Friday, May 11, 2012

Tim Colebatch has cracked it. Why the jobs data is flawed

Tim Colebatch in today's Age:


THE JOBS STATS: CAN WE BELIEVE THIS?

Year to Sept 10, Year to Sept 11

Population growth: +325,000, +320,000
Growth in adult population* +394,000, +224,000
Growth in jobs: +366,000, +111,000

Source: Bureau of Statistics. * Adult population estimate used in jobs data.


The official jobs figures published by the Bureau of Statistics have significantly underestimated recent job growth, due to forecasting errors that first overstated, then understated, the growth in the adult population.

The errors, which have serious implications for economic policy, began when the number of foreign students living in Australia fell rapidly after immigration laws were tightened in late 2009.

The unforeseen fall at first led Bureau forecasters to greatly overstate population growth — and When it realised the error, rather than correct it by revising the previous jobs figures the Bureau decided to understate population growth in future forecasts, depressing the labour force figures. These then reported a net loss of 900 jobs in 2011.

On one estimate, once the figures are adjusted for the erroneous forecasts, at least 100,000 of the jobs supposedly created in 2010 in fact arrived in 2011.

The errors are not in the official estimates of population growth, which are issued six months after the period to which they apply. They are in the estimates — in effect, forecasts — of the adult civilian population used in the labour force figures.

Usually the two series move together. But in the year to September 2010, population growth (including children) shrank rapidly, from 433,000 to 325,000, whereas the forecasts for the labour force estimated that adult population growth would remain steady at 394,000.

In the year to September 2011, that suddenly reversed. Actual population growth was little changed at 320,000, but the Bureau slashed the forecasts used in the labour force figures from 394,000 to 224,000.

Since most people interviewed in the labour force survey are employed, the effect of understating population growth was to understate employment growth.

The Bureau defended itself yesterday in an article published with the labour force figures, arguing that its main focus is on getting a correct reading of the unemployment rate and workforce participation rate — which come straight from the survey data.

But its approach seriously misled readers, commentators and ultimately the public, about the size of the slowdown in the jobs market — and hence, the true state of the economy.

One prominent commentator seized on the reported fall in jobs to describe the labour market as being in its worst shape since 1992....

The Australian Statistician, Brian Pink, yesterday stood by the Bureau’s figures. “We do not believe that the employment growth that we have shown has been biased in some way by the method - that’s our view,” he said.

Senior economic officials are aware that the data is flawed, but have refrained from making any public statement, so as not to reduce confidence in the Bureau.

But tax data released with the Victorian and Federal budgets confirm that the jobs markets in 2011-12 has been stronger than the official figures show.

The Bureau estimates that jobs in Victoria fell by almost 20,000 in the first nine months of 2011-12, with 38,000 full-time jobs lost. Yet the state’s payroll tax revenue rose 7.7 per cent in that time, with no fall in jobs.

Tuesday’s Federal budget showed PAYE income tax revenues up 9.7 per cent in 2011-12, faster than the 9.4 per cent growth in 2010-11. While officials believe job growth has weakened in recent months, the tax take is strong evidence that the Bureau’s estimates are wrong.

Westpac senior economist Justin Smirk said the bank’s economics team is uncomfortable with the way the Bureau has tackled its problem.

‘‘It does raise concerns about the accuracy of the data, and we still have questions about the actual employment levels and their growth path’’, he said.

Published in today's Sydney Morning Herald and Age







Related Posts

. A sudden pick up in employment?

. Employment graphs. The zigs are getting bigger.

. Bleak Christmas? For job seekers it was the worst in 20 years (or was it)



6202.0

61 comments:

Anonymous said...

Read the article in their publication Peter it explains why you mob in the media are barking up the wrong tree

Susan said...

Hello Peter,

"Growth in jobs: 366,000, 11,0001"

For the record, 11,0001 is not a number. At least not one, I'm familiar with ...

kymbos said...

Peter, this is a little unclear. Are you saying the previous employment estimates were biased downwards, but yesterday's figure is likely more accurate than what we have seen over the past year?

And how has Tim Colebatch 'cracked it' if the ABS has released an explanatory article addressing the issue? Or did they do that in response to Tim's article?

Peter Martin said...

Dear A, The article only tangentially touches on what Tim has discovered. It doesn't address his findings - for the obvious reason that they were not published until this morning.

Dear Susan, Corrected. Thanks.

Dear Kybos, I do not know whether yesterday's employment growth figures was accurate.

