In the future, so you don't sound stupid on the BLS statistics on unemployment

The BLS monthly report is a poll. It is not based on the number of people collecting unemployment benefits. It has nothing to do with the number of people filing for benefits. It does not measure people who have lost benefits, it is a large poll.

They call a bunch of people (around 400,000 to get 60,000 data sets) and ask a series of questions till they get the pre-described demographic data set that is representative of the population in their model.

The weekly number is another poll, except it is done of businesses to find out if they have increased or decreased their monthly payroll. Again, it isn't based on actual payroll numbers, it isn't based on layoff information submitted to states, it isn't based on any financial information. It is a poll.

For these reports, how they classify people and how they determine what the population is, like any poll, is probably about equally as important as how they ask the questions. The raw data is also adjusted depending on what month it is as a way to smooth the data. People build less houses in December and January in the North East and Midwest so adjusting the construction sector during that time frame makes sense. Also tourism seems to go away on the New Jersey and Delaware beaches around late September and pick up around late May (I can't imagine why?).

There could be actual job losses in the retail sector or any other sector, and the BLS could report hirings, because the job losses are lower than historical averages of job losses at that time of year. The reason why is not asked, assumptions are made about the past and taken into the present.

There is lots of data out there where you can get information on unemployment and wages. The BLS report is one data set. I have a preference for tax collections, unemployment claim payments and new claim filings (I'm an accountant, cash flows don't lie unless someone under or over reports the cash flow).

Joking.


Just noting the Pandora's box you open on DU when (correctly) noting that the surveys are, in fact, surveys.

I think I will start referring to the Labor Department as a right-wing pollster.

With one survey showing job losses of 20,000 and another survey showing employment increasing, is that it is at the end of the day a survey subject to a margin of error and also subject to the fact that the way the survey calculates data, over the past 50 years, has been set up to show the best possible picture and every administration since it was created has been tweaking it to make themselves look awesome. The weekly report, hasn't been tweaked as much, because the media doesn't pay as much attention to it as the monthly report. Don't get me wrong though, it has been seriously played with as well.

When I read, we have revised our model, added job losses to the period of time in the last administration, and hey look despite all other data points reporting something to the contrary, this survey has different data, my bullshit meter goes to from 50 (it's normal setting) to 100.

If I'm still around next month I don't want to have this debate again over what the BLS is.