A point that often gets missed in the day to day political hyperbole over discussion of the stimulus and unemployment, is the actual temporal nature of unemployment itself.
One of the usual clichés that get thrown around with these so called debates, is that the stimulus has “cost $X for every projected job saved” type thing. The problem with this argument is that it is not only wrong, it is dishonestly wrong. Massively, dishonestly, wrong.
The inherent “wrongness” of it all derives from the way unemployment operates across time. To describe it properly, we can do it best with a few charts. There have been three relatively recent economic downturns in Australia – one in the early 1980’s, another one in the early 1990’s, as well as the one we are currently in today. To start with, if we just take the 1980’s and 1990’s downturns and chart the unemployment rate caused by the two downturns from the time it first started to rise in each period through to the point where it fully recovered, this is what we get:
These two downturns brought about a sharp upswing in the unemployment rate that took literally years to wash out of the system. This is by no means unusual as it’s the way unemployment generally works all around the world after a downturn hits – it takes a lot longer to repair the employment damage than it does to create it in the first place.
Let us now bring in our current unemployment experience at the moment. To do this we’ll have to replace the actual real dates on the horizontal axis – so we’ll measure the change in the unemployment rate in terms of months, starting from the first month unemployment rose in all three periods.
We’ll also have to change the unemployment rate as our measure for the vertical axis to let all three downturns become comparable. So rather than use the raw unemployment rate, we’ll use the change in the unemployment associated with the respective downturns. So in June 1981 for example (the month before the 80’s recession started to hit unemployment), the unemployment rate was 5.4%. In July of that year (one month later) it grew by 0.5% to 5.9% – so the measure for the 1st month for the 1980’s series would be 0.5.
Similarly, in 1990’s recession, In December 1989 the unemployment rate was 5.6%. The next month, January 1990, it jumped to 5.8% – so the measure for the 1st month of the 1990’s/00’s series would be 0.2%.
When we combine the three downturns into the same chart, now that we have a comparable set of metrics, we get (click to expand):
This is where it gets interesting – the unemployment consequences from the 1980’s downturn took around 100 months to wash out of the system. The 1990s downturn took longer, around 130 months to wash out of the system.
What we’re about to do now is only for explanatory purposes, to highlight the importance of time when it comes to looking at unemployment.
You’ll notice that the rate of unemployment growth during the 80’s and 90’s downturns (the slope of the blue and red lines up to around month 30) was very similar, as was the rate of recovery in unemployment over the ensuing years (the broad slope of the red and blue lines after unemployment peaked). We can take a look at how the change in the unemployment rate we are currently experiencing might look in two scenarios.
Firstly, we’ll have a Scenario A. This “best case scenario” assumes that our unemployment rate has effectively peaked and that it will soon start to decline at the average rate of the two previous recoveries – about a 0.057% decline in the unemployment rate per month.
Secondly, we’ll have a Scenario B. This is a “worst case scenario” where the unemployment rate grew to the 8.5% Treasury was initially forecasting – which would be a change of 4.6% on our chart since we started with an unemployment rate of 3.9%. We’ll also assume that the change in the unemployment rate would grow at the average rate of the previous two downturns, and that the recovery in unemployment would also be the same as the average of the previous two downturns.
This is what the two previous downturns plus our two scenarios look like:
Under our best case scenario (the grey line), it will take us over 2.5 years to recover back to the position we started from. Under our worst case scenario, where unemployment hit Treasury’s initial forecasts – our unemployment rate wouldn’t recover fully until 2016.
Remember, these charts are only for explanatory purposes – the unemployment lines will wander in reality for all sorts of reasons from domestic economic issues through to broader global growth levels – but they highlight the basic trends we see with downturns, and one that we will inevitably see with our current situation.
The purpose is to highlight the importance of time with unemployment. For every job that the stimulus hypothetically saved – it didn’t just save it for a single year. In some cases it will have saved that job every year, for 5 years or more in net terms.
