A few weeks ago I heard on the morning news that ~16,000 jobs were lost in Alberta last month. Now, mine was one of them, but unless one of you guys told the government, I don’t see how they would know. I didn’t file for EI, didn’t register with anywhere, and even my bank probably wouldn’t be able to tell until the end of the week when I deposit a cheque instead of get direct deposit.
Are they looking at EI? Does -1 old job +1 new job actually equal no lost jobs for statistical purposes? Or am I just not worthy enough to even be a statistic?
In Canada, the Labour Force Survey is done once a month, which is how unemplyment figures for Canada are obtainded. The US does a similar survey (the Current Population Survey) once a month as well.
The Current Population Survey (CPS) that Iamametalrobot mentions is how it’s done in the US.
The CPS is conducted by the Census Bureau for the Bureau of Labor Statistics (BLS). The CPS is a household based survey and a household selected to participate in the survey will participate for four months, take eight months off, and then participate for four more months. Each month, approximately 50,000 households around the US are selected to participate. I don’t remember the sampling scheme off the top of my head, but they create some sort of representative sample.
The CPS asks a LOT of questions about work. Some sample questions (in no particular order) are:
-Did you work last week?
-If not, why not?
-How many hours did you work last week?
-How many hours do you usually work?
-If you are not working, why did you leave your last job?
-If you were laid off, do you expect to be called back?
-If you are not working, are you looking for work?
-If you are looking for work, what methods did you use during the past week to find work?
-How much do you earn? (there’s a whole series of irritatingly complicated earnings variables)
-What’s your occupation?
-What industry do you work in?
There are also plenty of data about household composition and demographic characteristics. From all this, you can get a very detailed picture of the state of the labor force. The data are publically available for anyone who cares to use them.
How would the state know that the people no longer being paid are unemployed - not fired, or people who moved out of state, or someone on a sabbatical, or someone who died?
“Unemployed” is a pretty specific term. It does not mean simply “no longer collecting a paycheck from the place I collected a paycheck from last week.”
Indeed. The unemployment rate is the percent of the non-institutionalized adult population that is not currently employed BUT is actively seeking employment. If you don’t want a job, you’re not unemployed. If you’d like a job but don’t actually look for one, you’re still not unemployed. If you’re in the military or in prison or in a mental institution, you don’t fit anywhere in the equation.
My company has been selected by theU.S. Bureau of Labor and Statistics. Monthly, I have to fill out a form with the number of employees that worked in the payroll period that covered the 12th of the month, how many of the employees were women, bonus and/or commissions if any, total gross payroll for the period and the total payroll for the previous month.
Been doing this for 5 years now although I’ve owned the business for almost 17 years. Pretty sure Canada has a similar situation…having the employer filling out many state and federal documents to get an idea for labor and tax stats.
If I read this table right (and I’ve only been looking around for 15 minutes, so who knows), all the provincial data is extrapolated from only 2,818 people, which is a much smaller sample size than I’d have expected.
Actually, no. The 2818 is the total population, in thousands. (So the total population of people in Alberta who are 15 and older is 2.8 million.) See the pdf of the same table - it specifies the units. Although, putting populations in thousands is sort of standard practice for this sort of table.
If you look at the full report, there is info on data collection at the end. Similar to the CPS, the Labor Force Survey samples 53,000 households per month. In scanning it quickly, though, I don’t see an estimate of how many of those households are in Alberta.
I don’t think I can do a good job of explaining how standard error and all that works, but in general, a sample of about 1000 will get pretty good estimates. If you look at opinion polls, you’ll notice that many of them have a sample around that size. The CPS and similar labor force surveys have gigantic sample sizes and are extremely accurate.
With the unemployment rate in particular, there’s plenty to quibble about in how it’s defined, but probably not in how the nuts and bolts of how the statistics work.