The difference between forecast and actual

TAFKAS has long maintained that spreadsheets have been one of the most destructive human inventions of all time.  They allow any idiot the authority to claim to predict the future based on some questionable analysis underpinned by specious mathematics and statistics.

Before spreadsheets, people who wanted to do this form of soothsaying had to either do the hard work of calculating and recording with a pencil or siting in a tent with a turban looking into a crystal ball.

With the advent of spreadsheets, however, any idiot can drag a column to the right and forecast ad infinitum without the actual hard work of doing the calculations.  Spreadsheets have given many intellectually lazy and incurious people the confidence and certainty to engage in many idiotic adventures – often with other people’s money and resources.

It is this idiocy that has now infiltrated our political discourse with demands for and quotes of modelling, modelling and more modelling.  And yet the people (yes you media industrial complex) who are the most responsive to false authority of modeling are the ones who would not know what a model is if you stumbled upon one and are usually the most innumerate of our population.

But the thing about predictive models is that, with over time, you can actually see whether they were right or wrong.  Accountability is funny like that.  But for some reason, the modelers and model quoters never seem to want to do this.

Covid death models.  Climate catastrophe models.   Green job models.  Consistently wrong yet always used and never questioned.  Yet the faceless people who develop, demand and quote these models have the same level of skin in the game as the turbaned crystal ball reader.  About none.

Which brings TAFKAS to the latest fiscal adventure by the Commonwealth Government of Australia.  Reported in the AFR this morning:

The federal government is working up a multibillion-dollar scheme to construct residential housing as part of its strategy to stimulate the economy and prop up the building industry as the nation emerges from the coronavirus crisis.

A billion here.  A billion there.  Pretty soon you might have real money.  Even with the RBA printing presses running overtime.  But the description of this proposed program as a scheme is probably more accurate than the reporter intended.

Scheme, a noun

an organized plan for doing something, especially something dishonest or illegal that will bring a good result for you

But why is this scheme required you may ask?  Well apparently according to the Prime Minister:

Mr Morrison again raised the issue of an extreme drop in migrant, estimated by Treasury at 85 per cent.

Is this the same Treasury whose forecast, using a model, of the cost of the Jobs-Seeker-Keeper-Reeper program was only $60 billion off.  Oh.  But wait.  There’s more.  So says Prime Minister Morrison:

“We’re looking at net overseas migration falling to 34,000-odd next year. When you think that – it was the great Professor McDonald who set a figure of between 160,000 and 210,000 as being what you need in this country to maintain GDP-per-capita growth – then there’s obviously a big gap there,”

Perhaps others have heard this elsewhere, but this is the first time that TAFKAS has heard a government official state that net overseas migration of between 160,000 and 210,000 as being what you need in this country to maintain GDP-per-capita growthIs this not news?  Where are the headlines?

But from where did these magic numbers come?  Tada.  From here a – a Department of Immigration and Citizenship (precursor to Home Affairs) piece of commissioned research.

And this is what the research says:

There is a range of NOM levels (160,000 to 210,000) which the modelling suggests would have the ‘best’ impact by 2050 on ageing of the population and the rate of growth of GDP per capita.

These findings are based on a demographic model MoDEM 2.0. The value of research that is based on models depends crucially on their quality.

So basically, this is a model using data spit out from a different model.  A forecast on top of another forecast.  Yeah.  That was always going to end well.

And when was this forecast from a forecast done you ask?  May 2010.  TEN YEARS AGO.

Yes.  The actual results, with respect to average GDP growth have subsequently shown that the model and its forecasts were WRONG.  But the Prime Minister is not only quoting this WRONG analysis, but the government continues to base economic policy on it.

What’s next.  Phrenology and leeches to treat WuFlu?

But here is another magic conclusion from the research the Prime Minister not only quotes but says comes from the great Professor McDonald:

Migration has a meaningful impact on the rate of growth of per capita GDP. From 2013 to 2020, the rate of growth of GDP per capita would be about 0.15 percentage points higher with migration of 180,000 than with zero migration.