But it doesn’t stop there.
Not only is the nostrum that “each job saved cost $X” utterly facile once you take the reality of time into account, it also completely ignores the opportunity costs involved – the costs that you would have had to spend anyway were unemployment higher.
The costs here are not only taxpayer costs such as higher welfare payments plus the inevitably higher government expenditure on labour market programs that high unemployment rates generate in Australia, but also the loss to GDP – the GDP we don’t have as a result of idle labour compared to the GDP we would have if that labour were engaged in paid employment.
When you dump a bucket of money into an economy like the government did, it has consequences – there is simply, absolutely no “debate” on this from anyone that can find their economic arse with a map, a GPS system and a set of directions.
The only real debate – and lets be blunt here, it is THE only REAL debate here – is over whether the squeeze is worth the juice, whether the costs are worth the benefits. Yet on this question, the popular cliché of “each job saved cost $X” is not only worthless, it is comprehensively dishonest to boot.
18 Comments
Instructive graphs and some very good points about the time lines of unemployment.
It may be impossible to figure out but how many perpetual unemployed are created by a down-turn? e.g. how many people who lost jobs in the 1990 recession were unemployed for 5-10+ years?
I think we have a problem with over analysis of monthly data and assuming that it provides more information than it did. e.g. Unemployment didn’t change this month, therefore everything is fine.
Your last plot shows that this time may not be that different to previous recessions and that unemployment takes 2+ years to peak (so in our case not until sometime next year at the earliest).
One of my collaborators used to say “You can fit two data points with any type of line you like”.
Makers sense – every job you save is one less you have to create in the recovery phase
So – the more you can save, the better
Also, saving that job is cheaper in the long run than the oncosts of not saving it, and the costs of restoring it.
Useful analysis Poss. Thanks for that.
In terms of the debate: was it worth the cost? Do you think that the difference in the “overall” consequences of higher unemployment to the economy because of a recession between your two scenarios is expressed by the difference in area underneath scenarios plot line??
Or is that too simplistic a way to look at that graph? Not sure if i’m espressing this well.
EP,
The interesting difference between this downturn and the last two is that this time the government, for the first time, acted early. In the previous two downturns, government action came effectively too late – once unemployment had already neared it’s peak. One of the things that will be interested over the next few years will be the actual shape of the unemployment series, whether it flatlines for a prolonged period then declines, or whether it peaks and then starts declining straight away.
Imacca,
Personally I think the stimulus delivered so far has produced a net benefit to human welfare. I also think most of what is to come will as well (basically hard infrastructure), although the government needs to keep an eye on what’s left of the schools program next year.
The area under the curves is certainly indicative of the size of the general difference between two any scenarios on unemployment (although there will be a fair bit of wiggle room with things like the composition of that unemployment and whatnot).
Was it worth the cost?
The Government chooses the unemployment rate. It has the capacity to employ every unemployed person in Australia who wishes to work.
I think we have to ask: why is it OK for people to be denied the right to work? Why should it be acceptable for the unemployment rate (never mind underemployment) to double over 18 months and not return for 8 years (for example in ’scenario B’)?
It’s a value judgment as to how much squeeze is OK to get a mil of juice. For my two cents, the squeeze was worth the juice. What is the point of running a surplus (effectively saving for a rainy day) if you won’t then spend when it’s bucketing down?
Kudos to the Rudd govt for squeezing as hard as they did. I think they got the balance right, and I think it was very courageous of them to do so in their first year or two in charge when they knew that the opposition would get hysterical about debt and deficits.
Of course you can squeeze too hard and if they did what Andos @ #6 says (ie employ everyone) then you would increase inflation, interest rates and threaten growth, nevermind the debt you would amass.
It would be a brilliant achievement if they could keep the rise in umemployment below 2% at the same time as keeping public sector debt to less than 30% of GDP that would be amazing.
