Tapped’s Ezra Klein, who is usually sharp, offers a seriously misguided argument in favor of increasing the minimum wage:

[W]hile reasonable people can disagree on the impact of minimum wage laws, it’s time they stopped. William Niskanen, in arguing against a federal boost to the wage, trots out the same old canards about wage increases decimating jobs. And yes, if you jack the wage up to $16 an hour, jobs will be lost. But up to $7 over a period of years? The evidence doesn’t back him up. And, luckily, it’s so easy to check that you folks can play along at home. Just crosscheck this list of state minimum wage laws with this rundown of state unemployment rates…

[A]ny attempt to correlate minimum wage increases with joblessness falls on its face. When Clinton raised the wage in the mid-‘90s, low income employment skyrocketed. Some catastrophe. And we can take this as far back as folks want. Check this graph, showing the real value of the minimum wage (now at a historical low). Its peak was 1968. The unemployment rate in ‘68? A brilliantly low 3.5 percent.

Unfortunately, correlation does not imply causation. The fact that some states have high minimum wages and low unemployment and others have the reverse does not prove that the minimum wage has no effect on employment. It’s perfectly plausible that states with robust labor markets tend to increase their minimum wages. Afterward, they still have low unemployment, but it might have been even lower had there been no minimum wage. And unemployment might be even worse in states with low minimum wages and high joblesseness. The same argument applies to the point that the minimum wage’s real value was highest in 1968, when unemployment was low. It might have been even lower without a minimum wage. We just don’t know.

To properly assess the effect of changes in the minimum wage, you need an “all others things equal” situation — essentially, a natural experiment. The canonical evidence supporting Klein’s position is a David Card/Alan Krueger study of the restaurant industry in the labor market comprising Philadelphia and southern New Jersey. They found a minimum wage increase in New Jersey had no effect on employment relative to Pennsylvania, which did not increase its minimum wage.

However, such findings are controversial among economists. For instance, Paul Krugman writes that “most of [Card and Krueger’s] colleagues are unconvinced; the centrist view is probably that minimum wages ‘do,’ in fact, reduce employment, but that the effects are small and swamped by other forces.”

I think Krugman’s view is probably right. But in any case, if we’re going to have a serious debate about the minimum wage, we need to be talking about these types of studies, not correlations between minimum wages and unemployment.