So.  For an extra 0.15 % (or 0.0015) of GDP growth, the states need to build expensive and questionable infrastructure, our schools, hospitals and public transport need to be over crowded.

Hows that for economic logic.

It’s no wonder that once your evidence in support of your questionable policy has been proved wrong, ignore the evidence and fall back on your rhetoric.

This is contemporary Australian political and economic leadership and why we are stuffed.

This entry was posted in Uncategorized. Bookmark the permalink.

32 Responses to The difference between forecast and actual

  1. cuckoo

    For an extra 0.15 % (or 0.0015) of GDP growth, the states need to build expensive and questionable infrastructure, our schools, hospitals and public transport need to be over crowded.

    To say nothing of doing everything to eradicate the last lingering traces of any historic Australian identity. One day some young Australian citizen may come across an episode of The Mavis Bramston show or My Name’s McGooley in a media museum, and the people he sees will be more exotic to him than the natives of Tierra Del Fuego, or Alpha Centauri.

  2. stackja

    Ronald Reagan said something like:
    Put predictions end to end, they never reach a conclusion.

  3. pbw

    More cheap shots against leeches!
    When will you stop the ad hominem attacks against the RE industry?

  4. Botswana O'Hooligan

    This housing plan sounds much like the NZ one of a 1000 completed houses a year. Only 47 were actually built.

  5. John A

    Covid death models. Climate catastrophe models. Green job models. Consistently wrong yet always used and never questioned. Yet the faceless people who develop, demand and quote these models have the same level of skin in the game as the turbaned crystal ball reader. About none.

    To which one could add modeling for the business case for government projects like bridges, tunnels and freeway upgrades, or even “major events.”

  6. stackja

    Liberty Quote
    A gruber (n.) is someone who grubes (v.i.) – they tell people, who they have lied to, how stupid they were to have believed the lie they were told.
    — Steve Kates

  7. John A

    Perhaps others have heard this elsewhere, but this is the first time that TAFKAS has heard a government official state that net overseas migration of between 160,000 and 210,000 as being what you need in this country to maintain GDP-per-capita growth.

    Well, since we have approximately 80,000 to 100,000 abortions per annum, thus forfeiting a lot of natural increase, the shortfall “has to be made up” via inbound migration, according to this “modelling.”

    Which of the regular Cat commenters points out the essential Ponzi scheme nature of migration policy? This report fully substantiates that assertion.

    By contrast, it also torpedoes the Zero Population Growth idiocy and the other anti-humanity assumptions which underlie the Watermelon Greens policy proposals.

  8. Dave of Reedy Creek, Qld

    Scary stuff. I think the “modelling” phenomena had its roots in Climate Change where the media touted every projection as gospel. Pity for them not one has ever come to pass, now modelling is spreading faster than the Wuhan virus and in many ways just as deadly. makes one wonder is this where the RBA got its madness from on interest rates. I wonder could we model the following and see what they come up with…”The definition of stupidity is doing the same thing over and over and expecting a different result.”

  9. Terry

    ‘…that spreadsheets have been one of the most destructive human inventions of all time.’

    You are blaming the wrong “tool” (it’s the “forecaster”, not the spreadsheet).

    You might as well blame gravity for a plane crash.

  10. duncanm

    leeches to treat WuFlu

    Bad example – that may be a good idea.

    Covid complications include post-infection clotting

  11. shady

    People nowadays think accountability means the ability to count. Modern corporate management in a nutshell.

  12. John Bayley

    Look at the bright side.
    This is all great news for the CFMMEU, who will be the ones building the ‘government-led recovery’.
    No doubt the Liberals think that will make the unions vote for them.
    Maybe they can ask John Setka if he wants to run as a LNP candidate. He could probably beat Adam Bandt in Melbourne. Or Adam may decide not to run at all, just in case something were to happen to him.

  13. MACK

    Cheap computing power and illicit drugs have a lot to answer for. Sometimes I think both may be in action at once.

  14. From 2013 to 2020, the rate of growth of GDP per capita would be about 0.15 percentage points higher with migration of 180,000 than with zero migration

    Except exactly the opposite happened. GDP per capita has been plummeting since about 2012.

