What the BLS survey modification will and will not do.

You may have heard that the Bureau of Labor Statistics (BLS) has modified its survey of unemployment.  There is probably going to be a good deal of confusion over what’s being changed, so let me summarize the situation.

  • Official unemployment numbers are derived from the Current Population Survey (CPS), which surveys American households every month in order to gather various statistical data.  The potential confusion lies in that the CPS isn’t uniform in how it defines unemployment; depending on the question, somebody may or may not be actually considered to be in the labor market.
  • So the CPS will (over the next four months) start including people who have been out of work  for between two and five years in their calculation of median length of unemployment, which the BLS pretty explicitly thinks is being under-reported.  Previously, the cutoff date was only two years; anybody out of work for longer than that would be considered effectively not part of the work force for the purposes of determining this specific statistic.
  • However, the CPS will not change the BLS definition of ‘unemployed‘ (no job, actively looking for work in the last month, ready to work) for the purpose of their most commonly reported-on statistic (the U-3, which is currently 9.8%).  As Ed Brayton – no friend to the Right – notes, this means that the currently reported unemployment rate numbers will not change because of this policy.
  • Take up any contradiction in the assumptions behind calculating median unemployment length and calculating the current unemployment rate with the BLS.

So the real story here is not that the BLS is on the verge of reporting that the unemployment numbers are going to go through the roof.  They’re not, and that’s because of a conscious and continuing decision by the government to use the U-3 numbers (9.8%) instead of the U-6 ones (17%) (permalink): there’s no point in fudging the unemployment numbers downward with one hand while raising them with the other[*].  No, the real story is that the BLS is quietly worrying that the long-term prospects for reducing unemployment are looking pretty grim.

Moe Lane (crosspost)

[*To clarify, because this came up in comments: this was only meant to implicitly push back against the notion that the government was trying to game the system by changing the criteria for the length-out-of-work question.  They’re already (in my opinion) gaming the system the other way by using U-3 numbers to reflect ‘official’ unemployment when they should be using the U-6 ones.]