If you’re waiting for honesty from the Liberals about this – or about most any other topic – you’ll be waiting till Hell freezes over.
Somewhat off topic, but both the data sets from previous downturns show a remarably similiar second hump in unemployment around 3 years after the peak. My guess would be the unwinding of spending/interest rate settings in order to keep debt and inflation in check give you this effect (essentially the w shaped recession, even if the w in not exactly symmetric).
In that case the level of debt accrued is a factor of importance and it is difficult to determine the best balance ahead of time. That being said, the major part of the debt that this downturn will leave us with does not derive from the stimulus spending but instead the structural problems with taxation that leaves us too dependedant on company tax. The excessive income tax cuts and middle class welfare that developed in the Howard/Costello years is as much to blame for the resultant debt we will have as the stimulus spending, although both will make recovery easier at least in the short term.
For some more thinking about unemployment levels, check out today’s post from Bill Mitchell: http://bilbo.economicoutlook.net/blog/?p=5060
John at #6 the jury is out on the question of whether the government being the employer of last resort is actually inflationary. You might want to do some reading on this site: http://e1.newcastle.edu.au/coffee/ Also there is a clearer introduction here: http://wapedia.mobi/en/Job_guarantee from which I quote
“The fixed JG wage provides an in-built inflation control mechanism. Mitchell (1998) called the ratio of JG employment to total employment the buffer employment ratio (BER). The BER conditions the overall rate of wage demands. When the BER is high, real wage demands will be correspondingly lower. If inflation exceeds the government’s announced target, tighter fiscal and monetary policy would be triggered to increase the BER, which entails workers transferring from the inflating sector to the fixed price JG sector. Ultimately this attenuates the inflation spiral. So instead of a buffer stock of unemployed being used to discipline the distributional struggle, the JG policy achieves this via compositional shifts in employment.”
Er, John at #7 I mean..
The lasting value of the stimulus funding is the extent to which it contributes to greater wealth and productivity and dare I say it, human happiness, many years from now. And a few years from now, this is what will be argued about, not what a job costs.
Possum, one graph you should look forward to is the dip in electricity supply caused by all those pink bats
Thanks Possum
Debunking feels SO good.
One thing the British discovered was that unstable employment/unemployment in the early part of your working life tends to lead to the same thing later on. From that point of view, the stimulus has obviously worked.
But why, oh why, has so much been spent on supporting jobs at the absolute bedrock (women in retail, men in construction) and so little on education, which was the obvious casualty of the Howard years?
Cud Chewer. Looking at electricity supply is an interesting proxy for GDP type figures.
One of the ways to judge what is happening in China is to check the electricity demand since the official figures are always a bit political. Another group has started looking at night time brightness from lights to judge growth in african countries where the data is sparse.
There is also a long-term social/psychological cost in having large numbers of people out of work and unable to get any. Having people spending a year or two or more not developing their work skills because they are stuck at home unemployed is also bad for society.
Perhaps the stimulus package should be judged by the job-years saved.
Jenny Lee: Perhaps one of the reasons the Government has to support “bedrock” (women in retail and men in construction) is because of permanently bringing in thousands of unskilled migrants.
It seems that the migrants not only arrive and have families, but they seem to bring with them a burning desire for their kids to have a better education than they did. Accordingly these children have a much better education than ones average Anglo/Irish/Scottish/Welsh descendants ever bothered to get. As far as the parents of this set of kids were concerned it was there. It was free. Too good to be true. Why bother? There’s always the footy.
Meanwhile, by the time all the migrant kids have passed their exams and entered the white collar professional market, the Government can afford to relax about a better education, because they-the migrants- have already paid for it themselves.
Time to bring in another lot of immigrants and lo the wheel turns. Why throw money at people who have already paid for the best education they can afford, when the next round of cannon-fodder will need to be educated.
I apologize if I have failed to communicate what I am trying to say.
But I did try.