I do know that it looks as if the ABS knowingly suppressed reported employment growth in 2011 to make up for mistakenly inflated employment growth in 2010.

That's the story. The release of the ABS paper (which only tangentially touches on what Tim has discovered) was a coincidence, as far a is know.

Anonymous said...

Hey Pete As far as I can tell there is little the ABS can do to improve this. There trying to predict what someone will do before they even arrive.
How would you like the ABS to improve this...

Susan said...

I note the problems with the labour force figures you raise, Peter. I personally would not be inclined to draw a straight line between them and the PAYG figures in the Budget. Any apparent inconsistency between them can be explained by wages growth since - as I'm sure you are aware - PAYG depends on growth in both wages and employment.

Peter Martin said...

Dear A,

Glad you asked.

"How would you like the ABS to improve this?"

At the moment when the ABS discovers its population estimates have been wrong (producing wrong jobs growth figures) it DOES NOT revise those job growth figures. It leaves the historical record wrong and overadjusts future figures to compensate.

So when late in 2010 the ABS discovered its population growth projections had been too high biasing up its jobs growth figures during 2010, INSTEAD OF REVISING THE 2010 JOBS FIGURES DOWN it pushed down its projections of population growth in 2011 to compensate, resulting in lower than accurate jobs growth figures in 2011.

What should the ABS do instead?

It should revise old figures when new information indicates they are wrong rather than making new figures wrong in order to compensate.

It is really important that it does.

Peter Martin said...

Hi Susan, Interestingly the finance department figures show PAYG growth holding up during the first nine months of 2011-12.

Bruce Bradbury said...

The ABS is right to focus primarily on getting the unemployment and employment/pop ratios correct (rather than the numbers of people in jobs).

This is what the Labour Force Survey is best at estimating (as explained in the ABS publication). It is also what is most relevant for both household living standards (what is the chance that someone will have job) and for macro-economic management (what is the chance that employers will be able to find extra workers).

The lesson to take from this is that the rates are both better estimated than the numbers _and_ they are more useful measures of how the economy is performing.

The challenge now is to get politicians and commentators to understand this.

Peter Martin said...

Bruce, the challenge is to get the ABS to acknowledge this.

It gives the monthly change in employment pride of place on page 1 of each labour force release.

It doesn't include the employment-to- -population ratio at all.

Susan said...

Hi Peter, you are right about the MFS. But isn't that revenue rather than receipts? I guess it doesn't matter all that much for PAYG since the wedge isn't really big. The first 9 months of the year would not - in any case - reflect the April jobs numbers (which, like you, I don't believe). The problem with the labour force figures other than their volatility is the fact that they are subject to revision and sometimes, significant revision. Over time, you would expect them to reflect other statistics like tax revenue but, there is often a lag before this happens. In the face of everything we know about the non-mining economy, I would say that the April jobs number is the outlier and that, in view of that, employment outcomes are likely to be headed in the direction outlined in the most recent Statement 2.

Anonymous said...

Peter, the argument that jobs growth has been very much stronger than previously thought must be reconciled with a raft of weak indicators over the same period.

If jobs growth was solid rather than stagnant, somebody forgot to tell the non-mining economy to act accordingly.

The Lorax said...

But GDP growth was stronger in 2010 than 2011.

As Anon says, it doesn't make a lot of sense in light of other indicators which were all weaker in 2011.

The credibility of the employment survey is shot to pieces. What with the huge MoE, the big revisions, and now this, its become almost meaningless.

BTW, Roy Morgan's survey has diverged markedly from the ABS survey in recent months. Morgan has unemployment in the nines, and trending up. The ABS has unemployment in the fours and trending down.

Bruce Bradbury said...

Peter, yes, you are right. The ABS should not be giving such prominence to the numbers in their summary release (as opposed to the rates). I also agree that they should be less shy about doing revisions.

The Lorax said...

Morgan says only 30 percent believe ABS survey, while 60 percent believe the Morgan survey.

Morgan now doing surveys to support its surveys!

Marek said...

Now who to trust - Morgan or the ABS? ;)

The Lorax said...

Morgan is more consistent with other data. Perhaps the ABS is right, and everything else -- job ads, inflation, GDP, consumer sentiment, business confidence, PMI, PSI -- is wrong?

Peter Martin said...

Dear Anon and Lorax,

"The argument that jobs growth has been very much stronger than previously thought must be reconciled with a raft of weak indicators over the same period."