  15. Kneel

    “All models are wrong – some are useful”

    Whether by pencil and paper, spreadsheet, custom written software, or anything else, models are merely codified theories of experts (well, hopefully of knowledgeable experts).

    They need to be validated and verified to ensure they are reasonably predictive.
    Validated: demonstration that the model is calculated what it is intended to calculate.
    Verified: demonstration that predictions match reality.

    There are several issues with various models that never seem to be addressed, such as:

    for climate models, there is no evidence that the discrete solutions to the PDE’s converge with the continuous solutions (example: use a photo editor to “pixelate” an image. Convergence is demonstrated when changing the size of the pixelation just makes it “blury” as opposed to “completely different”). This means the output may in fact just be “numeric noise”.

    for economic models, human adaptation appears to be ignored – this means that when people change their behaviour due to changed circumstances, the model rapidly diverges from reality.

    Both of these examples could be found if the model was verified and validated.

    A sensitivity analysis also helps – that is, what happens if you change just one of the inputs over its entire “reasonable” range. One at a time that is. If the output is not reasonable over the entire input range, there may be an issue with the model, or reality may also show discontinuities – either way, you need to understand why.

    It may surprise TAFKAS that economic models are not the worst – that position is occupied by medical and climate models (although economic ones are not far behind!)

  16. Tim Neilson

    “Don’t you somehow feel how false it all is, and how falsely reassuring – this nineteenth century gloss of statistical accuracy, as if the flood tide of history could run in rivulets tidy enough for garden irrigation, safe enough for a million taps in suburban bathrooms…”.

    [Charles Rainier, in James Hilton’s “Random Harvest”.]

  17. Tim Neilson

    I should complete that “Random Harvest quote – the other part irresistibly reminds me of someone in today’s world. (I’ll let you guess.)

    “… but when the storm does come, who’ll give a damn if the rows of little figures still add up – who’ll care if the sums are all wrong provided one man knows a right answer?”

  18. Faye

    Politicians think mathematical predictions are too mysterious and highfalutin for them to figure out even though they could rustle up the fundamental figures to work with. At least a “back of an envelope” try would be a check against what they are presented. So they leave it to the “don’t worry about that” credentialed high flyers who can push a button and presto – “there’s an answer” – a wrong answer!

    You would think with so much tax payer hard earned money at risk that a checking procedure would be enforced. Why do the people involved think computers (ie their inputs) are infallible?

    Too many wrong predictions are rolling off the peanut factory’s production line and the economy is choking. The $60B over estimate – was that ever explained in detail? Who is to say there aren’t more mistakes everywhere.

  19. Boambee John

    So they leave it to the “don’t worry about that” credentialed high flyers who can push a button and presto – “there’s an answer” – a wrong answer!

    “For every problem thete is a solution, neat, logical, wrong!”

  20. gafa

    To heck with the building industry, let it implode in on itself for a while. To use the building industry Ponzi scheme as an excuse for immigration is pathetic. Tradie rates need to return to realistic levels, $120,000 pa to swivel a stop/go sign, get out of here.
    Unions, Banks, Developers form a triangle of money sucking leeches.
    The Compulsory Superannuation scheme helps feed into this and is one of the worst, unfair and destructive schemes to have ever been forced onto the people. Needs to be majorly revised or suspended asap.

  21. Delta

    Briggs sums up the model farce quite accurately. Coronavirus Update XV: The Curse of Karen

    From his post:

    “I’ve tried to explain many times that models can only say what they’re told to say. All models. Every model. Even our models.

    I make my model say “the shape of the viral outbreak shall resemble a logistic curve.” That was all me. Indeed, way back when this started, I created the model for fun as a class project, and only because so many were chimping out about “Exponential! Exponential!” (Which was always an impossibility.)

    My initial model for the evolution of an insane reaction was woefully inadequate. I didn’t foresee how what was clearly going to be a routine pandemic (and all are deadly) would turn into a massive lurch toward the left singularity. There’s no excuse for this on my part, because I’ve been preaching its imminence for years!

    Anyway, those models predicting more coronadoom and how we should have liberty on Tuesdays and tyranny of Thursdays are only doing what their designers told them to do.

