In a post earlier this month, I reported on a Brief Analysis I did for the National Center for Policy Analysis, drawing on a journal article by Sabia and Burkhauser, showing that a hypothetical increase in the minimum wage from $7.25 an hour to $9.50 an hour would not be well targeted on poor households. The reason is simple: the vast majority of low-wage workers are not in poor households.

I was pleased to see that bloggers Tyler Cowen, Greg Mankiw, and John Cochrane linked to my post.

Tyler Cowen recently did a post pointing out that the Sabia/Burkhauser results on this issue have not, as far as he knows, been challenged.

I was careful not to say that an increase in the minimum wage would not reduce poverty. I did lay out how, if some of the low-wage workers lose their jobs, the gains to minimum wage workers would be less. But that job loss doesn’t mean that increasing the minimum won’t reduce poverty.

But now that I have thought more about it, there’s more to say. The reduction in poverty you estimate using a Sabia/Burkhauser approach will overstate the reduction in poverty. Why? Because it’s a partial equilibrium approach. Sabia and Burkhauser–and I’m not faulting them for this–don’t ask the following question: If the minimum wage rises and if the reduction in jobs is minimal, what happens to the households that buy the goods and services produced by low-wage workers?

When the minimum wage increases, costs for employers of low-wage workers rise. When costs rise, the supply curves and marginal cost curves shift up. When the supply curves and marginal cost curves shift up, prices rise.

Who pays those prices? In many cases, households that are close to poverty. When they pay higher prices, their real income falls. When their real income falls, some of them will become poor.

Moreover, when poor people pay higher prices they become poorer. Although that will not show up as increase in poverty–they were already poor–it’s not good.

As Steven Landsburg puts it:

The minimum wage takes from the (mostly) relatively poor people who buy a lot of fast food and gives to the (mostly) relatively poor people who serve it. When I go into McDonald’s in the morning, most of the customers strike me as less well off than the nice lady who serves me my Egg McMuffin.

So, whereas the direct effect of the minimum wage increase is to reduce poverty, this indirect effect will increase poverty. Which effect would dominate? It’s an empirical issue. My gut feel is that the direct effect would dominate and poverty would fall. But I don’t have a strong gut feel that the direct effect is much bigger than the indirect effect.

The clearcut implication, though, is that the hypothetical minimum wage increase would reduce poverty even less than I had thought.