Like this: Things weakened during 2011 but jobs growth was very much stronger than flat.

The Lorax said...

Are you buying the Victorian numbers Pete?

There are so many reasons not to believe jobs growth is surging in Victoria: Manufacturing is in recession, the AIG's PSI for April was an absolute shocker, the Melbourne apartment construction bubble has burst.

Colebatch has been banging on about the weakness in Victoria all year, but now we're supposed to believe the ABS survey was dodgy, but now its fixed and Victorian employment growth is going gangbusters.

Sorry, not buying it.

Peter Martin said...

I don't believe it Lorax. The margin of error for one month's employment growth is one state is massive.

Anonymous said...

Wow pete you and Tim have stirred up a hornets nest. I know a guy who works at the ABS and he told me that theyre ticked off at u guys. He was telling me you guys didn't consult with them about Tims article. Also he said they have been looking at this for some time now and their analysis shows it doesn't really change the pattern of growth. They also have a very good idea who your source is....treasury guy who doesnt work in the domestic price section any more. Can't wait for the reply...lol

Anonymous said...

Have to say Pete you guys have really over played your hand on this one. Get ready to wipe a sticky eggy mess off your face. Maybe you should have made a phone call to the abs before you started making sensational claims...no thats right you treasury guys (or ex treasury) think you know it all. Cant wait for the reply from abs

Anonymous said...

"Like this: Things weakened during 2011 but jobs growth was very much stronger than flat."

How much stronger in real terms Peter?

I have a feeling that Tim has jumped to a very big conclusion here. If the the labour market was reasonabley robust through 2011, logically we would expect to have seen some strength across a broad range of indicators. Instead we saw GDP growth trending down, historically weak credit growth (with personal credit falling below zero), a horror year for the housing market, retail struggling to rise above pre-GFC levels, rising business insolvencies, poor performance in manufacturing and tourism...and more. It wasn't a recession but was well short of an economy in robust health.

If employment really was solid in 2011, it would be almost like suggesting that strong jobs growth was the core reality while all other results across a broad range of indicators were outliers. This seems highly doubtful at best.

Tim's observation is interesting and certainly deserves investigation but I feel that he has allowed the belief that he is onto something big to cloud his judgement somewhat.

2011 was a rather lacklustre year for the non-mining economy - Tim needs to now demonstrate that this was not the case and that it was in fact reasonabley robust if his argument is to hold water

Anonymous said...

I really doubt that this article actually holds any weight. The reason being is that the numbers on a qrter to qrter basis are very small e.g. A qrter where the pop difference between erp and lfs pop is 10k, 6 k of this is employed, but this is spread over the qrter. This means that while there may be an "under estimation " in 2011 the numbers are very very small. Plus this really is a bit of an old story. I know that several banks have had conversations with the abs about this very thing and in the end theyve concluded that while there may be a small over estimation 2010 followed by an equally small under estimation in 2011 the pattern is the same i.e. strong 2010 followed by weak 2011. Question now is if the abs demonstrates this will Colebatch and Pete come out and admit they were wrong. I doubt it

Bruce Bradbury said...

The ABS estimate of the employment to population ratio is not really affected by the population revisions. This shows a growth of 0.7 percentage points for the year to April 2011 and -0.3 percentage points for the year to April 2012. (Grogs Gamut has this nicely plotted up).

Irrespective of whether you want to call -0.3 'flat' or 'weak' it does support the view that 2011 was weaker than 2010.

Anonymous said...

Nicely put Bruce. But say you follow Pete and Tim's preoccupation with the level estimates and apply the e to p to ERP to generate level estimates with the right population figure...does it look any different?

Anonymous said...

Just did some calculations pete. Looks as if there is no substantial difference. I went back and recalculated the labour force population using erp as a guide. Then recalculated then level estimates using the e to p ratio. While there are differences the pattern is the same i.e. very strong 2010 weak flat 2011.

I think you guys have stuffed up!

Anonymous said...

Playing with words here Pete.
"Things weakened during 2011 but jobs growth was very much stronger than flat."
So was it weak or not? From what you and Tim are saying I get the impression you are saying it wasn't weak.I think you will find that even after the figures are revised the figures will still be WEAK...not stronger than flat (whatever that means) but WEAK.
Have alook at the data the first time they revised
http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/6202.0Jul%202010?OpenDocument

Story stayed the same figures changed a little. And when they revised this time the difference was bigger over a longer period.

Anonymous said...

