Tuesday, October 23, 2007

How useful is the HLFS?

The Household Labour Force Survey is the source for our official unemployment rate. Sixteen thousand households are surveyed quarterly.

Being 'employed' is defined by working 1 or more hours in a week.

Employed: All persons in the working-age population who during the reference week worked for one hour or more for pay or profit in the context of an employee/employer relationship or self-employment; or worked without pay for one hour or more in work which contributed directly to the operation of a farm, business or professional practice owned or operated by a relative; or had a job but were not at work due to: own illness or injury, personal or family responsibilities, bad weather or mechanical breakdown, direct involvement in an industrial dispute, or leave or holiday.

In September 2006 the HLFS data showed that 43 percent of one parent with dependent children only households had no-one employed.

Yet my OIA data shows 107,000 single parents with dependent children were on some form of benefit. According to the Census data from only six months earlier there were 145,000 such families.

Allowing for very slight variation due to information gathered in different months, this provides an unemployment rate of 74 percent. Of course some on benefits are also working part-time but typically only around 1 in 5. That would drop the rate to about 59 percent. Still well above 43 percent.

Which probably shows the nonsense of using one or more hours paid employment as the official definition for being 'employed'.

The HLFS is useful - for those who would paint a rosier picture than is really the case.

9 comments:

Anonymous said...

That 1+ hour per week definition comes from the ILO. Not that it refutes your point, just saying.

Anonymous said...

I’d say it probably shows the nonsense of trying to cobble together rates with insufficient and/or incompatible data and making conclusions based on flawed logic.

The census count you’ve cited is for families with (any) dependent children; the HLFS percentage you’ve cited is from a subset of households with dependent children (those with dependent children only – ie, not including those also living with adult children and/or other people).

Lindsay Mitchell said...

Anon,

The problem with the other subsets you mention eg one parent with dependent and adult children, or one parent with adult child(ren) only, is the adult child is more likely to be employed than the parent. While the mother remains on benefit and 'unemployed' the household is no longer categorised as 'none employed'.

So the HLFS 'one parent with dependent children only' is the rate commonly used to cite the unemployment rate of single parents.

I maintain that it is misleading because it is too low.

But I'm open to a better estimate. What do you propose?

Anonymous said...

What is misleading is your conflation of “on benefit” and “not employed” with “unemployed”, when they are not the same thing. The unemployment rate for sole parents is derived in the same way as for other people: the percentage of the labour force unemployed. If you want to derive the unemployment rate of sole parents with dependent children counted at the 2006 census, you need a 2006 census count of the number of sole parents with dependent children who were unemployed and the number of sole parents with dependent children in the labour force (employed and unemployed). It should be possible to get such data from Statistics New Zealand (labour force status of sole parents with dependent children). I see they are releasing census work data on Table Builder at the end of this month. Perhaps we’ll be able to find the necessary data there.

It should also be possible to check with Statistics New Zealand how many employed people in the labour force survey work really short hours (my guess is very few). Your criticism of the survey’s definition of employed – one or more hours – could then be assessed against evidence rather than supposition.

For the other statistic you estimated - proportion of sole parents receiving a benefit – you should be using a March 2006 benefit figure (not September), and compare it at least with 2001 and preferably earlier censuses. On its own it doesn’t tell us much, certainly not the unemployment rate of sole parents.

Lindsay Mitchell said...

If the "unemployment rate for sole parents is derived in the same way as for other people" then it must be derived from the HLFS. So what is it?

The only table pertaining to sole parents is employment status by household. As you have already pointed out there are five subsets of sole parents each with a different 'none-employed' rate. As well as the problem of not knowing who in the household is employed I suspect that some of the groups are small enough to present some sampling error.

Yes, I should use March 2006 benefit figures but they are not available from the published factsheets.

For instance the total number on the DPB is not the same as the number on DPB Sole Parent and the factsheet doesn't provide a breakdown.