    Every single time you hear of a model—any model anywhere—ask yourself these two questions: (1) Why does this guy want me to believe it? (2) Has that model proven itself before in making skillful prediction of data never before seen or used in any way?

    If the answer to (2) is “No”, be skeptical. If the answer to (1) is “THE SCIENCE says so”, run away.”

  22. Ian of Brisbane

    The only stimulus required and the only one that would work is a big tax cut

  23. Let’s insist all immigrants must live in Canberra for three years and bolster their population. Policy would be modified I’m guessing.

  24. cohenite

    The chunk virus alarmists are just like the climate alarmists: they just make shit up and are never held to account. Here’s Braganza, the head of the Bureau of Meteorology at the bushfire RC claiming the BoM forecasts were totally accurate:

    http://theclimatescepticsparty.blogspot.com/2020/05/the-bureau-of-meteorology-gets-it-wrong_27.html

  25. Dr Faustus

    As a rule of thumb, modelling based on uncertain parameters will never have any serious predictive power. The Rev Thomas Bayes saw to that.

    That’s not to say they are useless: but the usefulness is mainly in testing input assumptions and looking at possible ranges of outputs.

    Our current Covid position was brought about by the political need for a blame-free compass – with experts and mathematics and modelling and sciency stuff as air cover.

  26. H B Bear

    End of the Population Ponzi will create a few problems for Treasury.

    Expect to see some real bullsh1t.

  27. Damon

    Models are necessary to justify political decisions that have already been made.

  28. Alex

    “The only stimulus required and the only one that would work is a big tax cut”, sorry Ian we don’t have a model for that.
    Head of Treasures
    Canberra

  29. Squirrel

    The stuff about reviving manufacturing in Straya is just window-dressing – the only real plan is to get the Big Australia Ponzi scheme, and all its related money-spinning rackets back into top gear as soon as possible.

    In ten years time, we’ll still be relying on mining and agriculture to pay our way in the world, and debt-fuelled consumption to keep most of the punters happy at home.

  30. duncanm

    We’re having quite an amusing episode in our (tech) company at the moment, which is related.

    A few of us engineers have been quietly writing down the sales forecasts for a number of years, almost a decade.

    Through a few revisions, we worked out the most efficient way to present this data was to plot “Forecast period” vs. Revenue; ie: at the left of the X axis, you have the most distant forecasts (whenever they were made), and at the right of the X axis, you have ‘actual sales’ at forecast time zero.

    It doesn’t matter when the forecast was made, all that is important is how far into the future it looks.

    The trend which resulted has been amusing us for years – we can predict the future actual sales quite accurately from some years out; sales/marketing were always wildly optimistic (something like +20%/year. So four years out, they’d be out by a factor of over 2x (= 1.2^4)).

    No-one in management above them seemed to hold them to their forecasts, and we’ve kept this beautiful little nugget secret, updating it at every sales/marketing update when the forecasts are presented.

    In the heat of an argument in a fairly senior meeting, the graph got thrown up, and apparently marketing/sales are most upset!

    Blowback is expected soon.

  31. If the people we’re bringing in are less productive than the average productivity of residents, then per capita GDP will fall, not increase.
    The people we’re bringing in include those from shit hole countries like Iraq, Syria, Somalia and Afghanistan as well as family reunion oldies.
    I don’t need no steeenking models. Bringing in 160,000 to 200,000 migrants may increase GDP but they will DECREASE PER CAPITA GDP. Have no doubt.

  32. Beachcomber

    The fabrication and invention continue unabated.

    Lockdowns Saved Trillions! …. Say Models


    The government has spent decades studying what a life is worth. It hasn’t made a difference in the covid-19 crisis.

    Economists at the University of Wyoming estimated the economic benefits from lives saved by efforts to “flatten the curve” outweighed the projected massive hit to the nation’s economy by a staggering $5.2 trillion. Another study by two University of Chicago economists estimated the savings from social distancing could be so huge, “it is difficult to think of any intervention with such large potential benefits to American citizens.”

Comments are closed.