Correction*when they revised in 2010 the difference was larger and over a longer period then now. so the differences will be smaller now

Anonymous said...

Hi Pete,

Couple of points, some have already been made. As Susan and others have pointed out employment and PAYE revenues aren't always correlated for a few reasons.
One employment doesn't take into account wage growth.
Two the LFS doesn't measure the number of jobs. This second point is important especially in recent times with the growth in PT employment. The growth in PT employment may mean that more people have more than one job and thus pay more tax without employment changing.
I also agree with others about other indicators pointing to a weaker 2011. Unless Tim has more evidence I would suggest this is a bit of a red herring and the changes aren't as big as he is suggesting...if they are then this might change my mind but i doubt they are. Maybe you could post the stuff that he has on this so we can see the size of the change that he's talking about?

Peter Martin said...

"Maybe you could post the stuff that he has on this so we can see the size of the change that he's talking about"

Tim Colebatch, above:

"On one estimate, once the figures are adjusted for the erroneous forecasts, at least 100,000 of the jobs supposedly created in 2010 in fact arrived in 2011."

Anonymous said...

Details would be nice Pete. was it all one month or was it spread...how was it put together?

Scott Wallace said...

I think you guys seem to have got it in reverse. I thought if 2010 population was over forecast it means that the actual unemployment rate was lower and if they decided to lower the population number in 2011 then the actual unemployment rate was higher or real employment growth was slower or negative.

Anonymous said...

Hi Scott

Two things

1st the ABS did not delibrately push population growth down. The growth rate was reduced due to the way they derive the figures (read the article it explains it). They use this method so you dont get SUDDEN jumps or drops in the figures. Pete and Tim have written this article probably not knowing how its done and therefore unknowingly misled readers into believeing the ABS manually intervenes with growth rates.
2nd employment is correlated with population not unemployment. So the most noticeable changes will be for employment not unemployment.

Scott Wallace said...

Thanks Anonymous I was not careful with my expression how I understood what Pete said was that they lower the forecast population growth number to adjust the total population data closer to actual number.

Yes, sorry what I meant was employment growth in 2010 was faster and in 2011 was slower. Now when I think about unemployment I am confused and I think I know why Tim and Pete got it mixed up. When you use the number of unemployed people divided by a smaller population number (2010)that means unemployment was higher but if you use the total number employment created divided by a smaller population that means employment growth was faster. I think that's it.

Scott Wallace said...

Too early in the morning and not enough sleep, my expression still is not clear. But anyway if we recalculate 2010 number with a smaller population size, I don't know whether we need to readjust the participation rate as well. I'm more confused.

Peter Martin said...

Dear A,

Than you for your quality contribution.

You say:

"the ABS did not delibrately push population growth down. The growth rate was reduced due to the way they derive the figures (read the article it explains it). They use this method so you dont get SUDDEN jumps or drops in the figures.

Indeed. A much better approach would be NOT to adjust future growth rates to correct for past mistakes, but to revise the previously published figures to CORRECT the past mistakes (more than once every five years)

That way the present jobs growth figures would genuinely be the Bureau's best guesses, rather than being biased up or down to compensate for previous wrong guesses.


Dear Scott, The unemployment rate has nothing to do with the size of Australia's population. Nothing at all. It is the outcome of a sample survey (with the caveat that the population survey is used to weight individual respondents to the sample survey).

The accuracy of the Bureau's estimate of the unemployment rate is not in dispute.

Anonymous said...

Hi Pete
Yeah it would be nice if they revised more, but really not necessary given other measures already provide an accurate measure that are relatively stable (i.e. E to P is stable/ no need to revise). Plus where would the money come from? Like other departments I'd imagine the ABS has had an increased "efficiency" dividend applied.
So from what I can tell most ppl agree that the EtoP is a reliable indicator that is not impacted by problems with population projections..I think you agree on this point? So instead of working with the ABS you and Tim have successfully eroded confidence in the ABS unneccesarily...well done. Next time there is a cut in the funding to the ABS give yourself a pat on the back.

Peter Martin said...

"most ppl agree that the EtoP is a reliable indicator"

Most people don't even know what the EtoP is. The ABS doesn't even include it in its publication.

Most people do know what employment growth is. It is a measure the ABS chooses to highlight, each month.

Did you really mean to suggest Tim and I ought to keep quiet about its deficiencies in order to maintain confidence in the ABS?

Are there other things you like us to keep quiet about?

