We've actually tried negative income taxes, and they seem to work

The latest issue of Chicago magazine has a great piece on the 1970s experiments with the Negative Income Tax (NIT) including in the deprived city of Gary, Indiana (famous for being the birthplace of the Jacksons). Inspired by Milton Friedman, and in an effort to reduce time, effort and effort spent administering welfare as well as stigma in receiving it, some of the poorest residents of Gary and four other poor areas received cash in randomised controlled experiments.

A lot of research was done into the treatment groups in Gary and across the other NIT experiments.

One study found that kids born to mothers in the treatment group had birth weights 0.3-1.2lb higher. Another found significant and substantial improvements in reading scores for children in treated families. And what's more, kids whose families had been in the programme for a number of years performed significantly better than those in it for a shorter time. School performance also increased significantly across a wide variety of metrics for early-grade students in a rural experiment in North Carolina, although the effect did not appear in the other major experiment, in rural Iowa.

It's not clear exactly where this effect came from, but the most plausible source is probably better nutrition and spending the extra money on housing in better areas. Most of the evidence suggests that recipients did not spend their windfall on expensive consumption goods.

There were no overall robust effects on marital stability, but this was misreported, and the mistaken belief that the NIT had led to black families breaking up was a significant factor in killing the proposal as a political possibility under Richard Nixon.

However, as well as the effects seen, the experiments seemed to find that the income effect—having more money overall—outweighed the substitution effect—lower and more predictable effective marginal tax rates making it more attractive to work—especially when it came to women. A glance at the table below makes this clear.

But it's possible to conclude that the fall in the amount of labour those getting the NIT supply (something like 5% for the poor groups studied, and around 2% estimated for the population as a whole) is quite small, and within the bounds of what we'd be willing to accept to substantially reduce poverty.

What's more, there are countervailing factors. One issue is the level of the guaranteed income. Some of the families received were guaranteed an income 150% of the poverty line. With a benefit level closer to the existing system, merely structured more clearly and predictably, we might expect a weaker or even positive response (although we alleviate less poverty).

A second issue is the long-term response. If the negative income can counteract large environmental problems, allowing families to move away from pollution and feed their kids better and achieve more in school, we might see these people enjoy improved long-term life outcomes. Even looking at things from a narrow labour supply perspective, we know that more educated and more intelligent people supply more labour over their lives, so the long-term effect may be neutral.

(Tables sourced from Widerquist (2005))