Garett Jones has a really bad post at econlib. And I mean really bad. Jones seems to think he has hit upon something really clever by analyzing the income effect and substitution effect in a way that is sensitive to the resulting government spending. For those unfamiliar, the income effect refers to the way taxes might cause people to work more by making them have to work more to achieve the same take-home pay. The substitution effect refers to the way taxes might cause people to work less by making every given marginal unit of labor less rewarding.
Jones seems to think that we should look at how the revenues raised from the taxes are actually spent to determine whether these effects are actually operative. In what has to be the worst analysis I have possibly ever seen, he uses the following example:
If the tax hike is used for pure redistribution from the “average person” back to the “average person,” then the tax hike doesn’t make the “average person” poorer: The government is taking money out of everyone’s right pocket and slipping it into their left.
But if the income effect is gone, what’s left? The disincentive to work: The pure substitution effect.
Jones claims that in this scenario there is no income effect (after all, income is unchanged), and then — seemingly by process of elimination — concludes that therefore a substitution effect is present. This is wrong. If we analyze the taxing and spending together in one big lump, we conclude that this policy has no income effect or substitution effect. Because the prevailing policy is to give the taxed amount right back, there is no disincentive to work. An individual worker knows that the eventual income collected on a marginal unit of work is completely undiminished, and is thus undeterred from work by the initial tax. The marginal tax rate is effectively 0%.
That Jones does not see this is honestly stunning. I guess we all make mistakes, but this one is right in your face. It is so obvious that I have a slight fear I am missing something completely, and maybe someone in the comments can show me where I have gone wrong.
This is just one of the problems as well. One other (and there are more still) is that government transfer programs do not involve putting the money back in the hands of the person it was taxed from. And that really does matter. For instance, if you tax money from a middle income person, they might be most impacted by the income effect, and therefore work more and produce more. That would be the instant consequence. If the revenue was then transferred to a poor person, that might — via the income effect — cause them to work somewhat less.
Whether this leads to a net reduction in total production depends, but it’s plausible that it might actually lead to a net increase in total production. If the middle income person has a higher marginal productivity than the poor person, the extra labor induced by the tax could produce more than the reduced labor induced by the transfer. Such a program would thus make the overall economy more productive.
I am genuinely impressed by how off-base Jones is in this argument. It is like the troll physics version of economic arguments.