Since unemployment data is based off of surveys, I assume the announced data reflects some period in the past.

I cannot find information on the "currency metric" of unemployment data.

Does anyone have any ideas?

In case I'm falling for the "XY Problem", here's my initial conundrum:
Imagine I have a real-time unemployment model. If for some week my model claims unemployment is relatively high. At the end of the week the monthly unemployment rate will be announced. Should I bet that the announced monthly unemployment rate will be high? If the data was collected from surveys taken a week ago, it would not be. I need to know when the data was collected.

  • 1
    What is the scope of your request? Which areas of the world are you seeking unemployment survey research/administrative data from?
    – Kotebiya
    Apr 9 '15 at 20:45
  • @Kotebiya I would like to know this for US labor statistics released by the BLS, but there may be standards set by the ILO, since "it is used internationally so comparisons can be made between countries". Having a standard on the timeliness of the data would be necessary to make comparisons I would think. Apr 9 '15 at 21:02

There are many sources of surveys/administrative in the United States data you might find useful.

  1. Unemployment Insurance Claims - the number of people in a given week who: newly file for unemployment insurance, are continuing their unemployment insurance. This would be the most timely data.
  2. Current Population Survey (CPS) - (reference period, data source) This is the monthly survey conducted since 1940 that is used to take an assessment of the country's labor force status. Whenever you hear about the "official" unemployment rate, it is usually coming from this survey which are published "Employment Situation" reports such as the ones here. This is a fairly well-organized source of information since the response rate is usually 93% and above and has less statistical noise since in any given month ~72% of the sample is carried over from the previous month and ~45% is carried over from the previous year.
  3. Current Employment Statistics (CES) survey - (reference period, data) This is the monthly survey that polls random companies about their employment. Be warned, however, its timeliness is questionable since it takes up to three months for the average response rate to go above 93% ( usually in the first month, its response rate is ~60% on average).
  4. Quarterly Census of Employment and Wages (QCEW) - (Data: source 1, source 2) - This is a far less timely source of data. However, it is a publicly released report of all payroll employment within a given quarter. Historically, it takes on average nine months for the data to release so by the time it is ready you are typically looking at data from a year ago.
  5. Local Area Unemployment Statistics (LAUS) - This source of information is different from all the rest since it uses small area estimation to predict state and local unemployment rates. It uses the CPS as its primary source, along with other data such as UI claims or the CES survey act as secondary sources. This source is also less timely for your purposes.
  • 1
    Gallup also does a 30 day rolling average of employment statistics at gallup.com/poll/125639/Gallup-Daily-Workforce.aspx. Their methodology is different than the Bureau of Labor statistics, based on telephone surveys of a large sample (hence the 6.5% unemployment rate versus 5.5% from BLS) but if you're looking at trends, it should move in the same direction.
    – UpQuark
    Apr 10 '15 at 16:08
  • 1
    The Basic Monthly version of the CPS is conducted by the Census Bureau. However, the Bureau of Labor Statistics (BLS) and the Census Bureau work in heavy collaboration together as the link briefly mentions.
    – Kotebiya
    Apr 10 '15 at 17:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.