I’ve seen many posts today about our sluggish jobs recovery. Most people are pointing to data that show most of the job creation is in ultra low-wage, crappy sectors like fast food. (Here are James Pethokoukis, Tyler Cowen, and Mark Thoma on the matter.)

I think there’s a silver lining to this. Without considering part time jobs (which are not relevant to this post) there are three types of households: dual income, single income, and no income. The latter two indicate that one or both earners, respectively, cannot hold their job consistently, have high turnover, and are living off insurance.

Dual-income families have higher median incomes not only because there are two earners, but each individual earns more on average. Educated people from healthier backgrounds are more likely to get married – and stay married.

Let’s say I’m a genie. I can create jobs as I want. The process is simple, I specify the income said job will earn, and the market efficiently allocates my magical capital into productive labor. So I can wish for the marginal job to command $100,000, $50,000, or $20,000. Or less. Naturally, within this framework anyone with a genie would just hit some uselessly large number until all of America is rich and prosperous. But let’s stay within a reasonable bound.

Let me pose a question: what creates more marginal welfare: an efficiently allocated job pulling $20,000 or an efficiently allocated job earning $40,000. Most people will be quick to say the second job, of course.

But here’s the thing. Better jobs – like the ones we want to create – are more likely to go to families that already have an earner. That’s because the spouse of someone earning $75,000 a year may be “unemployed” because she (or maybe he) can’t get a great position, but will hold out instead of becoming a fry cook. She’ll take the executive assistant position paying $40,000, though.

As you can see, if your intention is to help America’s poorest – and that is precisely where the marginal dollar generates the most utility – you want to create jobs that are suitable to America’s poorest. Obviously there’s a point at which my tenuous argument breaks down. I would hands down rather disemploy a fry cook over an engineer or scientist (but not doctors – definitely not doctors).

There’s another, subtler point. Two $20,000 jobs are better than one $40,000 job – sure one shows up better on the output/hr productivity statistics: but the other employs two people. Furthermore, you may not believe wage flexibility is good (I don’t), but if you do the best way to achieve said flexibility is not through deflation of existing jobs but creation of crappier ones. That’s basic arithmetic.

This is not just a cyclical story. As many have claimed, automation and globalization will hurt the solid middle before it touches the bottom of the barrel. America is divided as much on income as it is on culture. A solidly middle-class housewife or husband will – to my approximation – rather remain unemployed because she can afford it rather than “stoop” to the jobs that are being created.

We may see more single-earner middle class families with a dual-earning poor who need it. This isn’t consolation for the grim path of our economic future, but it suggests there is reason to be optimistic at this stage in the recovery. These dynamics and frictions cannot be ignored: sometimes a higher paying job is not what the economy needs, hard as it is to believe.