Here’s the interesting set of questions from the “Mish Mailbag” summarized as follows: How Accurate and at What Cost is BLS Data? Why isn’t there a private industry alternative?”

The questions are from reader “Trevor”, a grad student, forwarded from his professor.

When I first read the questions, I thought they were about jobs because the BLS stands for Bureau of Labor Statistics.

Just as I was about to hit “send” with my analysis on jobs, I re-read the email and realized the question was about the CPI, not jobs.

This is what happens when you get hundreds of emails and you make a concerted effort to answer them quickly.

So, Trevor gets a bonus two answers for one question. Let’s deal with jobs first as that was my initial response.

Private Jobs Reporting

  • Gallup produces employment data continually. I believe its reporting is as accurate as the BLS.
  • TrimTabs estimates jobs via tax collection data, also with free reporting.
  • ADP offers free advance reporting every month, but then goes to great lengths to smooth its data to match the BLS.
  • BLS Jobs Conversation 

    I still believe the BLS double counts part-time jobs. In telephone conversations, the BLS admits my assertion it is possible.

    A simple sort-merge program weeding out duplicate social security numbers would settle the debate.

    I contacted ADP twice on this issue and received no response.

    The BLS told me they would like to weed out duplicates but they do not have access to the data for security reasons.

    This about that for a second. What is the problem here?

    The answer is unencrypted social security numbers. If social security numbers were encrypted in government databases, then matches on encrypted hash keys would reveal duplicates without disclosing the underlying social security numbers to anyone.

    A simple sort merge program on raw data could do the same, coming up with a count only, so there would be no reason for anyone to see the social security numbers.

    Print Friendly, PDF & Email