Perhaps electoral irregularities so as to maintain confidence in the electoral system, budget irregularities so as to maintain confidence in the budget process...

Anonymous said...

No Pete you have agreed that EtoP is a reliable indicator.
Furthermore, the case you and Tim have mounted is grossly exaggerated and I think you know it. Its absolutely ridiculous for you to compare this to electoral irregularities...please be serious! Also your wrong, the E to P is published, yes it isn't given prominence which is a mistake but that isnt the argument. The argument is that you and Tim believe somehow the figures have been dramatically altered because of problems with population projections...which is false and again I think you know your wrong. When the ABS comes out and shows Tims claim of 100'000 jobs being transfered to one year to another as wrong will you come out and admit your wrong or will you move the goal posts and weasel out of your argument...I wait with interest

Anonymous said...

Further to my previous point take alook at some more considered opinion than your own. Still convinced that your onto a big scoop?
http://ricardianambivalence.com/2012/05/14/its-the-unemployment-rate/

Peter Martin said...

Thanks. I read the RA post. It is fine, but it doesn't address the point Tim is making - that instead of revising employment numbers when new population information shows them to be wrong the Bureau usually adjusts future population growth estimates to compensate. This has the effect of suppressing (or inflating) actual employment growth.

Anonymous said...

No it doesnt Pete. Look at the revisions they did in July 2010. Do sum sums Pete, take the eto p ratio and apply it to the erp...you will notice it doesnt change significantly...I think you and Tim need to acknowledge that.

Peter Martin said...

As I read Colebatch he is talking about the year to September 2011.

Anonymous said...

For comparison...you guys didn't look at the July 2010 revision as a guide to the extent of revisions before you wrote these articles? Good investigative journalism!

Anonymous said...

Think you missed a zero on the top figures, should be 110,000 for year to Sept 2011 employment growth.

Peter Martin said...

Fixed. Thanks.

Anonymous said...

The fact that the last post had to point out you got the 2011 figures wrong makes me wonder how much care was taken to get facts right and whether you and Tim did much prep work before publishing.
Did you even speak to the ABS before publishing? Might have helped?

Peter Martin said...

That was a simple cut and paste error by me.

No blame should attach to Tim. What he wrote is in The Age error-free.

Much as I would love to discuss who Tim spoke to I can't, for the usual reason. Unfortunately.

But what's more important is what he has reported. Published employment growth in 2011 was lower than reality and has not yet been corrected.

Anonymous said...

Peter, I think employment was revised by the ABS when they revised the population figures. See their comment below (and look at their chart):

If we now examine the impact on the employment level estimates we see that, like the Labour force civilian population, the differences appear to be quite large. The graph below shows the largest revision was 156 thousand for September 2009. This does not mean that there were 156 thousand more people employed in September than was first estimated, it means in broad terms the weight assigned to each individual in sample in September was much more after the revisions were carried out.

So why do you say that the history of employment isn't revised?

Peter Martin said...

Why do I say:

instead of revising employment numbers when new population information shows them to be wrong the Bureau usually adjusts future population growth estimates to compensate.

Because the Bureau normally only revises past employment numbers for mistakes in population projections once every five years.

Here's how the Bureau put it last week:

"Normally the Labour Force population benchmarks are fixed once they are forecast. Benchmarks are revised once every five years, after data from the Census of Population and Housing has been calculated and rebasing of ERP has been finalised. However, the benchmarks were revised in July 2010 after significant revisions were made to NOM in ERP. These revisions included all data from July 2006 to June 2010.

Anonymous said...

Have to say Pete your readers are onto something with the E to P ratio.
Took their advice and pulled out the data and recalculated the estimates using the E to P ratio. I was thinking surely Pete did his own checks first to check the veracity of the claims put forward by a source. Turns out you mustn’t have.
First I checked the previous revisions as a guide to see if the method works. As a rough guide it does. As you have claimed the problem with what the ABS has done is that it has suppressed growth. In your own words:
“But what's more important is what he has reported. Published employment growth in 2011 was lower than reality and has not yet been corrected.”
So I used Growth from year to year as my guide to see if the method worked. It works well and confirms that the pattern of employment growth does not change. Below it shows that the revised figures both the real and E to P ones show the same pattern as the original figures, that is that employment in 2009 was weak and strengthened in 2010.

