The differences in predictions coming from industry, the Opposition and the Government on the effect of the superannuation levy are unbelievably wide. They are so wide that the great British statesman’s warning about the moral status of statistics is worth bearing in mind.
The extremes are the Confederation of Australian Industries top limit of 60,000 jobs being lost and the prediction by the Treasurer, John Dawkins, that job losses will be minimal and might even create jobs.
The National Farmers’ Federation, citing the Murphy model, says there will not be much impact this financial year, but in the next two years the impact will be 30,000 jobs.
How do these tally with Mr Dawkins’s view that the levy would boost domestic savings and reduce dependence on foreign capital and the CAI’s figures are “”dodgy”.
The argument is over statistics, but it is really a question of predicting human behaviour.
To predict that the CAI relies heavily on surveying businesses on what they think they might do. It also does some straight-forward arithmetic. It’s arithmetic is clearly correct. It adds up the amount the levy will cost ($1.7 billion) on the present Australian wages bill and divides by the average wage cost per employee ($25,000) and comes to 67,000. (When the proposed levy was cut from 5 per cent to 4 per cent it cut its estimate accordingly to 60,000).
It did not say all those jobs would go, because the levy cost could be passed on in higher prices, postponing investment slowing employee promotions and so on. This is the “”dodgy” bit, not the figures.
It is “”dodgy” on two counts, both of them have much to do with human behaviour. To determine how much of the levy will go in jobs and how much in other methods, the CAI must rely on surveying its members. The breadth of the survey is not in question; the CAI represents 65 per cent of the private sector workforce. But the nature of the answer might be. Business does not like the compulsory levy. It prevents it from making its own choice about what to do with the money: capital investment, pay rises, profits and so on. Thus in answering any question about the levy business is likely to paint the grimmest picture in the most alarmist way. However, when it comes to action, the human element messes up the best model.
It is a difficult human thing to dismiss employees who have done nothing wrong. I was at an industry conference recently where people consistently said that the hardest part of this recession for employers has been being forced to sack loyal employees for economic reasons. It was done only as the very last resort.
It is fair to conclude therefore that people in industry are likely to ^ say @ the super levy will be catastrophic, while doing their damnedest to ensure that the very opposite takes place.
The NFF Murphy model is quite complicated. It has many variables. So the NFF has detailed predictions on the inflation rate, production and employment. The model takes into account an enormous array of factors like saving rates, the external economy, tax incentives, retirement figures and so on. But I bet nowhere does it have the factor: “”Oh God, how can I tell Bloggs (spouse, two kids, mortgage and car payments) who has worked for me for 10 years that he has lost his job?”
The human factor goes wider than that. It affects the outcome of every so-called economic decision made by Governments.
The trouble with economic modelling is that it relies on an assumption that people will act rationally. That would not be so bad if most of the people most of the time acted rationally. But some people act irrationally most of the time and all people reaction irrationally some of the time. Nearly everyone acts with imperfect information all of the time. How else would the economic modellers differ if that were not the case?
Who can tell, for example, the collective affect of a compulsory superannuation levy on individual savings? A sort of collective security might come over the country that our retirements will be much more affluent now, so we can give up saving in other ways. Or perhaps the collective fear of losing one’s job (exacerbated by pessimistic industry predictions) makes people save even more. The myriad of irrational human economic responses to major government announcements is impossible to predict.
The difference between an ordered economic model and those who think the randomness of human reaction makes financial prediction impossible is mirrored in the physical sciences.
The Newtonian tradition has it that the universe is understandable and predictable, provided one gets all the inputs right. Mathematicians have for centuries worked on theories which predict things and then prove the theory by proving what was predicted is actually the case in the physical universe. Stephen Hawking is now attempting to come up with a unified theory of the universe largely through mathematics.
That side of scientific thought in some ways is like the economic modellers. Oddly, the economic modellers are more ambitious. Whereas the mathematicians are only predicting the present state of matter (which will only be observed later when telescopes get more powerful), economic modellers are predicting the future.
The other side of scientific thought is chaos theory. Chaos theorists are uncomfortable with the ordered universe. They think prediction is impossible. This is best exemplified by an incident recorded by meteorologist Edward Lorenz. He tired of keying in a long figure when doing computer weather predictions so he rounded off a few decimal points. It threw the computer’s weather predictions out wildly. Those decimal points were the border between order a chaos.
Other physical examples quickly turned up. Then a counter theory was proposed that order could be found in the chaos within a range of inputs: that some processes in chemistry and biology were self-ordering. They were not self-ordering like neat linear equations, and precise results could not be predicted. But patterns could be. Coastlines, cauliflowers and the now famous Mandelbrot fractals, where paisley-style coloured patterns can be generated on a computer screen, are examples. Anyone can identify a pattern as a Mandelbrot fractal on a computer, but no-one can predict how the screen will look five minutes later after a few tiny changes on the program work their way through, other than to say it is another Mandelbrot fractal.
If this happens with only a tiny change of input into only five megabytes of program, how can one predict with any precision the result of a tiny change of input (the superannuation levy) to the gigantic gigabytes of program that is the Australian economy. The best we can hope for is that the input will not cross the boundary of chaos like Lorenz’s decimal point and that at the end of the day we will still have something that can be identified as the economy.
And this does not consider another damnation for economic modellers: the uncertainty principle. This states that the mere act of observation changes the nature and quality of that being observed. In the physics it means: look at something one way and it is mass; look at it another and it is energy. In economics, the mere observation of something will change the behaviour of those being observed.
Under observation, in the political arena, business people change their behaviour.
The worst of it is that chaos theory and the uncertainty principle are so powerful that even in two years time we will not be able to say whose prediction was correct: the NFF’s, the CAI’s or Mr Dawkins’s. Too many other tiny inputs could have caused the job gains or losses.
Meanwhile, those suffering greatest in the recession might be more prosaic about the whole prediction business and adopt Lloyd George’s words: “”You cannot feed the hungry on statistics.”