In Jobs: How Good are the Jobs the Economy is Creating? I referenced a Wall Street Journal report on jobs by Joshua Zumbrun.

I thought Zumbrun provided an excellent starting point for discussion. But I also noted what I labeled a “data error” as well as over-lapping categories that could lead to double-counting.

I received a nice email from Zumbrun in response to my email to him.

Hi Mish

Thanks for the email. We were using subsectors here and the figure we have labeled “durable goods” is the “durable goods: wholesale trade” subsector. You can check that those numbers are correct here:

As we’re working with a somewhat large set of data we didn’t realize the potential ambiguity, but we’ll update to reflect that these figures are “Durable goods, wholesale trade” sector. Your figures on durable goods manufacturing are correct — we just have the “durable goods, manufacturing” industry broken down by all its subsectors.

If you added them back together I think you’d see they align… however, if you add durable goods manufacturing to our sectors (which include all the durable goods manufacturing subsectors) then you’ll be double counting the job losses in that industry.

You’re right that food and drinking places and specialty trade subsectors overlap with their sector. We took the overlap out of the graphic but not the table. I’ll update the table to remove.

It didn’t quite occur to me someone would use the data the way you have and that the double counting in the table could pose an issue.
Thanks for the email.


In a second followup email Zumbrun replied …

“Here’s a link to the hospital data if you’re interested.

I certainly agree it’s broad and would obviously love it if this were available more fine-grained. But the data is right here if you’re interested, and the rising average wage over time doesn’t provide much support for the notion that “maybe the mix of hospital jobs is shifting toward lower wage.”

That said, I certainly do concede that this a very broad category and we don’t know the exact mix.

One person made a comment on WSJ about totaling up the weighted averages of their table. As you can see, that could not be done.

And as Josh implied, I inadvertently would have introduced an error if one took my numbers and plugged them back into the entire spreadsheet.

Yet, my numbers are accurate as far as I went.

In aggregate the numbers are not pretty, as I stated.

Mike “Mish” Shedlock