tax are a bit lower than before so the overall size of government can go up a bit in equilibrium. But whether that next extra lump of spending should go to a UBI or to other spending - you're back in the trade-offs world. You can't just magic up a new tax and pretend the best use for it is your pet project - some other proposal might be a better use of the funds. 

Shifting from cash transfers to in-kind benefits for some kinds of in-kind benefits solves part of one of those problems. If there are benefits that are valuable to someone in need, but useless to others, then you don't have to worry about people lying to get access to that benefit. Cash benefits require monitoring systems and intrusive questions to avoid diffusing the benefits beyond where they're most needed. Some in-kind benefits are self-targeting. So things like literacy programmes for example - people who are literate won't try to get access to them, and you might have reason to expect that improving literacy might help reduce need.

Cash should always be the baseline against which other things are measured. If an in-kind benefit is less valuable to the recipient than cash is, that's a pretty big strike against it. But say that every dollar's worth of spending on an in-kind benefit is valued at $0.95 by the recipient, but providing a $1 cash transfer would require paying out and extra $0.10 in monitoring costs and in leakage to those who weren't really eligible - then some in-kind benefits can wind up being better overall.

So everything above is terrible trade-offs. The most promising option remains what they government's been trying under the investment approach - better evaluation of what programmes can cost-effectively move people from benefit and poverty into self-sufficiency, where possible, and otherwise seeing what's most cost-effective in reducing misery. But there are still piles of problems there too - like difficulty in writing outcome-based contracts for NGOs delivering services; defining outcomes; and, need for monitoring to ensure that reductions in the government's long-term fiscal liability is a good proxy for what the government is trying to achieve. But it still looks the most promising.