Anonymous said...

part 2
Example
Growth to June 2009 before revisions: -28.5
Growth to June 2009 after revisions: 25.8
Growth to June 2009 using revised population figure and E to P: 20.5
Growth to June 2010 before revisions: 353.2
Growth to June 2010 after revisions: 311.0
Growth to June 2010 using revised population figure and E to P: 318.7
You can find the data that I used on the ABS website free for any diligent journalist to download and do his research. It’s the article_July2010 excel spreadsheet. The E to P ratios are published the month before as unlike your post and Tim’s article suggest the ABS does revise, so the old data is in the month before.
http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/6202.0Jul%202010?OpenDocument
http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/6202.0Jun%202010?OpenDocument

Anonymous said...

3rd and final
So now I applied this to the current figures (Using ERP minus 0-14) and guess what same thing. Strong 2010, weak 2011. No 100,000 change in 2011, no radical change in the pattern of employment growth. Here it is:
Unrevised: 3.33% growth in the year to September 2010 or an increase of 365,500
Revised using E to P ratios: 2.79% or 305,910
Then in the year to September 2011
Unrevised: 0.99% or 111,850
Revised using E to P ratios: 1.33% or 149,430
That’s a revision of 0.5% in 2010 and 0.3% in 2011. Pictures the same as it was before, no understatement of reality as you and Tim claim. Furthermore, slap some standard errors around the estimates and guess what, the revisions look minuscule. Bottom line your wrong!
So you would like the ABS to do regular revisions to gain a measly 0.3 to 0.5% increase in “accuracy” over 3 years (2009 to 2010 and 2010 to 2011). Revisions that mind you don’t change the pattern in employment growth. And before you spin out the whole “its stronger than flat” consider this, would you reported any differently if the employment growth was 2010: 305,910 2011: 149,430. I don’t think you would have. Well done, great journalism, now let the grownups get on with their work because obviously numbers are a bit too much for you.
You can find ERP here and the data for 0 to 14 year olds (Table 4 and 59):
http://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/3101.0Sep%202011?OpenDocument

Peter Martin said...

You have found that in the year to September 2011 employment growth was 37,580 stronger the ABS labour force figures suggest.

In the year to September 2010 it was 59,590 weaker than the ABS labour force figures suggest.

Here’s what I got when I attempted to isolate what we know so far about the calendar years:

In the first nine months of 2011 the ABS labour force figures say employment grew just 28,000.

In fact it appears to have grown by well over 50,000.

Throughout 2010 the ABS labour force figures say employment grew 364,400.

In fact it appears to have grown by 284,200.

Anonymous said...

So whats your point Pete? That you were wrong to suggest that the trend of employment significantly changed? Because by your own calculations (although i think you've made a mistake with the 2010 figure) average employment went from on average 24,000 growth per month in 2010 to around 6,000 per month. To me that tells basically the same story i.e. employment growth was weak in 2010.

Anonymous said...

Correction Weak in 2011

Anonymous said...

Jut in case my previous point wasn't obvious enough. The difference that you calculated between the original and revised, i.e. 3,000 employed persons / month for 2011. That is roughly equivalent to approximately 10 more employed people per month out of a sample of 56000 (see sampling fractions http://www.abs.gov.au/ausstats/abs@.nsf/mf/6269.0#PARALINK7). Which is roughly a change of 0.002% per month.
Pete the difference is noise! You and tim overplayed your hands and if you were decent journalists you both would come out and admit to it!

Garry Shilson-Josling said...

Hi Peter.

This, from July 2010:

http://www.tradingroom.com.au/apps/view_breaking_news_article.ac?page=/data/news_research/published/2010/7/186/catf_100705_174500_1938.html

The problem has been evident for some time.

Peter Martin said...

Wow. I always knew you were clever.

It makes the ABS decision to persist with its practices all the stranger.

Garry Shilson-Josling said...

Thanks Peter.
It is a bit strange, I agree.
I think most people who use these figures would rather see more frequent revisions if it meant more accurate estimates of the growth rates.
And they'd prefer more accurate estimates to start with, of course.
My impression, based on recent long-term arrrivals and departures data, is the the working age population is actually growing at something like 30k/month rather than the current ABS assumption of about 20k.
That is not huge, but enough to make a significant difference to impressions of the state of the labour market.
It will be interesting to see how this plays out.
One thing that maybe needs to be looked at is whether it makes sense to incorporate Dept of Immigration and Citizenship projections into the bureau's extrapolations, especially when we have quite timely data on how many people actually arrived in and departed Australia. (Migration flows are by far the biggest effect on population growth year to year.)

Post a Comment

COMMENTS ARE CLOSED