This is a tangent from the last post to add some harder evidence to my critique of Mulligan. I've been doing a lot of reading on poverty, some of this has bearing on discussions of incentive effects. It's generally been found that the level of benefits granted by welfare, or other anti-poverty programs, tends to have no statistically significant impact on employment (and the magnitude tends to be small, as well as insignificant). This is simply another piece of data that makes me sceptical of the supply side and incentive argument.
This of course isn't to say there is no impact due to government policies. Welfare benefit levels tend to be below survival level anyway, it's not surprising that variation in a too small benefit will have little impact. Larger amounts do show significant incentive effects, in particular means testing on Medicaid. People do work less to maintain Medicaid since it is not uncommon that losing Medicaid would result in a far greater loss of income than any additional earnings. What this shows is that incentive effects do matter, but only when they're quite chunky. They don't have any impact at the margins. If I ever get around to finishing my rationality posts I'll explain why this is in more detail. The basics of my perspective is that most marginal, maximizing, and individualistic behavior is the result of particular institutional structures that drive this behavior, it's purely situational. Absent these situational factors, people's more natural, less market oriented cognitive scripts gain prominence, which tend to be more granular and not very prone to either marginal or maximizing behavior. Government programs that are not designed to encourage marginal behavior, such as unemployment insurance, Medicaid, and a few others, tend to push people out of market based behavior and back into pre-market methods of thinking. The problem is not the government program, but that unlike voting, taxing, and many other government institutions structured to drive individualistic behavior, many anti-poverty programs are built according to more primitive impulses that do not encourage market based behaviors and instead erode these norms. Means testing being a primary culprit. But I've digressed.
The issue with all government programs is that as popular as it is to say things along the lines of we need to use a scalpel instead of a hatchet, the choice is better described between that of a chainsaw and a hatchet. We have to worry very much about unintended consequences. What this means is that we have to throw out the idea that we can impact either the economy or society with any precision or targeting. We can make programs that reduce poverty, joblessness, or lessen the impacts of unemployment. But if we try to target these programs narrowly at those who most need them we will make things worse because there is no institutional method that gives this degree of precision. What we can do is make programs that gradually taper off, such as unemployment benefits that slowly reduce or are reduced at a slow rate as if they were being taxed down. This wouldn't be fair, but we can't do fair with government, it's beyond our capacity. But to do otherwise is to create perverse incentives and this we should avoid to the degree possible.