The number of sole parents on the DPB does not include the number of sole parents on all benefits. Neither is that information routinely released.

Therefore I was relying on a response to a OIA question I had to ask to access this information. It pertained to September rather than March. There isn't huge variation over a few months (except during December/January) and any there is wouldn't change the percentage by more than 1 or 2 points. No, it's not a scientific calculation but I haven't claimed it is.

I think it is useful for the public to understand how the official unemployment rate is calculated (via a survey) and that is has no connection with any unemployment rate calculated by numbers on benefits. Hence only 3.7 percent of the workforce are officially unemployed yet 10 percent of working age people are on a benefit.

I think our discussion has shown that technicalities tend to obscure more than reveal the true picture. But I am happy to be held to account. Readers can then make their own minds up.

Anonymous said...

You wrote: If the "unemployment rate for sole parents is derived in the same way as for other people" then it must be derived from the HLFS.

Your conclusion doesn’t follow (non sequitur). The Household Labour Force Survey has only been going since 1986. For decades before that, the unemployment rate was derived from the five-yearly census. The census is still used to calculate unemployment rates for those groups that aren’t well represented in the survey (eg, people in small communities, as in SNZ’s Community Profiles).

It’s not always hard to find information. I just googled “labour force status of sole parents” and came up with this table of 2001 census data: “Work and labour force status and sex of sole parent by age group of youngest dependent child in family, for one-parent with dependent children families in private occupied dwellings, 2001”.
http://www.stats.govt.nz/NR/rdonlyres/DF8562BB-1FBF-49B7-BB2B-2A8CABCCDE49/0/Table30.xls

Here’s how to calculate the unemployment rate of sole parents in 2001 from this census table. Add the number employed full-time, the number employed part-time and the number unemployed. That’s the labour force (84,000). Divide the number unemployed (15,800) by the labour force and multiply the result by 100. I get 19 percent. With a similar table for 2006 (or even a simple count of sole parents by labour force status, without the child age breakdown) you could see whether the unemployment rate of sole parents changed in the five years to 2006. I would guess that it's still relatively high but has gone down somewhat, along with the overall trend.

Lindsay Mitchell said...

So in 2001 the unemployment rate (using the technically correct definition) of sole parents was 19 percent. Fine.

And in 2001 there were 140,000 single parents with dependent children and 116,000 on benefits. Or 83 percent.

One presumes they are on benefits because they don't have jobs.

I think we have come full circle.

Anonymous said...

I agree there is something circular about this discussion. You ended your post with these words:

"The HLFS is useful - for those who would paint a rosier picture than is really the case."

In 2004, you were using that HLFS statistic to show that things were getting worse: the proportion of one parent households with no-one employed had reached 52%.

http://www.scoop.co.nz/stories/PO0408/S00102.htm

Now that it's fallen, to 43% in Sept 2006 (and to 39% in June 2007), you're finding fault with the survey and insinuating that people use it to paint a biased picture...

Even your own estimates (83% in 2001, 74% in 2006), show that there has been a substantial drop in reliance on benefit income. Elephant pate, anyone?

"One presumes they are on benefits because they don't have jobs."

No, they are on benefits because they have responsibility for dependent children and have insufficient income from other sources, such as a spouse. Many of them do have jobs. Looking at that census table, you can't get anywhere near 74% without adding in the part-time workers.

I'm getting dizzy, so I'll stop there.

Lindsay Mitchell said...

Your last comment is a fair criticism.

3 years ago I was taking the HLFS at face value when I should have been questioning the anomaly between the information it provided and the information available about benefit dependency.

I accept there has been a drop in the number of single parents relying on a benefit but some have moved sideways onto In Work tax credits. Others have moved to other benefits and are not identified in published fact sheets.

Most importantly the reduction is not happening among the youngest. The ongoing problem of thousands of young people starting on welfare and staying there long term is not improving. The overall drop is mainly due to the most able taking advantage of a strong work market.

And if you check I did take account of part-time workers in my original post.